The patches are submitted
https://github.com/apache/nuttx/pull/16365

Thanks for enhancing the AVR platform

Best regards
Alin

On Tue, May 13, 2025 at 7:44 AM <kr....@kerogit.eu> wrote:

> Hello,
>
> I would like to submit some fixes/improvements for the AVR DA/DB family
> according to previous discussion in this thread. Summary of the changes
> follows:
>
> Patch 0004 - AVR_LINUXGCC_TOOLCHAIN_IS_GCC configuration option is now
> removed. Instead, AVR_LINUXGCC_TOOLCHAIN selects ARCH_TOOLCHAIN_GCC
> directly. Default configurations for all chips (except the DA/DB family)
> adds CONFIG_DEBUG_OPT_UNUSED_SECTIONS set to N to prevent linker from
> removing seemingly unused sections that need to be present in the
> resulting binary. (To be removed when the corresponding linker script is
> checked.) I tested this by building nsh configuration for
> mega1284p-xplained. As far as I could see, the only difference in the
> binary was addresses of some global variables caused by different length
> of g_version.
>
> Patch 0001 - I also replaced Hello, World configuration for breadxavr
> board with a NSH one - it provides better starting point for further
> testing and development in my opinion.
>
> Patch 0003 - This is a small fix in breadxavr board Kconfig file - the
> code for buttons depends on AVR_AVRDX_GPIO_ISR_MUX being set but the
> depends clause did not reflect that.
>
> Patches 0002 and 0005 - Finally, I tried to make some improvements in
> documentation of both the breadxavr board and the DA/DB chips.
>
> All patches are attached to this message and are also available in the
> Git repository linked above in a branch named avrdx_fixes. Comments are
> welcomed. If these changes are deemed acceptable, I would again like to
> ask someone with GitHub account to open a PR.
>
> On 2025-05-08 17:23, kr....@kerogit.eu wrote:
> > Hello,
> >
> > On 2025-05-07 19:02, Xiang Xiao wrote:
> >> Yes, it's better to not introduce AVR_LINUXGCC_TOOLCHAIN_IS_GCC since:
> >>
> >>    1. Other arch doesn't have this similar option
> >>    2. For compatibility, we can either unselect
> >> DEBUG_OPT_UNUSED_SECTIONS
> >>    in defconfig or verify ld add KEEP correctly
> >>
> >
> > Alright, I'll remove that and unselect DEBUG_OPT_UNUSED_SECTIONS in
> > existing configurations. Since the patch series is already merged, I'll
> > post a patch that alters this in the master branch.
> >
> > I'll also remove the hello board configuration and replace it with nsh,
> > that seems a bit more useful. Will also try to provide some
> > instructions for working with the chip.
> >
> > Should be able to get to it during next week at most.
> >
> > Thank you for the review and help.
>

Reply via email to