Re: [PATCH v3 2/4] xen/arm: Handle cases when hardware_domain is NULL

[Jan Beulich <jbeulich@xxxxxxxx>]

On 12.04.2021 12:52, Luca Fancellu wrote:
> --- a/xen/include/xen/sched.h
> +++ b/xen/include/xen/sched.h
> @@ -1022,6 +1022,9 @@ static always_inline bool is_hardware_domain(const 
> struct domain *d)
>      if ( IS_ENABLED(CONFIG_PV_SHIM_EXCLUSIVE) )
>          return false;
>  
> +    if ( !d )
> +        return false;
> +
>      return evaluate_nospec(d == hardware_domain);
>  }

On v2 I did say on the respective code that was here (and my
suggestion of this alternative adjustment): "Can you point out
code paths where d may actually be NULL, and where [...] would
not behave as intended (i.e. where bad speculation would
result)?"

Since you've taken the suggestion as-is, and since the commit
message says nothing in either direction here, did you actually
verify that there's no abuse of speculation possible with this
extra return path? And did you find any caller at all which may
pass NULL into here?

Jan