The line with the asm is
asm volatile("msr msp, %0"::"g"
(*(volatile uint32_t *)BLDR_ADDRESS));from (https://github.com/1Bitsy/1bitsy-examples/blob/master/examples/1bitsy/common/button_boot.c)
arm-none-eabi-gcc --version returns:arm-none-eabi-gcc (GNU Tools for ARM Embedded Processors 6-2017-q2-update) 6.3.1 20170215 (release) [ARM/embedded-6-branch revision 245512]
So it turns out that the TFT really like NSS (chip-select) to be cleared before each command and set after each command. Which mea ns that I need to wait for an spi_send to complete before setting NSS back to high. Is there a good way to do this in libopencm3? I ended up having to do the following:
while (!(SPI_SR(SPI2) & SPI_SR_TXE));
while ((SPI_SR(SPI2) & SPI_SR_BSY));
The reason that works is because spi_xfer waits for RXNE flag which is only set after the full number of bits has been sent (and read)
@esden Check out the TFT example in my fork on (https://github.com/MarkOsborne/1bitsy-examples/tree/master/examples/1bitsy/tft-st7735)
Let me know what you think. I've tried to keep it as simple as possible. Adafruit sends 60 odd commands on init - I send three.
Let me know what you think. I've tried to keep it as simple as possible. Adafruit sends 60 odd commands on init - I send three.
thank you for putting it together!
On the first glance I only found a cosmetic issues :D https://github.com/MarkOsborne/1bitsy-examples/blob/master/examples/1bitsy/tft-st7735/st7735lcd.c#L215 something is off in the indent departament. :D
I will pull it in a sec and test compile it. It looks great to me, do you want to send in a pull request?
for future reference, you want to use a branch for new additions, that makes it easier to manage pull requests: https://guides.github.com/introduction/flow/
A branch in my fork or a branch in your repo. Either way I should have started with a topic branch (I know better).
after you do that github will pop up with a pull request button
then you send in a pull request to the main repositary using that branch
this allows you to amend and change things in that branch untill the pull request gets accepted and merged into the main repository.
this way you do not have to delete pull requests and you can send those pull requests early even if you are not fully ready for the code to be merged, we can discuss what you did and you can make changes including force push untill it is “right” :D
It is a very nice way of using git branching for collaboration. :)
this way you can have even more than one pull request pending in parallel.
I hope that makes sense. :)
OK - reset master, moved change to a topic branch and fixed whitespace. The change is now in a topic branch: (https://github.com/MarkOsborne/1bitsy-examples/tree/st7735/examples/1bitsy/tft-st7735)
I'll submit a PR now so you can comment. I'll retest this evening.
I'll submit a PR now so you can comment. I'll retest this evening.
You are welcome. Writing the TFT example satisfied one of my learning goals for the 1bitsy. Which was to write something on an ARM chip from the datasheet up and understand every line of code. And libopencm3 satisfied my desire not to have to map all the memory addresses.
But it got me to wondering, what is the driver model for 1libsy/libopencm3? One of the things that makes Arduino popular, and probably applies to STM32CubeF4, is the availability of pre-written drivers. Being able to easily grab a driver that works is a huge boost to productivity and reading other peoples drivers furthers learning.
I wonder, would it be interesting to build up a set of libraries for the 1bitsy (based on the core libopncm3) that people can add to overtime? 1bitsy-examples would probably not be the right repo for that as the samples should ideally be simple and easy to understand - flexible, performant libraries don't always make good examples.
@BecomingMaker_twitter I wrote an SPI test and that did the wait between command and data the same as you. I use defines in include/libopencm3/stm32/common/spi_common_all.h instead of the doc examples. SPI2_SR is the same as your (SPI_SR(SPI2) . I used figure 253 in the stm32reference.pdf to figure out the timing.
@esden My BMP JTAG scan stopped working and I panicked a fired off an email with an unanswerable question, so please ignore it. I suspect the cable and was wondering how I can figure out what pogo pin pads are the JTAG signals on both ends of the cable. I found the pcb files, I think, but I need a viewer because I only use KiCad.
also you might want to try out SWD to scan your target instead of JTAG
it uses less electrical connections for communication
usually you should be able to get around it with either using “connect_srst” … this is needed when you use WFI instructions. Also if you have a firmware that is setting TRST pin to be GPIO. (PB4) you will have to use SWD not JTAG as it is an errata related bug that JTAG is not working on the STM32F415 silicon when TRST pin is set to GPIO.
Ohh and a broken ribbon cable does behave erraticly I have seen that happen with my ribbon cable I use on daily basis for everything including programming production boards. I assume these were at least 5k cycles that resulted in an intermitten connection on one of the strands.
@BecomingMaker_twitter I was having a big problem where I could not connect. I could use
monitor connect_srst enable before scanning and then I could scan and attach, but upon running it would fail. It turned out that my code was disabling the pins used for jtag , killing the connection. Make sure your code leaves the jtag/swd pins you need configured correctly.