Hello everyone,
I am a little bit late in this topic, but I would like to share what I have
found.
The problem is basically the clock. Once we boot our BBB, clocking systems
for GPIO 1, 2 and 3 are disabled (GPIO0 will be always enabled).
We can enable them by:
- Exporting any pin on each port (and then the system enables the clock
automatically);
- Modifying and setting some bits in the register CM_PER_GPIOx_CLKCTRL
(AM335x Technical Ref Manual - pages 1195, 1196 and 1197).
Some registers from CM_PER (Clock Management - Peripherals) will enable the
clock for each module.
You can find GPIO1 with an offset 0xAC, for example.
Setting bit 1 will enable the clock for this module so that we will be able
to read/write GPIO0 through memory mapping.
You can change the values from this register by mapping it, just like you
did for setting/clearing a bit on a gpio.
Hope it helps!
Regards,
Patricia.
Luigi very good, I also had this problem, and your solution worked for me
> also.
>
> But I still do not understand why it is necessary to export at least one
> pin on each port to be able to access the GPIO with mmap.
> Someone can explain me?
>
> Em terça-feira, 11 de junho de 2013 23h41min10s UTC-3, Jacek Radzikowski
> escreveu:
>>
>> Hello,
>>
>> I'm getting bus errors whenever my program tries to access registers
>> controlling pins on GPIOs 0,2 or 3. I'm not trying to set pinmuxing, I
>> try to write to registers controlling the pins.
>> GPIO1 works fine and I can enable outputs by writing to the OE
>> register and change the values on the pins by writing to the OUT
>> register.
>>
>> Here are some details of my simple test program:
>> The base addresses for the GPIO blocks are defined as follows:
>> const uint32_t gpioAddrs[] =
>> { 0x44E07000, 0x4804C000, 0x481AC000, 0x481AE000 };
>>
>> The memory blocks are mapped to process address space with the following
>> mmap:
>> gpios[i] = (uint32_t *) mmap(NULL, 0xfff,
>> PROT_READ | PROT_WRITE, MAP_SHARED, gpioFd,
>> gpioAddrs[i]);
>>
>> Printing values from OE, IN and OUT registers:
>> printf("i=%i\n",i);
>> printf("OUT[%i]=0x%08x\n",i,gpios[i][DATA_OUT_REG/4]);
>> printf("IN[%i]=0x%08x\n",i,gpios[i][DATA_IN_REG/4]);
>> printf("OE[%i]=0x%08x\n",i,gpios[i][GPIO_OE_REG/4]);
>>
>> The full source is on gist:
>> https://gist.github.com/piranha32/4fd285cc8333eeb4ec39
>>
>> Program works fine when 'i' is set to 1, crashes for 'i' set to 0,2 or 3.
>> Am I missing something obvious, or is this an overzealous memory
>> protection?
>>
>> Board: BBBlack
>> Kernel: Linux beaglebone 3.8.13 #1 SMP Fri Jun 7 09:49:12 CEST 2013
>> armv7l GNU/Linux
>>
>> thx,
>> j.
>>
>>
>> --
>> Given a choice between two theories, take the one which is funnier
>>
>
--
For more options, visit http://beagleboard.org/discuss
---
You received this message because you are subscribed to the Google Groups
"BeagleBoard" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
