SPI 2 Channels

[ephraim]

Dear Sir .

I have on my custom board 2 spi , spi1 and spi3.

&spi3 {
status = "okay";
compatible = "nordic,nrf-spim";
sck-pin = <31>;
mosi-pin = <9>;
miso-pin = <30>;
cs-gpios = <&gpio0 26 1>,<&gpio0 15 1>;
};

&spi1 {
status = "okay";
compatible = "nordic,nrf-spim";
sck-pin = <3>;
mosi-pin = <8>;
miso-pin = <5>;
cs-gpios = <&gpio0 4 1>;
};

The spi works within an interrupt and spi3 works fine with 2 slaves.

void AFE4900_Interrupt(struct device *gpiob, struct gpio_callback *cb,u32_t pins)
{
     unsigned char a;

switch(_AFE4900_RECIPE.test_type)
{
case none: if(_AFE4900_AQUIRE_MODE.aquire_mode == afe4900_aquire_sample)
{
        SPI_read_continuous(0x2a,2);
        flash_page_counter += afe4900_mem_chunk;

       if(++afe4900_int_counter >= 40)
       {
          ReadAccelerometerXYZ();

          MT25qlxx_WriteEnable();

           MT25qlxx_WriteFullPage(&flash_buffer[flash_bank][flash_section_counter * 0x100],0);

           flash_section_counter++;
           flash_page_counter += afe4900_mem_chunk;

            afe4900_int_counter = 0;
           flash_page_counter = 0;      

   }
}

 Routiunes SPI_read_continuous(0x2a,2); , MT25qlxx_WriteEnable(); , MT25qlxx_WriteFullPage(&flash_buffer[flash_bank][flash_section_counter * 0x100],0);

SPI_read_continuous(0x2a,2);    works with spi3  1st cs, 

 MT25qlxx_WriteEnable();

 MT25qlxx_WriteFullPage(&flash_buffer[flash_bank][flash_section_counter * 0x100],0);   works with spi3  2nd cs, 

ReadAccelerometerXYZ(); works with spi1.

As long as only one spi works 1 or 3 , The spi works fine ,  ( I comment out the unrelated spi call ).

When the 2 spi works 1 and 3 , The system is stuck after the first interrupt .

 

The Interrupt occurs every 2 mSec , there is no time overflow .

What can cause this phenomena ? 

Attached a scree shot , the top is spi3 , the bottom is spi1 , after 1 interrupt the system is stuck

Please Advise.

[CheMax]

Why not use the PPI to run an SPI that has only one slave? And cause an interruption already at the end of the exchange?

Can you track what happens in the interrupt? Suspicion that the function of working with flash memory may be blocked. And by the way, this is not the best solution for starting slow work with memory from an interrupt.

And yes, when you paste your code, it is better to use: Insert-> </.Insert Code. 

First, it visually separates code from text, which improves readability.
Secondly, the question will look more neat

[ephraim]

Hi.

As I mentioned either I work with spi1 or spi3 only , one at the time , There is no "lock" of the cpu.

Only when spi1 works after spi3 the systems stuck .

Is there a way to enable / disable each SPI before working with the second spi ?

[CheMax]

I believe that in this way you can disable  (I do not use HAL, so I propose a solution on registers):

NRF_SPIM1->ENABLE   = SPIM_ENABLE_ENABLE_Disabled;

or:

NRF_SPIM3->ENABLE   = SPIM_ENABLE_ENABLE_Disabled;

but it’s in manual mode. as for the automatic depending on the interruption, I don’t know, maybe there is such a solution. In general, this should be considered)

edit: 

look for a function "nrf_spim_disable(p_spim) " on the project

[Simon]

Have you figured this out ephraim? Did the solution from CheMax fix your problems?

Best regards,

Simon