Re: [PATCH] xen/gnttab: Shrink struct gnttab_copy_buf

[Jan Beulich <jbeulich@xxxxxxxx>]

On 03.11.2026 14:01, Andrew Cooper wrote:
> Move the booleans into earlier padding.  No functional change.
> 
> Signed-off-by: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>

Acked-by: Jan Beulich <jbeulich@xxxxxxxx>

> On arm32, it's 40 bytes before and after.  Given that we can't reuse the
> tail-padding from struct gnttab_copy_ptr, I'm not sure there's an obvious
> improvement to be had:
> 
>   xen.git/xen$ pahole xen-syms-arm32-after -E -C gnttab_copy_buf
>   struct gnttab_copy_buf {
>         struct gnttab_copy_ptr {
>                 union {
>                         /* typedef grant_ref_t -> uint32_t */ unsigned int 
> ref;          /*     0     4 */
>                         /* typedef xen_pfn_t -> uint64_t */ long long 
> unsigned int gmfn; /*     0     8 */
>                 } u;                                                          
>            /*     0     8 */
>                 /* typedef domid_t -> uint16_t */ short unsigned int domid;   
>            /*     8     2 */
>                 /* typedef uint16_t */ short unsigned int offset;             
>            /*    10     2 */
>         }ptr; /*     0    16 */
> 
>         /* XXX last struct has 4 bytes of padding */
> 
>         /* typedef uint16_t */ short unsigned int         len;                
>            /*    16     2 */
>         _Bool                      read_only;                                 
>            /*    18     1 */
>         _Bool                      have_grant;                                
>            /*    19     1 */
>         _Bool                      have_type;                                 
>            /*    20     1 */

Making these bitfields would help Arm32.

Jan

PS: Can we also get the other pending gnttab change sorted? As said,
feel free to post your patch, and I'll ack it. Alternatively an ack
on my variant would be nice.