Forum : ARM
| Post Information | Post |
|---|---|
|
August 12, 2009 - 7:41pm
|
Hello, I'm using the lastest version of Ride7 7.20.09.0162 and I have an STM3210E EVAL board which contains the STM32F103ZET6 processor. Now no matter what program I create the stack always seems to start(end) at 0x20005000 in a RAM build. This is proving a problem as the stack quickly overwrites my code now that my code is almost 0x5000 bytes. How can I change the initial stack pointer to closer to the 32K limit (i.e. 0x20010000)? (NOTE: I am still using the free RLink version at the moment until I exceed 32K) I have tried many methods of changing the startup and LD scripts and changing the minimum stack size and the estack value etc, but nothing seems to change the stack start(end) position. Any help would be very much apprecited. |
Hi,
I see this bug here. Thanks for reporting it.
It seems that the compile and link is fine, but the problem is in the debugger. This is why you could not solve it by playing with link options and linker scripts. Note that the problem appears only in RAM mode.
I will need more time to investigate it further and I will correct it for the next release of Ride. In the meantime, the simplest thing for you would be to work in Flash mode, which does not show the problem. The 32K limit that you have because you are working with a standard RLink does allow you to debug applications that use 32K of Flash and 32K of RAM, so it should be fine for you in Flash mode. And probably better, in fact. ;)
Is it possible for you to work in Flash mode?
If not, then you can work around the problem by modifying the startup. I can explain how, but it will not be simple. So working in Flash mode should be preferred.
Best Regards,
Vincent
Vincent,
Thank-you so much for confirming this so fast. I'm starting to really like Raisonance for their excellent customer service (being British we're not very good at it ourselves).
I am able to work in Flash and the fact that this bug existed meant I really had to get involved with the Linker scripts and understand them, which can only be a good thing.
I look forward to the next release.
Je vous remercie de votre aide
Adam