Re: [PATCH v3] x86/domain: adjust limitation on shared_info allocation below 4G

[Jan Beulich <jbeulich@xxxxxxxx>]

On 05.02.2026 09:03, Roger Pau Monne wrote:
> --- a/xen/arch/x86/pv/domain.c
> +++ b/xen/arch/x86/pv/domain.c
> @@ -247,6 +247,34 @@ int switch_compat(struct domain *d)
>      d->arch.has_32bit_shinfo = 1;
>      d->arch.pv.is_32bit = true;
>  
> +    /*
> +     * For 32bit PV guests the shared_info machine address must fit in a 
> 32-bit
> +     * field within the guest's start_info structure.  We might need to free
> +     * the current page and allocate a new one that fulfills this 
> requirement.
> +     */
> +    if ( virt_to_maddr(d->shared_info) >> 32 )
> +    {
> +        shared_info_t *prev = d->shared_info;
> +
> +        d->shared_info = alloc_xenheap_pages(0, MEMF_bits(32));
> +        if ( !d->shared_info )
> +        {
> +            d->shared_info = prev;
> +            rc = -ENOMEM;
> +            goto undo_and_fail;
> +        }
> +        put_page(virt_to_page(prev));
> +        clear_page(d->shared_info);
> +        share_xen_page_with_guest(virt_to_page(d->shared_info), d, SHARE_rw);
> +        /*
> +         * Ensure all pointers to the old shared_info page are replaced.  
> vCPUs
> +         * below XEN_LEGACY_MAX_VCPUS will stash a pointer to
> +         * shared_info->vcpu_info[id].
> +         */

To cope with the toolstack-dependent ordering of actions, may I suggest "... may
have stashed ..." instead?

Jan

> +        for_each_vcpu ( d, v )
> +            vcpu_info_reset(v);
> +    }
> +
>      for_each_vcpu( d, v )
>      {
>          if ( (rc = setup_compat_arg_xlat(v)) ||