Hi Mateusz,

True, disabling NSH will reduce a lot of space, but see a simple example
with hello world running on 13.0 and on 7.22:

Same command: $ ./tools/configure.sh stm32f103-minimum/hello

13.0:

$ arm-none-eabi-size nuttx
   text   data    bss    dec    hex filename
  24011    292   1908  26211   6663 nuttx

7.22:
$ arm-none-eabi-size nuttx
   text   data    bss    dec    hex filename
  15068     84   1676  16828   41bc nuttx

It means that just to print a hello message now we require almost 10KB more.

I agree with Peter, maybe there are other places where we can improve,
without disabling LIBC_LONG_LONG.

Since the LIBC_LONG_LONG used to affect the printf family behavior,
probably that is the place where we need to optimize (Michael's suggestion
to look at picolibc is a good idea).

BR,

Alan

On Wed, Jul 8, 2026 at 12:00 PM raiden00pl <[email protected]> wrote:

> FLASH growth is less of a problem than stack usage growth. FLASH growth is
> immediately noticeable, while stack growth silently crashes the system.
> A recent example: 2KB stack isn't enough for NSH to run "ps" with certain
> system
> settings.
>
> Regarding small systems: 64KB of FLASH is too little to do anything useful
> with
> NSH, and it always has been. If you want to do something useful with such a
> small
> FLASH, the first thing you do is disable NSH.
>
> śr., 8 lip 2026 o 16:49 <[email protected]> napisał(a):
>
> > On 2026-07-08 15:06, Karel Kočí wrote:
> > > Hi
> > >
> > > My two cents:
> > >
> > > * Nobody is suggesting to break POSIX compatibility for everyone.
> > > Alan's
> > >   suggestion is to have option to break it to reduce the size.
> >
> > If memory serves, that was discussed before and the prevalent opinion
> > was that even having such option makes NuttX POSIX-noncompliant. (And
> > therefore any such option is unacceptable.)
> >
>

Reply via email to