Re: [Xen-devel] [PATCH] xentrace: handle sparse cpu ids correctly in xen trace buffer handling

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 30.08.18 at 09:52, <jgross@xxxxxxxx> wrote:
> @@ -202,7 +202,7 @@ static int alloc_trace_bufs(unsigned int pages)
>       * Allocate buffers for all of the cpus.
>       * If any fails, deallocate what you have so far and exit.
>       */
> -    for_each_online_cpu(cpu)
> +    for_each_present_cpu(cpu)
>      {
>          offset = t_info_first_offset + (cpu * pages);
>          t_info->mfn_offset[cpu] = offset;

Doesn't this go a little too far? Why would you allocate buffers for CPUs
which can never be brought online? There ought to be a middle ground,
where online-able CPUs have buffers allocated, but non-online-able ones
won't. On larger systems I guess the difference may be quite noticable.

Jan



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel