Code and dev environment

[skew]

[srinath]

[Max]

Hi Srinath,

I think we're using the CW compiler because no-one has tested any others yet with AQ. I don't see any reason it won't work with others though.

BTW, in anyone is interested, the same compilation method described above for Linux will also work on Windows, even with the same makefile (just have to adjust the path to CW). You do need "make" and a couple other utilities, and the easiest way to get those on Windows, if you don't have Cygwin already, is to download Yagarto Tools -- http://www.yagarto.de/ (first download package).

I've also managed to set up a dev environment in Eclipse, including compiling and debugging.

If anyone wants more info, let me know.

-Max

[Jessestr]

Don't get my activation id from crossworks

[Max]

Took them ~8 hours to send mine. Check your spam if you can.

BTW no activation needed to just use their compiler/toolchain (with Make or Eclipse or whatever). I pulled the needed parts out of the CW install folder and copied them to another location (so I can remove CW later). Also the CW toolchain code is released under GPL, if I'm not mistaken, since it's based on GNU tools.

-Max

[srinath]

[Max]

I'm definitely curious about using a different toolchain... was going to investigate using Yagarto or Code Sourcery at some point, but for now just happy to have it all working in Eclipse (Crossworks just wasn't for me and I already use Eclipse for all my other programming work).

As for STLink, that's of course not a Rowley product but from STM, and it's hard to beat a $15 SWD debugging setup (using the discovery board). Especially if there is already Linux support for it too. But if someone could build something more compact for a similar price (and w/out all the damn LEDs ), that would be great.

-Max

[srinath]

My only thing against STLink is that it doesn't have a swd interface and i have to 'put together' some connection.

If a device 'conforms' to a swd standard can't there be a universal way of accessing/programming/debugging it? For example if i want to program a NXP chip, i need to buy another little gadget and so on.
It seems redundant and inefficient. Like having a car to go the beach, another one for groceries, another one for daily commutes etc

[Max]

Come now, making custom wire harnesses for everything is all part of the fun, right?

It's so hard to cut through the marketing hype sometimes... everyone wants you to buy THEIR interface/adapter/programmer/cable even though it does exactly the same thing as what you might already have. The whole SWD thing seems to be very slow to be adopted, but I did find tools and info. I think OpenOCD does support SWD now, though I haven't investigated that yet (using a GDB server from Atollic for now).

-Max

[srinath]

I am editing Meister's makefile to be Rowley free . These are the only lines that reference crossworks.

CW_PATH = /opt/crossworks_for_arm_2.2
BIN_PATH = $(CW_PATH)/gcc/arm-unknown-eabi/bin
LIBPATH = $(CW_PATH)/lib
INCPATH = $(CW_PATH)/include

CC = $(BIN_PATH)/cc1
AS = $(BIN_PATH)/as
LD = $(BIN_PATH)/ld
OBJCP = $(BIN_PATH)/objcopy

I plan to replace the 4 lines above with
CC=arm-none-eabi-gcc
#LD=arm-none-eabi-ld
LD=arm-none-eabi-gcc
AS=arm-none-eabi-as
OBJCP=arm-none-eabi-objcopy

That still leaves LIBPATH AND INCPATH. Where do i get the content in those 2 directories if i didn't want to cheat at look at crossworks. These must be arm or stm32f4 specific header files.

I have never written a makefile for an embedded target before, so i am puzzled with the following line in linkerOpts
--gc-sections -T./STM32/STM32f4.ld -Map aqv6.6.map -u_vectors
The file STM32f4.ld seems to be a memory map. Is that correct? How did Meister come up with it?