corguy
Member
I'm interested in using a STM32F103VET6 100 pin mcu in a
project, has anyone done anything along those lines with Maple?
Is it as simple as changing some pin mappings or would it
be more involved?
Also, has anyone loaded Maple on the STM Discovery board? It might
have posed the same type challenges.
zoofdxp
Member
Roberto Navoni aka Redfox74 has..see a schematic here:
http://www.slideshare.net/redfox74/ln-multipilot-32schematic
his website at: http://www.virtualrobotix.com/
He has a modified leaflabs ide called VRIDE that supports this
processor, I know that he has changed pin mappings for various
ports, so that would be good stuff to inspect/look into:
corguy
Member
Thanks, it looks like it uses a STM32F103VET6 but in the hardware file it
references a STM32F103ZET6 which is a 144 pin device. Not that it matters just
odd that the code does not match the specs.
I'll have to do some more digging but it seems like it's possible and
maybe even easy to use the 100 or 144 pin MCU's.
corguy
Member
Can anyone give me a HIGH level view of what changes would be required to support
the STM32F103VET6? It is basically changing the pin mappings in a few files? I've
looked through the Maple lib source and it seems to be doable, just want to get
others views. Thanks!
poslathian
LeafLabs
I am sure mbolivar could give you a more thorough answer (I cant) but check this thread here:
http://forums.leaflabs.com/topic.php?id=70&page=2
also check out https://github.com/leaflabs/libmaple/blob/master/notes/portable.txt
mbolivar
LeafLabs
based on this page, it looks like this is similar enough to the ZE used on the native for this to be very straightforward, if a little time consuming.
the good news is that once you've done this once, you get a solid understanding of how everything in libmaple fits together, and adding new MCUs/boards becomes a trivial matter.
to get libmaple proper working, you should only need to add the appropriate MCU config to stm32.h. that is, the only thing you should need to do for libmaple proper is add an #elif defined(MCU_STM32F103VET6) section around here in stm32.h. you can imitate the sections for the other MCUs, above.
in order to get more complete support going with our Makefile, you'll need to pick a new BOARD name (i.e., compile libmaple with BOARD=yourboardname), and:
1. modify support/make/target-config.mk for your BOARD. use the maple native parts as a starting point. the error LED stuff is for throb() in util.c. if you don't define those, then everything will still work, but the LED won't throb on assertion failures, etc., which can make debugging harder.
2. add a new yourboard.h and yourboard.cpp to wirish/boards/. maple.h and maple.cpp have the best comments about what's going on in those files, while maple_native.h and maple_native.cpp are probably the closest to what you're doing. you can read the board-specific values documentation for more info on what goes into the header file.
3. add an include for yourboard.h appropriately around here in wirish/boards.h.
4. add linker scripts for your board to support/ld/yourboardname/<flash,ram,jtag>.ld. don't panic if you've never written a linker script; you mostly just need to configure memory. see the existing board linker scripts in the subdirectories under e.g. support/ld/maple_RET6/ for examples. since the maple bootloader has not been ported to the VET6, your best bet is probably just to define a jtag.ld, which is meant for "bare metal" (i.e., no bootloader) builds. you'll need a JTAG programmer for this to work, of course.
i started a wiki page with some information on porting libmaple a while back:
http://wiki.leaflabs.com/index.php?title=Porting_libmaple
it could definitely use some more love. i'll start by adding the information in this post to that page. if you succeed in your port, please add any extra information you found useful to the wiki, so that future porters have an easier time.
hope this helps, and happy hacking!
edit: typos, added boards.h info
Update: see this commit in torfbolt's libmaple tree for an example:
https://github.com/torfbolt/libmaple/commit/24d632323be65ecbd2d636854367f8ab73f7e576
That patch is for the STM32F103RBT6 (like what's used on the Maple), so they had less work to do, as there wasn't a new MCU being added, but it should give you a starting point.
corguy
Member
Thanks guys, the only way to try this is to jump in with both feet so
I'm going to do a run in the next week or two of PCB's with the VET6
100 pin footprint so I can see what the problems are. I'm sure I'll have some
questions as I get into it.
The test PCB will not be Maple compatible as far as size, pins, etc, just
something quick and dirty and simple. I do want to make the standard boot loader
work and it will have a jtag connector and breakout a few digital/analog lines.
mbolivar
LeafLabs
corguy -- have fun!
porting the bootloader in particular should be ... interesting. it was written before libmaple existed, and hasn't changed much since. we keep saying we'll get around to rewriting it in terms of libmaple, but only so many hours, if it ain't broke, and so on.
some advice:
- get in touch with redfox74; as mentioned above, he may already have finished this
- DO NOT use the bootloader master branch as a starting point. (as of today) it's maple-only. the bootloaders for the other boards are in the the other branches (we know; sorry). the native-boot branch is the one used for maple native; i'd recommend starting there.
- use gitk to get an idea of how the different branches relate to each other. this will let you use git log to see the changes that were necessary to get the bootloader working between boards.
mbolivar
LeafLabs
The repository is a bit of a mess, is all. Over time, we've gotten to the point where adding new boards to libmaple follows a pretty well-defined procedure, and there's documentation. Adding new boards to the bootloader has been different each time (although we now know the changes necessary to make it nice and generic and all within a single branch, we haven't made them).
mbolivar
LeafLabs
I took a look at maple-bootloader and it appears that all the board changes are branches rather than forks. It's not clear to me how to pull one specific branch like maple-native. Do you get all branches when you clone maple-bootloader?
yup; as i mentioned previously:
- DO NOT use the bootloader master branch as a starting point. (as of today) it's maple-only. the bootloaders for the other boards are in the the other branches (we know; sorry). the native-boot branch is the one used for maple native; i'd recommend starting there.
just $ git checkout native-boot for the maple native branch.
to see the diff between mini and native (which may be helpful for picking out what you might need to change), do $ git diff origin/mini-boot..native-boot.
corguy
Member
Right, I was trying to follow your directions to start with just the maple-native code but that's where I did not get how git works to pull a branch. I see now, thanks for the help.
For future reference for git newbies like me:
git clone https://github.com/leaflabs/maple-bootloader.git (grab master)
git branch -a (lists all the branches)
git checkout native-boot (get the branch)
git branch (now lists both master and native-boot locally)
git diff master native-boot > /tmp/diffs.txt (gives me all the diffs in a file)
gitk --all & (starts gitk GUI to browse the files)
You must log in to post.