Hi Miro,I've posted a reply to the sourceforge bug report you mention. Let's try to continue this discussion on our launchpad page and if we can confirm it is a bug in the compiler we will look into it and provide a fix. Correctness is of the utmost importance to us, so while bugs do happen fixing them is our priority.Please note that we cannot monitor every single forum out there for people reporting bugs so if you suspect you found an issue with our toolchain in the future, please let us know either here, on our launchpad pages [1][2] or on GCC's bugzilla (only if you are confident that the bug is also present on stock GCC compiler)[1] https://bugs.launchpad.net/gcc-arm-embedded[2] https://answers.launchpad.net/gcc-arm-embedded[3] https://gcc.gnu.org/bugzilla/Best regards,Thomas [on behalf of the GNU Arm Embedded Toolchain team]
Hi Miro,
We are aware of this sub-optimal code generation issue for M0 etc. It is on our list of things to fix. Rest assured we will get to it in the near future. In the meantime, the code generated for M3 looks like T1 code and mostly valid for M0. You could compile this code snippet for M3 and link for an M0 target if you wanted a work-around.
Thanks,Tejas.
Hi Thomas,
Thank you for your reply. I was not sure where to post a suspected bug report.
I have posted the bug to: bugs.launchpad.net/.../1722849
In this bug report I was able to distill the problem to a short code example without any external references.
Thanks again,
Miro