Re: [PATCH v4] builtin/gc: correct total_ram calculation with HAVE_BSD_SYSCTL

From: Junio C Hamano <gitster@pobox.com>
Date: Jul 7, 2025, 13:39 UTC
Message-ID: <xmqq8ql06ry9.fsf@gitster.g>
In-Reply-To: <20250703080111.64276-1-carenas@gmail.com>

Carlo Marcelo Arenas Belón <carenas@gmail.com> writes:

Show 17 quoted lines

>  #elif defined(HAVE_BSD_SYSCTL) && (defined(HW_MEMSIZE) || defined(HW_PHYSMEM) || defined(HW_PHYSMEM64))
> -	int64_t physical_memory;
> +	uint64_t physical_memory = 0;
>  	int mib[2];
>  	size_t length;
>  
> @@ -529,9 +529,18 @@ static uint64_t total_ram(void)
>  # else
>  	mib[1] = HW_PHYSMEM;
>  # endif
> -	length = sizeof(int64_t);
> -	if (!sysctl(mib, 2, &physical_memory, &length, NULL, 0))
> +	length = sizeof(physical_memory);
> +	if (!sysctl(mib, 2, &physical_memory, &length, NULL, 0)) {
> +# ifndef __LITTLE_ENDIAN__
> +		if (length == 4) {
> +			unsigned mem;

Don't we guarantee that uint32_t is always available? If not, we should and use that type instead of "unsigned", I think.

Also I do not quite see the point of limiting this fallback to little-endian architectures. It wouldn't be called from any performance critical codepaths, would it?

Thanks.

Show 8 quoted lines

> +			if (!sysctl(mib, 2, &mem, &length, NULL, 0))
> +				physical_memory = mem;
> +		}
> +# endif
>  		return physical_memory;
> +	}
>  #elif defined(GIT_WINDOWS_NATIVE)
>  	MEMORYSTATUSEX memInfo;