Re: [PATCH 11/12] xen/hypfs: add scheduling granularity entry to cpupool entries

[Jan Beulich <jbeulich@xxxxxxxx>]

On 26.10.2020 10:13, Juergen Gross wrote:
> @@ -1057,6 +1063,43 @@ static struct hypfs_entry 
> *cpupool_dir_findentry(struct hypfs_entry_dir *dir,
>      return hypfs_gen_dyndir_entry_id(&cpupool_pooldir, id);
>  }
>  
> +static int cpupool_gran_read(const struct hypfs_entry *entry,
> +                             XEN_GUEST_HANDLE_PARAM(void) uaddr)
> +{
> +    const struct hypfs_dyndir_id *data;
> +    struct cpupool *cpupool;

const?

> +    const char *name = "";
> +
> +    data = hypfs_get_dyndata();
> +    if ( !data )
> +        return -ENOENT;
> +
> +    spin_lock(&cpupool_lock);
> +
> +    cpupool = __cpupool_find_by_id(data->id, true);
> +    if ( cpupool )
> +        name = sched_gran_get_name(cpupool->gran);
> +
> +    spin_unlock(&cpupool_lock);
> +
> +    if ( !cpupool )

May I suggest to check !*name here, to avoid giving the impression
of ...

> +        return -ENOENT;
> +
> +    return copy_to_guest(uaddr, name, strlen(name) + 1) ? -EFAULT : 0;

... success (but an empty name) in this admittedly unlikely event?

Jan