Author Topic: Calling all Linux Kernel developers!  (Read 348048 times)

0 Members and 1 Guest are viewing this topic.

Offline tangrs

  • LV5 Advanced (Next: 300)
  • *****
  • Posts: 229
  • Rating: +98/-0
    • View Profile
    • tangrs blog
Re: Calling all Linux Kernel developers!
« Reply #435 on: December 02, 2012, 03:55:30 am »
@Vogtinator: Don't bother with the LED. There's some weird protection that TI put on that noone could work out.
Good work on the GPIO and power management interfaces though :)

In other news, Clickpad support is officially working! Lionel and I are also working to get nightly builds on wheels :D

Offline Vogtinator

  • LV9 Veteran (Next: 1337)
  • *********
  • Posts: 1193
  • Rating: +108/-5
  • Instruction counter
    • View Profile
Re: Calling all Linux Kernel developers!
« Reply #436 on: December 02, 2012, 07:17:04 am »
Quote
Good work on the GPIO and power management interfaces though
I don't know whether it actually works..
Turning off by disabling access to all peripherals is likely to be a bug or it's not off at all, but we can't see it's actually on.
Also I don't know whether GPIOs do work, I can't test it, but some values make sense.
But what could be the cause for the adc always returning 0?

Offline tangrs

  • LV5 Advanced (Next: 300)
  • *****
  • Posts: 229
  • Rating: +98/-0
    • View Profile
    • tangrs blog
Re: Calling all Linux Kernel developers!
« Reply #437 on: December 02, 2012, 07:20:17 am »
Quote
Good work on the GPIO and power management interfaces though
I don't know whether it actually works..
Turning off by disabling access to all peripherals is likely to be a bug or it's not off at all, but we can't see it's actually on.
Also I don't know whether GPIOs do work, I can't test it, but some values make sense.
But what could be the cause for the adc always returning 0?

Does it give you the same thing if you try to do the reads as an Ndless program? I haven't got access to my calculators at the moment.

Offline Vogtinator

  • LV9 Veteran (Next: 1337)
  • *********
  • Posts: 1193
  • Rating: +108/-5
  • Instruction counter
    • View Profile
Re: Calling all Linux Kernel developers!
« Reply #438 on: December 02, 2012, 07:21:15 am »
Is there already a program for ndless to test it or I do I have to do it wrong again? :D

Offline tangrs

  • LV5 Advanced (Next: 300)
  • *****
  • Posts: 229
  • Rating: +98/-0
    • View Profile
    • tangrs blog
Re: Calling all Linux Kernel developers!
« Reply #439 on: December 02, 2012, 07:22:50 am »
Is there already a program for ndless to test it or I do I have to do it wrong again? :D

Nope, but it shouldn't be too hard to whip one up quickly :P

Offline Vogtinator

  • LV9 Veteran (Next: 1337)
  • *********
  • Posts: 1193
  • Rating: +108/-5
  • Instruction counter
    • View Profile
Re: Calling all Linux Kernel developers!
« Reply #440 on: December 02, 2012, 07:25:28 am »
What if I make the same mistakes again and it's returning the wrong values?

Offline tangrs

  • LV5 Advanced (Next: 300)
  • *****
  • Posts: 229
  • Rating: +98/-0
    • View Profile
    • tangrs blog
Re: Calling all Linux Kernel developers!
« Reply #441 on: December 02, 2012, 07:28:51 am »
What if I make the same mistakes again and it's returning the wrong values?

I meant for the ADC. If you can get an ndless program to read the correct values, then you can get linux to read the same.

Offline Vogtinator

  • LV9 Veteran (Next: 1337)
  • *********
  • Posts: 1193
  • Rating: +108/-5
  • Instruction counter
    • View Profile
Re: Calling all Linux Kernel developers!
« Reply #442 on: December 02, 2012, 07:30:27 am »
Ok, I'll try that.
Edit: Nope, zero.
I'm reading directly from 0xC4000100 + 0x10 (measured voltage) + 0x20*channel, or am I too dumb to read the wiki page?
The OS should have started a measurement already, so something should be there..
Edit2: On nspire_emu it works (ADC0 = 930), but not on the actual hardware
Edit3: In the diagnostics menu, there are some GPIOs higher than 32 (GPIO39). Why are these not mentioned in the wiki?
« Last Edit: December 02, 2012, 09:04:05 am by Vogtinator »

Offline mdr1

  • LV6 Super Member (Next: 500)
  • ******
  • Posts: 303
  • Rating: +21/-2
    • View Profile
Re: Calling all Linux Kernel developers!
« Reply #443 on: December 02, 2012, 10:13:58 am »
Really great !

X and directfb are working!
Now we're one step closer to minecraft :D
Note: I can't login on SSH because of an german umlaut in my password and I haven't changed the keyboard-layout yet.
Note2: I don't know where that segfault came from, maybe too much ctrl-c? ;-)
[/youtube]

I don't understand how you can launch this graphical program, can you help me to do so please ?



Offline Vogtinator

  • LV9 Veteran (Next: 1337)
  • *********
  • Posts: 1193
  • Rating: +108/-5
  • Instruction counter
    • View Profile
Re: Calling all Linux Kernel developers!
« Reply #444 on: December 02, 2012, 10:16:55 am »
You only have to run "startx" on your calc.

BTW: 0x1ACCE551 is sent to the watchdog for unlocking, problably not related to the leds.

Edit: I can read the ADC via ndless, but the driver does still not work :-(
« Last Edit: December 02, 2012, 10:47:06 am by Vogtinator »

Offline InspiredByCas

  • LV3 Member (Next: 100)
  • ***
  • Posts: 44
  • Rating: +4/-1
    • View Profile
Re: Calling all Linux Kernel developers!
« Reply #445 on: December 02, 2012, 10:41:16 am »
There is one thing which I don't understand:
How can the OS lock the LED when it is overwritten by linucx in the RAM?
Or how can the OS restart the calc when reset button is pressed? (In earlier OS all user data was deleted when pressing reset button on the back.)
Or why can nClock run in background with linucx? I think it does because after the 'reboot' command in linucx nClock has the right time on startup, rebooting with reset button messes up nClock and it shows the wrong time.

Are there two different RAMs?

And if you write to these addresses in the RAM to do something special (watchdog, LED, power management), wouldn't the OS or linucx (or whatever runs) be overwritten partially?
« Last Edit: December 02, 2012, 10:44:33 am by InspiredByCas »

Offline Lionel Debroux

  • LV11 Super Veteran (Next: 3000)
  • ***********
  • Posts: 2135
  • Rating: +290/-45
    • View Profile
    • TI-Chess Team
Re: Calling all Linux Kernel developers!
« Reply #446 on: December 02, 2012, 10:58:28 am »
Quote
How can the OS lock the LED when it is overwritten by linucx in the RAM?
TI's OS does not lock the LED when it's overwritten by Linux; it's just that Linux doesn't activate it.
EDIT after Vogtinator's post: by the sentence above, I meant that while Linux is being executing and overwritten, TI's OS is not in a running state; but indeed, the state of the PTT is stored somewhere, in the hardware ports and/or the Flash memory (see what occurred during the beta-testing of Nleash).

Quote
Or how can the OS restart the calc when reset button is pressed? (In earlier OS all user data was deleted when pressing reset button on the back.)
Pressing this button triggers an interrupt, which is handled by TI's OS.

Quote
Or why can nClock run in background with linucx?
It cannot ;)
Like the rest, it is completely replaced by Linux and its userspace.

Quote
Are there two different RAMs?
Yes, there are indeed two different RAMs, but the main parts of both TI's OS and Linux use the same RAM, in the same area.
« Last Edit: December 02, 2012, 11:28:55 am by Lionel Debroux »
Member of the TI-Chess Team.
Co-maintainer of GCC4TI (GCC4TI online documentation), TILP and TIEmu.
Co-admin of TI-Planet.

Offline Vogtinator

  • LV9 Veteran (Next: 1337)
  • *********
  • Posts: 1193
  • Rating: +108/-5
  • Instruction counter
    • View Profile
Re: Calling all Linux Kernel developers!
« Reply #447 on: December 02, 2012, 11:23:43 am »
Quote
TI's OS does not lock the LED when it's overwritten by Linux; it's just that Linux doesn't activate it.
Partially wrong: We don't know how the OS can unlock the LED. Whether the LED registers are locked is saved somewhere in hardware.

In my adc test prog I had to do the following:
Code: [Select]
*((uint32_t*)0x900B0018) = 0;
*((uint32_t*)0x900B0020) = 0;
What is this register "0x900B0020"? In nspire_emu *(0x900B0020) & 0x10 is tested, if it returns false, the adc value is 0x3ff=1023.
« Last Edit: December 02, 2012, 11:36:06 am by Vogtinator »

Offline mdr1

  • LV6 Super Member (Next: 500)
  • ******
  • Posts: 303
  • Rating: +21/-2
    • View Profile
Re: Calling all Linux Kernel developers!
« Reply #448 on: December 02, 2012, 11:26:29 am »
You only have to run "startx" on your calc.

BTW: 0x1ACCE551 is sent to the watchdog for unlocking, problably not related to the leds.

Edit: I can read the ADC via ndless, but the driver does still not work :-(
But when I enter "startx", the console answers "not found".



Offline Vogtinator

  • LV9 Veteran (Next: 1337)
  • *********
  • Posts: 1193
  • Rating: +108/-5
  • Instruction counter
    • View Profile
Re: Calling all Linux Kernel developers!
« Reply #449 on: December 02, 2012, 11:28:21 am »
You are not using the right rootfs image.