nuttx: Suppress UNUSED potential warnings

[pbarada]

Summary

Update UNUSED(a) for GCC to cast 'a' to void to suppress potential "address of 'xxx' will always evaluate to true" warning when enabling warnings in system headers.

Impact

User experience: No adaptation required.
Build: No impact.
Hardware: Should not affect GCC builds.
Documentation: No impact.
Security: No impact.
Compatibility: No impact.

Testing

Build Host:

• OS: Ubuntu 24.04.4 LTS
• Compiler: arm-none-eabi-gcc 13.2.1

Target: None
Built all ARM builds using modified tools/testbuild.sh script with increased warnings (-Wall -Wextra -Wsystem-headers), verified no warnings from invocations of UNUSED() macro.

[raiden00pl]

@xiaoxiang781216 what are the consequences of this change? looking at git history, changes in this macro were made by Xiaomi, so they probably had some purpose:

1. change from ((void)(a)) to ((void)(1 || (a))):
   fs/lfs: Update version to 2.4.0 #3212

commit 3813634dc98f3b77122bb1fcc17f818f08521225
Author: Xiang Xiao <xiaoxiang@xiaomi.com>
Date:   Mon Mar 29 14:08:03 2021 +0800

    Refine UNUSED macro to avoid any side effect
    
    Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
    Change-Id: I35d858fc6fb2e0e24b993ab7dd2203c9dd35232a

2. change from ((void)(1 || (a))) to ((void)(1 || &(a))):
   optimize UNUSED macro #6779

commit 0083a2e4f29ac8574eeda895803f2a7447d89680
Author: liyi <liyi25@xiaomi.com>
Date:   Thu Aug 4 14:54:31 2022 +0800

    optimize UNUSED macro
    
    Signed-off-by: liyi <liyi25@xiaomi.com>

[xiaoxiang781216]

@xiaoxiang781216 what are the consequences of this change? looking at git history, changes in this macro were made by Xiaomi, so they probably had some purpose:

1. change from ((void)(a)) to ((void)(1 || (a))):
   fs/lfs: Update version to 2.4.0 #3212

commit 3813634dc98f3b77122bb1fcc17f818f08521225
Author: Xiang Xiao <xiaoxiang@xiaomi.com>
Date:   Mon Mar 29 14:08:03 2021 +0800

    Refine UNUSED macro to avoid any side effect
    
    Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
    Change-Id: I35d858fc6fb2e0e24b993ab7dd2203c9dd35232a

my change try to avoid evaluating the argument to generate the side effect, but @pbarada said other OS(e.g. Linux) always evaluate the argument. so it's fine restore the change to align the behaviour.

[pbarada]

Its not that the expression generates code to evaluate it, more that the semantics of the expression involving the address of the variable always evaluates to true - which GCC potentially will warn about. Note this change is only enabled for GCC(and derivatives).