Posted at 2025-01-23 by @gfwilliams

Hi - I'm afraid I'm not sure why it would be broken, no - however @SimonGAndrews has recently done a bunch of work on I2C and that needed a bit of fettling to work nicelty with the C3, so SPI might be the same.

Honestly, I wouldn't be put off using software SPI - whichever way you write the data Espruino will wait until the transmission is complete, so using hardware over software really doesn't gain you very much.

Posted at 2025-01-23 by @SimonGAndrews

Hi @devsnd , I would be happy to help figure this out. My current project needs to use SPI for an analog input chip. And also wanted to prove the I2c was working on S3 also.
Exactly what board are you using I can’t find one with that pinout. Please sent a link for it. Let me know and I will take a look.
My immediate thought is maybe the code is interfering with the chips use of SPI to access its flash.

Posted at 2025-01-23 by @devsnd

Thanks @gfwilliams, I thought the SPI bus would be buffered in hardware, but if it's a blocking operation either way, I'll just continue using software SPI and call it a day! Thank you for clearing that up.

Posted at 2025-01-23 by @devsnd

Hey @SimonGAndrews,

This is the board I'm working with: https://www.aliexpress.com/item/1005006599448997.html

Except for the SPI everything has been working nicely so far (BLE, WiFi), only exception being that the LED1 is on pin 8 for the super mini board. So I have to use D8 instead of LED1 to use the built in LED. Also, the LED is inverted, so I have to write(false) to turn it on

Posted at 2025-01-23 by @enaon

everything has been working nicely so far (BLE, WiFi),

I am sorry if the question is a bit irrelevant. Is this the 2.25 the ESP32 version of espruino, using https://www.espruino.com/Espressif+Flash?

I am having troyble with an esp32 wemos board, the wifi stops after some time, I get an ERROR: Connect failed (err 118). I just want a verification that esp32 works stable in general.

Posted at 2025-01-23 by @SimonGAndrews

@devsnd
Ok I can see the same failure with another S3 board (Olimex ESP32-S3-DevKit-Lipo Development Board). It looks like there is a problem with the initialisation of SPI1.setup() in the S3 build of Espruino. The error is being reported in Espruino (before the restart) as:

"spi: spi_bus_initialize(762): invalid dma channel, chip only support
spi dma channel auto-allocm"

spi_bus_initialize is called in Espruino from targets/esp32/jshardwareSpi.c
So its a similiar shape to the recent I2C issues. I have raised an issue #2601, in Github and can investigate over the next few days.

Attachments:

• Screenshot from 2025-01-23 16-01-23.png

Posted at 2025-01-24 by @gfwilliams

Thanks for looking into this @SimonGAndrews - I just posted on the GitHub issue

if it's a blocking operation either way

Yes - for higher speeds hardware SPI will be able to push data faster but you'll never hit that with PCD8544 anyway.

At some point it might make sense to make it async, but a lot of JS code expects that it is synchronous (eg setting a pin, sending by SPI, disabling that pin)

Posted at 2025-01-26 by @devsnd

@enaon for me the WiFi is also kinda flaky. I'm trying to work around it in software now, by adding retries and reconnects, but it's far from ideal.

I stumbled upon this article here (https://roryhay.es/blog/esp32-c3-super-mini-flaw) which basically says that there's a bunch of boards out there that have flawed PCB designs. I still have to get out my calipers to check if my board is one of the bad ones...

@SimonGAndrews very interesting that you can reproduce on your S3. I won't be a big help in the investigation, except for trying out any new firmwares you prepare that you want tested in a c3 board... my time is very limited currently and software SPI is good enough for my purposes, as Gordon attests.

Posted at 2025-01-27 by @SimonGAndrews

@devsnd the problem you found is in the c3 and s3 builds. I’ve found a fix for it that works in both. The testing is going ok so far. It’s great that you found it , the more people using the builds , the better to help make them stable.
Others have worked on the S3 , to get it to the point where you can build it yourself. It’s allot easier (and the only successful way for me) if you use a Linux system to do it.
I will update the issue and send a PR to Gordon, hopefully this week.

PS - that’s a good practical article you found regarding the Wi-Fi

Posted at 2025-01-27 by @gfwilliams

for me the WiFi is also kinda flaky.

Yes, I hit this too, and in the end I think it killed the RF stage of my C3 board. I wasted so much time on this thinking it was an Espruino problem - maybe try and return them and get some decent ones. There's not a bunch we can do in software

Posted at 2025-01-28 by @enaon

@enaon for me the WiFi is also kinda flaky. I'm trying to work around it in software now, by adding retries and reconnects, but it's far from ideal.

I stumbled upon this article here (https://roryhay.es/blog/esp32-c3-super-mini-flaw) which basically says that there's a bunch of boards out there that have flawed PCB designs. I still have to get out my calipers to check if my board is one of the bad ones...

thank you for you time to answer, I guess my case may not be related to the s3's issues, as I have an old esp32 wemos module that I know wifi was working fine hardware wise, I used it with moongoose os for some years now, it was stable. I will play some more, thank you again.