Hi KR,

What avr-gcc version are you using?

I'm getting this error: "avr-gcc: error: device-specs/specs-avr128da28: No
such file or directory"

I found a reference to it here:
https://www.avrfreaks.net/s/topic/a5C3l000000UlWyEAK/t193338

I downloaded the Microchip.AVR-Dx_DFP.2.7.321.atpack file renamed it to
Microchip.AVR-Dx_DFP.2.7.321.zip and extract it, but I think it is
something ot xc8 from Microchip...

Are you using Linux as well? I'm using Ubuntu and just ran: "sudo apt
install avr-gcc". Could you please

BR,

Alan

On Sun, Apr 27, 2025 at 7:49 AM <kr....@kerogit.eu> wrote:

> Hi,
>
> thanks for the response and for the PR on GitHub.
>
> > Ok. If possible, could you explain the issue with github?
>
> Well, it's a personal thing. Aside from the usual gripe about big
> corporation there is also a clause in their Terms of Service that
> requires the user to "agree to submit to the exclusive jurisdiction and
> venue of the courts located in the City and County of San Francisco,
> California." And that's certainly something I don't agree with.
>
> > I think in this case we could add _bm and _bp as exceptions as well.
>
> The _bm case is the most prevalent in the current source so that's a
> great help. There are currently two cases of _gp and around 10 cases of
> _gc so that should be manageable to redefine in the architecture code.
> It also seems like I missed a 5th suffix - _gm - but that also has
> around 10 uses currently.
>
> Even with the bm exception though, the nxstyle tools reports mixed case
> errors. Apparently, it cannot apply the exception for the line of code
> which looks like this:
>
> ret = (VPORTA.IN & (PIN2_bm | PIN3_bm)) >> 2;
>
> Splitting it into two lines works though and it should yield the same
> result in the machine code.
>
> ret = (VPORTA.IN & (PIN2_bm | PIN3_bm));
> ret >>= 2;
>
> I'll incorporate these changes to the next revision of the patch series
> and also submit a patch for the nxstyle.c
>
> >
> >> 2. These patches depend on "nuttx/clock: make NSEC_PER_USEC and others
> >> long" (ac42add946) being applied. It is only present in NuttX master
> >> branch and needs to be cherry-picked for testing on the branch based
> >> on
> >> release 12.9
> >>
> >>
> > Could you please give more details here?
>
> Since the fixes were applied to the master branch, I expect the
> remaining patches in the series to go there as well - in which case
> everything is sorted and nothing else needs to be done, the patch
> ac42add946 is present there. The issue why that patch is needed is
> explained in
> https://lists.apache.org/thread/xzor3tjy9jozlmox82frd27v969b37l4 . In
> short - without this patch, avr-gcc will produce malfunctioning code
> when converting between ticks and time.
>
> Hope I didn't miss any questions, please point it out if I did. Other
> than that, I will try to submit second version of the patch series
> during the next week.
>
> Thanks for the review.
>

Reply via email to