Mf4.2 + gcc +stm32

Has anyone tried to compile the STM32 code with GCC? It’s NO-GO.

[ol]There is some code missing in the STM32.settings file that defines the -mcpu=cortex-m3 setting
There is code missing to define that cortex-m3 doesn’t support ARM
The ASM files use “;” for comments, GCC wants "@ "(or “#”? dunno anymore)[/ol]

I solved 1 and 2, but 3 is harder.

I’m currently working from the RTM 4.2 code. Might have changed in the SVN repo. Will get that when I can…

Have you tried downloading the port from the people who made it, i dont remember its name…

I’ll be playing with it next week when I get my discovery board…

Port is in the PK - STM32Stamp made by Oberon Microsystems

Oberon wouldn’t have used GCC, I assume…

Architect, have you attempted to compile the port for your STM32 Discovery yet?

Not yet :frowning: I got myself in those free Stanford classes - advanced tracks. That is eating my free time right now.I also have that stamp board that has the port in the PK. I will try that one first.

It’s super low on flash and RAM, right? I’m skeptical that GCC can produce binaries small enough to fit…

Flash - 512K, RAM - 64K

side toppic:
Has anyone else seen the new freescale Cortex-M4 chip? Upto 200MHz, upto 4MB flash and 512K ram… ;D

No actual details yet though…


No package info either…

Ah, yeah, it’s the 64k RAM. They say they used the built-in bootloader instead of the NETMF bootloader to save space (though, that’d be flash, not RAM?)

I’m tempted to pay the $25 for the stamp board and give the Oberon port a shot. Then, on to my discovery board!

I think Oberon used RVDS to build thier port and not GCC.

Dont hold me to this but compiling and getting it to work (even on the stamp) may be challenging.

Anyways have fun.

…as an after thought may be you should hold on to that $25 and shipping and use it towards a Hydra, I am not selling for GHI but trying to save some hair pulling :smiley:

Yeah, they use RVDS. And RVDS uses a lot of compiler specific commands.

I have been struggeling the last two days to get the code to a point where GCC will compile the port. At this point I can’t even see how big the image will be.

Got the ASM files sorted. Don’t know if they still do what they are supposed to do, as I had to replace a few RVDS commands with GCC commands, and I don’t know if the substitution was correct.

Currently stuck on the GCC equivalent for defining a whole function as ASM inside a C file…

I’m 100% uninterested in Hydra, Spider, and Gadgeteer as a platform.

In my opinion, FEZ Mini is the most interesting module GHI produces. Bigger, faster, and more RAM do not excite me. Smaller, cheaper, and lower power consumption are what get me going.

STM32F205RET6 (120 MHz, 512K flash, 132K RAM, LQFP64) which sells for $11.44 in ones (and $6.45 in high quantities) is something that I’d love to see in a tiny little module. What if a module like that could be produced in quantity and sold for $19? What if a LQFP144 version (with a LOT more IO and connectivity) was available for, say, $23? That’s what I want (and it’s not a pipe dream…).

I want to have LOTS of microcontrollers doing LOTS of things, and Gadgeteer (while it is a great hobbyist and educational platform) isn’t suited to that. I don’t want one board that I put together and take apart, I want to make useful things that will live on.

@ godefroi, I know how you feel. However, I think Gadgeteer will be great for getting more people interested in microcontrollers and hopefully that ends up meaning more things on more microcontrollers. As I’m preparing a presentation for my .NET users group, Gadgeteer is going to be awesome to have when you only have ~ 1 hr to build a project from scratch and you can actually show the participants every step from concept to working prototype in that amount of time. For my own personal projects, I’m struggling to see how Gadgeteer will be as useful when my end goal is to leave the micro there permanently.

Dont get me wrong :slight_smile:

I love the ST chip set. I have applications running on the older STR7 and STR9 chips. I have applications running on the STM32 in a variety of confgurations (including the F1/F2 M3 and the F4 M4). However none of them run NETMF.

Compiling NETMF 4.2 with GCC on to these chips is going to be lot of fun. If you have invested in other compilers (RVDS / MDK) then you may get good results.

GHI was able to compress a lot of stuff into a single chip, so it is possible. However it does require investment in skills, time and resources (lots of these).

I will be very interested in a STM NETMF port with GCC that can do something usefull. Please keep us posted on your progress.

LinkSprite makes available the source for their port of NETMF for their STM32 board. It has external flash and RAM, but it looks like it might be set up for compiling with GCC:


It claims to be for the 3.1 framework but what I saw while poking around it indicated it was for the 4.0 PK.

Hmm, I dont know if things have changed that much since 4.0, but that rar only contains the board’s port. It doesn’t include the actual CPU core stuff…

Looked further.
Looks like it uses the core files for CortexM3. Is that still part of PK? Why does the STM32F use it’s own core files and not CortexM3?

Will have to look on Monday…

Unrelated question. Since you are working with PK, did you contact GHI about fez hydra?

Unrelated answer.

Only started with PK this week. Too soon to take liberties… :slight_smile: