Freeplay CM3 GBA clone Kit.

[Dragu]

@Substring
I found a copy now. It's an utility that should write a framebuffer to the device by a certain driver. So more for testing. ili9341 is writing only newly changed parts of the framebuffer to the display, as waveshare driver is doing the whole. This should be about real 60fps instead of 30fps. As I am not able to compile and have no Linux system for this, I will just use what exists without special branches for my toy.

[Substring]

@dragu are we talking about https://github.com/juj/fbcp-ili9341 ?

[Substring]

you shouldn't have THAT many framebuffer devices ! just /dev/fb0 or /dev/fb1. The latter is usually the SPI one, so make a dirty assumption

[Dragu]

@Substring
WIth fbcp-ili9341 you are totaly right, as Juj has a working version of this, but you find no compiled executable for Freeplay from him. Perhaps I can ask him for this, if he want to share.

[Dragu]

@Substring
Anyway I get more and more confused about this, just a copy of dmesg. Means that waveshare32b.dtbo load fb-ili9340, with in config

dtoverlay=waveshare32b:rotate=270,speed=99999999,fps=60

[    2.361892] fbtft: module is from the staging directory, the quality is unknown, you have been warned.
[    2.370502] fb_ili9340: module is from the staging directory, the quality is unknown, you have been warned.
[    2.370951] fbtft_of_value: buswidth = 8
[    2.370958] fbtft_of_value: debug = 0
[    2.370963] fbtft_of_value: rotate = 270
[    2.370968] fbtft_of_value: fps = 60
[    2.370973] fbtft_of_value: txbuflen = 32768
[    2.378627] hid-generic 0003:1241:1233.0001: input,hidraw0: USB HID v1.11 Keyboard [HID 1241:1233] on usb-3f980000.usb-1.4/input0
[    2.459203] input: HID 1241:1233 as /devices/platform/soc/3f980000.usb/usb1/1-1/1-1.4/1-1.4:1.1/0003:1241:1233.0002/input/input2
[    2.528989] hid-generic 0003:1241:1233.0002: input,hidraw1: USB HID v1.11 Device [HID 1241:1233] on usb-3f980000.usb-1.4/input1
[    2.529208] usbcore: registered new interface driver usbhid
[    2.529214] usbhid: USB HID core driver
[    2.631365] graphics fb1: fb_ili9340 frame buffer, 320x240, 150 KiB video memory, 32 KiB DMA buffer memory, fps=100, spi0.0 at 99 MHz

[Substring]

@dragu have you read https://github.com/juj/fbcp-ili9341/blob/master/README.md#installation ? No dtoverlay should be specified, the driver handles everything

[Dragu]

@Substring
Yes. More than 10x
But we have not fbcp-ili9341, and my head too much small to build this.
There is also this page that described the display area with new adapted viewport :
https://github.com/juj/fbcp-ili9341/blob/master/freeplaytech_waveshare32b.h

[Substring]

@dragu I haven't checked in-depths how it works, but you might simply build it in raspbian and copy it over to recalbox

[Dragu]

@Substring
Now, problem, what version of Raspian should I need to stay compatible with Recalbox ? I have spare SD's and can always shoot one for my bartop if it will work. Otherway my spare PI's are used up.

[Substring]

@dragu the most "secure" version would be one with kernel 4.4

[Dragu]

@Substring
My bartop dont stops with crying, but I arrive to install, and activate ssh (what is not activated by default after nov16, beep) Also no cmake inside, but installed also now. And this cmake give me an error,

pi@Raspian:~/fbcp-ili9341/build $ cmake -DFREEPLAYTECH_WAVESHARE32B=ON
CMake Error: The source directory "/home/pi/fbcp-ili9341/build" does not appear to contain CMakeLists.txt.

Sure and after juj description you make mkdir build and after cd build (empty) so he cannot find this file there but a level lower, somebody make a mistake or cmake changed...
Finally I arrive, but errors

pi@Raspian:~/fbcp-ili9341/build $ cmake -DFREEPLAYTECH_WAVESHARE32B=ON /home/pi/fbcp-ili9341
-- Doing a Release build
-- Board revision: a02082
-- Detected this Pi to be one of: Pi 2B rev. 1.2, 3B, 3B+, CM3 or CM3 lite, with 4 hardware cores and ARMv8-A instruction set                  CPU.
-- Enabling optimization flags that target ARMv8-A instruction set (Pi 2B >= rev. 1.2, 3B, 3B+, CM3 or CM3 lite)
-- Scaling source image to view. If the HDMI resolution does not match the SPI display resolution, this will produce blurrines                 s. Match the HDMI display resolution with the SPI resolution in /boot/config.txt to get crisp pixel perfect rendering, or alte                 rnatively pass -DDISPLAY_CROPPED_INSTEAD_OF_SCALING=ON to crop instead of scale if you want to view the center of the screen p                 ixel perfect when HDMI and SPI resolutions do not match.
-- Preserving aspect ratio when scaling source image to the SPI display, introducing letterboxing/pillarboxing if HDMI and SPI                  aspect ratios are different (Pass -DDISPLAY_BREAK_ASPECT_RATIO_WHEN_SCALING=ON to stretch HDMI to cover full screen if you do                  not care about aspect ratio)
CMake Error at CMakeLists.txt:127 (message):
  Please define -DSPI_BUS_CLOCK_DIVISOR=<some even number> on the CMake
  command line! (see files ili9341.h/waveshare35b.h for details) This
  parameter along with core_freq=xxx in /boot/config.txt defines the SPI
  display speed.  Smaller divisor number=faster speed, higher number=slower.


-- Configuring incomplete, errors occurred!
See also "/home/pi/fbcp-ili9341/build/CMakeFiles/CMakeOutput.log".

also not normal.

[Dragu]

@Substring
Ok, I understand, he wanted one more parameter, not easy to read between lines. So he build something, must just find it

pi@Raspian:~/fbcp-ili9341/build $ cmake -DFREEPLAYTECH_WAVESHARE32B=ON -DSPI_BUS_CLOCK_DIVISOR=8 /home/pi/fbcp-ili9341
-- Doing a Release build
-- Board revision: a02082
-- Detected this Pi to be one of: Pi 2B rev. 1.2, 3B, 3B+, CM3 or CM3 lite, with 4 hardware cores and ARMv8-A instruction set CPU.
-- Enabling optimization flags that target ARMv8-A instruction set (Pi 2B >= rev. 1.2, 3B, 3B+, CM3 or CM3 lite)
-- Scaling source image to view. If the HDMI resolution does not match the SPI display resolution, this will produce blurriness. Match the HDMI display resolution with the SPI resolution in /boot/config.txt to get crisp pixel perfect rendering, or alternatively pass -DDISPLAY_CROPPED_INSTEAD_OF_SCALING=ON to crop instead of scale if you want to view the center of the screen pixel perfect when HDMI and SPI resolutions do not match.
-- Preserving aspect ratio when scaling source image to the SPI display, introducing letterboxing/pillarboxing if HDMI and SPI aspect ratios are different (Pass -DDISPLAY_BREAK_ASPECT_RATIO_WHEN_SCALING=ON to stretch HDMI to cover full screen if you do not care about aspect ratio)
-- SPI_BUS_CLOCK_DIVISOR set to 8. Try setting this to a higher value (must be an even number) if this causes problems. Display update speed = core_freq/divisor. (on Pi3B, by default core_freq=400)
-- USE_DMA_TRANSFERS enabled, this improves performance. Try running CMake with -DUSE_DMA_TRANSFERS=OFF it this causes problems, or try adjusting the DMA channels to use with -DDMA_TX_CHANNEL=<num> -DDMA_RX_CHANNEL=<num>.
-- Targeting WaveShare 3.2 inch (B) display with ILI9341
-- Configuring done
-- Generating done
-- Build files have been written to: /home/pi/fbcp-ili9341/build

make -j seems to work, incredibly

pi@Raspian:~/fbcp-ili9341/build $ make -j
Scanning dependencies of target fbcp-ili9341
[  5%] Building CXX object CMakeFiles/fbcp-ili9341.dir/fbcp-ili9341.cpp.o
[ 16%] Building CXX object CMakeFiles/fbcp-ili9341.dir/mem_alloc.cpp.o
[ 22%] Building CXX object CMakeFiles/fbcp-ili9341.dir/display.cpp.o
[ 16%] Building CXX object CMakeFiles/fbcp-ili9341.dir/diff.cpp.o
[ 27%] Building CXX object CMakeFiles/fbcp-ili9341.dir/dma.cpp.o
[ 33%] Building CXX object CMakeFiles/fbcp-ili9341.dir/ssd1351.cpp.o
[ 38%] Building CXX object CMakeFiles/fbcp-ili9341.dir/mailbox.cpp.o
[ 44%] Building CXX object CMakeFiles/fbcp-ili9341.dir/spi.cpp.o
[ 50%] Building CXX object CMakeFiles/fbcp-ili9341.dir/st7735r.cpp.o
[ 55%] Building CXX object CMakeFiles/fbcp-ili9341.dir/ili9486.cpp.o
[ 61%] Building CXX object CMakeFiles/fbcp-ili9341.dir/gpu.cpp.o
[ 66%] Building CXX object CMakeFiles/fbcp-ili9341.dir/keyboard.cpp.o
[ 72%] Building CXX object CMakeFiles/fbcp-ili9341.dir/text.cpp.o
[ 83%] Building CXX object CMakeFiles/fbcp-ili9341.dir/hx8357d.cpp.o
[ 83%] Building CXX object CMakeFiles/fbcp-ili9341.dir/statistics.cpp.o
[ 88%] Building CXX object CMakeFiles/fbcp-ili9341.dir/mz61581.cpp.o
[ 94%] Building CXX object CMakeFiles/fbcp-ili9341.dir/ili9341.cpp.o
[100%] Linking CXX executable fbcp-ili9341
[100%] Built target fbcp-ili9341

[Dragu]

@Substring
Ok, I have an executable, now somebody must tell me where is equivalent to rc.local in Recalbox ?

[Substring]

@dragu no such files. Startup scripts are in /etc/init.d. create a S02iliblabla, get the basic structure of an init file, and launch it at start.

But first, you'd better try by launching it manually through ssh and make sure it works. If ypu just write an init script, you'll never get any ettor message

[Dragu]

@Substring
Petite question, pour recompiler, vaut-il mieux d'effacer le contenu du build précédent ?

Bonne idée de le lancer en ssh. Bon, il affiche rien pour le moment, mais me sort tout un tas de messages sophistiqués

# ./fbcp-ili9341
bcm_host_get_peripheral_address: 0x3f000000, bcm_host_get_peripheral_size: 16777216, bcm_host_get_sdram_address: 0x40000000
BCM core speed: current: 400000000hz, max turbo: 400000000hz. SPI CDIV: 8, SPI max frequency: 50000000hz
Allocated DMA channel 5
Allocated DMA channel 1
Enabling DMA channels Tx:5 and Rx:1
DMA hardware register file is at ptr: 0x75b65000, using DMA TX channel: 5 and DMA RX channel: 1
DMA hardware TX channel register file is at ptr: 0x75b65500, DMA RX channel register file is at ptr: 0x75b65100
DMA RX channel 1 was assigned another peripheral map 17!
DMA channel 0 has peripheral map 0 (is lite channel: 0, currently active: 0, current control block: (nil))
DMA channel 1 has peripheral map 17 (is lite channel: 0, currently active: 1, current control block: (nil))
DMA channel 2 has peripheral map 0 (is lite channel: 0, currently active: 0, current control block: (nil))
DMA channel 3 has peripheral map 5 (is lite channel: 0, currently active: 1, current control block: 0xfe538d80)
DMA channel 4 has peripheral map 0 (is lite channel: 0, currently active: 0, current control block: (nil))
DMA channel 5 has peripheral map 0 (is lite channel: 0, currently active: 0, current control block: (nil))
DMA channel 6 has peripheral map 13 (is lite channel: 0, currently active: 1, current control block: (nil))
DMA channel 7 has peripheral map 0 (is lite channel: 1, currently active: 0, current control block: (nil))
DMA channel 8 has peripheral map 0 (is lite channel: 1, currently active: 0, current control block: (nil))
DMA channel 9 has peripheral map 0 (is lite channel: 1, currently active: 0, current control block: (nil))
DMA channel 10 has peripheral map 0 (is lite channel: 1, currently active: 0, current control block: (nil))
DMA channel 11 has peripheral map 0 (is lite channel: 1, currently active: 0, current control block: (nil))
DMA channel 12 has peripheral map 0 (is lite channel: 1, currently active: 0, current control block: (nil))
DMA channel 13 has peripheral map 0 (is lite channel: 1, currently active: 0, current control block: (nil))
DMA channel 14 has peripheral map 0 (is lite channel: 1, currently active: 0, current control block: (nil))
DMA RX channel was assigned another peripheral map!

[Dragu]

Ok, I also found this cmake option
"-DDMA_RX_CHANNEL=<num>: Specifies the DMA channel number to use for SPI receive commands. Change this if you find a DMA channel conflict."
What would you propose, Master ?

[Dragu]

Yes, we can. I used DMA channel 4 for RX, and the result

I will try to stretch a little the sides if possible, but already very nice

[Substring]

@dragu To recompile, the best is to make clean

Regarding the values, the best is probably to ask Flavor

[Dragu]

@Substring
I was more brutal using rm -r
But @Flavor dont use fbcp-ili9341, there is even no word about in the forum. Anyway, I am near the end with this one also. Pretty well working all this with standard Recalbox and just small addons. When I am finished with scaling here, there is just staying one point to close the thread.

[Dragu]

Vive la France, so I hope everybody recuperate for tomorrow....
Ok, concerning fbcp-ili9341, I am finished and adjust now the TFT screen to the viewport matching the physical window in the GBA shell. After 2 dozens of recompilations, and I have a version for all other people that will buy this kit. As you can see in the fotos, this fit. That they are round is fault of the lens.

So this story also over.

Now only thing what stays is the software for the brightness controller of the display background.