Re: [VOTE] Apache NuttX 12.13.0 RC0 release

[Tomek CEDRO <to...@cedro.info>]

On Tue, Mar 31, 2026 at 3:15 PM <kr...@kerogit.eu> wrote:
>
> 0 for AVR - AVR128DA28
>
> $ tools/configure.sh -l breadxavr:nsh
>    Copy files
>    Select CONFIG_HOST_LINUX=y
>    Refreshing...
> ...
>
> $ make -j16
> ...
> clock/clock_delay.c:70:1: error: static assertion failed: "Configure
> BOARD_LOOPSPERMSEC to non-default value."
>     70 | static_assert(CONFIG_BOARD_LOOPSPERMSEC != -1,
>        | ^~~~~~~~~~~~~
> CC:  group/group_signal.c make[1]: *** [Makefile:65: clock_delay.o]
> Error 1
> ...
>
> Set the value to something else, try again
>
> $ make -j16
> ...
> CC:  misc/lib_openat.c avr/avr_schedulesigaction.c: In function
> ‘up_schedule_sigaction’:
> avr/avr_schedulesigaction.c:98:15: error: ‘struct xcptcontext’ has no
> member named ‘saved_pc0’
>     98 |       tcb->xcp.saved_pc0  = up_current_regs()[REG_PC0];
>        |               ^
> avr/avr_schedulesigaction.c:99:15: error: ‘struct xcptcontext’ has no
> member named ‘saved_pc1’
>     99 |       tcb->xcp.saved_pc1  = up_current_regs()[REG_PC1];
>        |               ^
> avr/avr_sigdeliver.c: In function ‘avr_sigdeliver’:
> avr/avr_sigdeliver.c:64:19: error: ‘struct tcb_s’ has no member named
> ‘sigdeliver’
>     64 |         rtcb, rtcb->sigdeliver, rtcb->sigpendactionq.head);
>        |                   ^~
> avr/avr_sigdeliver.c:64:37: error: ‘struct tcb_s’ has no member named
> ‘sigpendactionq’
>     64 |         rtcb, rtcb->sigdeliver, rtcb->sigpendactionq.head);
>        |                                     ^~
> In file included from avr/avr_sigdeliver.c:31:
> avr/avr_sigdeliver.c:65:19: error: ‘struct tcb_s’ has no member named
> ‘sigdeliver’
>     65 |   DEBUGASSERT(rtcb->sigdeliver != NULL);
>        |                   ^~
>
> Reconfigure "Signal support level" to "Enable full signal support" in
> RTOS features, try again
>
> $ make -j16
> Create version.h
> LD: nuttx
> Memory region         Used Size  Region Size  %age Used
>             flash:       51215 B       128 KB     39.07%
>              sram:        1004 B        16 KB      6.13%
>            eeprom:           0 B        512 B      0.00%
>            rodata:         592 B         4 KB     14.45%
> CP: nuttx.hex
> CP: nuttx.asm
>
> $ minicom
> NuttShell (NSH) NuttX-12.13.0
> nsh>
>
> In other words - cannot currently be built out of the box now but
> workarounds/solutions exist. (Taking that into consideration together
> with the fact that that I am probably the only person trying to use
> NuttX with this chip, anything more negative than 0 vote does not seem
> appropriate.)


Yes this value should be measured and provided in board defconfig to
produce valid timings, please take a look:

https://nuttx.apache.org/docs/latest/applications/examples/calib_udelay/index.html

Can you please measure and provide valid value for that board
configurations? :-)

--
CeDeROM, SQ7MHZ, http://www.tomek.cedro.info

LOOPSPERMSEC on AVR board (was: [VOTE] Apache NuttX 12.13.0 RC0 release)

[kr...@kerogit.eu]

> Yes this value should be measured and provided in board defconfig to
> produce valid timings, please take a look:
> 
> https://nuttx.apache.org/docs/latest/applications/examples/calib_udelay/index.html
> 
> Can you please measure and provide valid value for that board
> configurations? :-)
> 
> --
> CeDeROM, SQ7MHZ, http://www.tomek.cedro.info

Hi, thanks for the reply. I did spot the relevant discussion here on 
mailing list when this change was introduced but - in my opinion - there 
is no correct value that can be provided. I went into more details here 
- https://lists.apache.org/thread/xjpyjb8r2v50bpx1j8c50podr90wr6t6 
(starting with paragraph 3)

As far as I can see, there is one other architecture - 
arch/risc-v/src/mpfs - that does not use CONFIG_BOARD_LOOPSPERMSEC for 
up_udelay (or at least it tries not to - again, more in that thread and 
the one linked there.)