On 05.10.2026 18:59, Roger Pau Monné wrote:
> On Wed, Sep 02, 2026 at 08:35:47AM +0200, Jan Beulich wrote:
>> While _set_status_v2() indeed doesn't alter the grant_entry_header_t it
>> is handed a pointer to, _set_status_v1() does. Drop the const from the
>> parameter of the latter (and then necessarily also from _set_status()'s),
>> while adding const to the local variable of the former.
>>
>> No functional change.
>>
>> Signed-off-by: Jan Beulich <[email protected]>
>
> Acked-by: Roger Pau Monné <[email protected]>
Thanks.
> Could we also use a union in the functions themselves to avoid the cast?
>
> union {
> const grant_entry_header_t *shah;
> uint32_t *raw;
> } grant = { .shah = shah };
Possibly. However, irrespective of us using such all over the place, it
still is - strictly speaking - UB.
>> ---
>> Really I think it would be best if we did away with the raw_shah local
>> variables (which looks reasonably simple for at least _set_status_v2()).
>> I'm unconvinced that we really need to use ACCESS_ONCE() here. Torn reads
>> aren't a problem; what we require is that we look at a stable local copy,
>> and that can be achieved by putting barrier() after the reads.
>
> Possibly - I would be fine with that, which kinds of render my
> previous suggestion to use a union not very useful, if we can get rid
> of those variables in the first place.
I'll add a todo item to my list to give this a try.
Jan
