How would I add an attribute to a QSTR def

[8888clockradio]

In my python script for the mimxrt port, how would I add an attribute to all the data generated for the frozen state?

I would want to add either FLASHMEM or EXTMEM to all of them. How would I pass a compiler attribute to the QSTR's that are generated. They look like they're macros so I have to dig really deep probably. The root definition of QSTR. Ideally I would like too FLASHMEM QSTR(def), but that doesn't work. Maybe I can convert them to const objtype loose definitions instead of using the QSTR macro. Please help, I'm porting libraries for the cross compiler/mpcross. I can upload some of these, I have the latest numpy ported but all these defs fit into RAM and there's not enough RAM for the QSTRs if I could place them in FLASHMEM or EXTMEM I would be really happy. I also have Kivy ported, that one was easy since it's current python syntax. I could upload these libraries If you would like somewhere. Numpy took like 4 hours to port.

[8888clockradio]

#define FLASHMEM attribute((section(".flashmem")))
#define EXTMEM attribute((section(".externalram")))

these are my attribute DEFS

[8888clockradio]

I would be nice to know the data type of a qstr I may be just qstr but that might help me figure out how to do it with instantiation instead of the macro.

[robert-hh]

Which board/hardware do you use?

[8888clockradio]

teensy 41
QDEF(MP_QSTRnull, 0, 0, "")
QDEF(MP_QSTR_, 5381, 0, "")
QDEF(MP_QSTR___dir__, 36730, 7, "dir")
this stuff

[8888clockradio]

I may release my python build it supports hard double floats right now too I had to upload the qstrdef generator too

[robert-hh]

The MIMXRT port uses already the hard double float setting, expect for the i.MX RT 1010 and 1015 MCU, which do not have a hardware FPU.

[jimmo]

@8888clockradio I'm not quite sure what you're trying to do. QSTR's should be pretty much automatic. All you need to do is use MP_QSTR_foo anywhere in your C code, and the foo QSTR will be added to the build automatically.

There are a small number of cases where you need to define a QSTR manually if it has special characters in it.

I assume you've seen https://docs.micropython.org/en/latest/develop/qstr.html

[robert-hh]

Support for the external serial flash and RAM slots of the TEENSY41 is on my list, but not at a high rank. If you need more RAM space, you could use the SEEED ARCH MIX board with 32MByte parallel attached RAM.

[8888clockradio]

text data bss dec hex filename
5510224 60807 82396 5653427 5643b3 build-SHAREDLIB/micropython.elf
/Users/georgerosar/Documents/GitHub/iridescentCoconutSynth/arm-gnu-toolchain-12.2.mpacbti-bet1-darwin-x86_64-arm-none-eabi/bin/arm-none-eabi-size build-SHAREDLIB/micropython.elf
text data bss dec hex filename
5510224 60807 82396 5653427 5643b3 build-SHAREDLIB/micropython.elf
/Users/georgerosar/Documents/GitHub/iridescentCoconutSynth/arm-gnu-toolchain-12.2.mpacbti-bet1-darwin-x86_64-arm-none-eabi/bin/arm-none-eabi-objcopy -O ihex -R .eeprom build-SHAREDLIB/micropython.elf build-SHAREDLIB/micropython.hex
#-/Applications/Teensyduino.app/Contents/Java/hardware/tools/teensy_post_compile -file=/Users/georgerosar/Documents/GitHub/iridescentCoconutSynth/iridescentmicropython/ports/libmicropython/build-SHAREDLIB/micropython -path=/Users/georgerosar/GitHub/iridescentCoconutSynth/iridescentmicropython/ports/libmicropython -tools=/Applications/Teensyduino.app/Contents/Java/hardware/tools
/Applications/Teensyduino.app/Contents/Java/hardware/tools/teensy_size build-SHAREDLIB/micropython.elf
teensy_size: Memory Usage on Teensy 4.1:
teensy_size: FLASH: code:4103800, data:52115, headers:9072 free for files:3961477
teensy_size: RAM1: variables:49783, code:4220, padding:28548 free for local variables:441737
teensy_size: RAM2: variables:30528 free for malloc/new:493760
teensy_size: EXTRAM: variables:32

[robert-hh]

So it looks like you made good progress with your adaption of the port. If you thing that some of that would be helpful for the port, pleas make a pull request. Do not forget to verify, that change works on all supported boards.

[8888clockradio]

So it looks like you made good progress with your adaption of the port. If you thing that some of that would be helpful for the port, pleas make a pull request. Do not forget to verify, that change works on all supported boards.

I have my own git server, and it's public. However my code changes are too much to upload to GitHub. If you're interested in details send your email address. I'll tell you how to download my iridescentmicropython makefile builder

[8888clockradio]

So it looks like you made good progress with your adaption of the port. If you thing that some of that would be helpful for the port, pleas make a pull request. Do not forget to verify, that change works on all supported boards.

It's something like 77GB on my git

[8888clockradio]

So it looks like you made good progress with your adaption of the port. If you thing that some of that would be helpful for the port, pleas make a pull request. Do not forget to verify, that change works on all supported boards.

I basically have a whole graphical OS in development. I have to make a board with a big enough flash ram

[8888clockradio]

I'm getting this right now? Too many qstrings for the datatype?

build-SHAREDLIB/frozen_content.c:7040465: warning: unsigned conversion from 'int' to 'short unsigned int' changes value from '167431' to '36359' [-Woverflow]
7040465 |     MP_QSTR__hyphen__hyphen_onedir,
        | 
build-SHAREDLIB/frozen_content.c:7040466: warning: unsigned conversion from 'int' to 'short unsigned int' changes value from '167432' to '36360' [-Woverflow]
7040466 |     MP_QSTR__hyphen__hyphen_onefile,
        | 
build-SHAREDLIB/frozen_content.c:7040468: warning: unsigned conversion from 'int' to 'short unsigned int' changes value from '167433' to '36361' [-Woverflow]
7040468 |     MP_QSTR_test_pyinstaller,
        | 
build-SHAREDLIB/frozen_content.c:7040469: warning: unsigned conversion from 'int' to 'short unsigned int' changes value from '104582' to '39046' [-Woverflow]
7040469 |     MP_QSTR_importorskip,
        | 
build-SHAREDLIB/frozen_content.c:7040471: warning: unsigned conversion from 'int' to 'short unsigned int' changes value from '95700' to '30164' [-Woverflow]
7040471 |     MP_QSTR_with_name,
        | 
build-SHAREDLIB/frozen_content.c:7040473: warning: unsigned conversion from 'int' to 'short unsigned int' changes value from '134928' to '3856' [-Woverflow]
7040473 |     MP_QSTR__hyphen__hyphen_workpath,
        | 
build-SHAREDLIB/frozen_content.c:7040474: warning: unsigned conversion from 'int' to 'short unsigned int' changes value from '71478' to '5942' [-Woverflow]