Re: [PATCH] x86: correct socket_cpumask allocation for AP

[Boris Ostrovsky <boris.ostrovsky@oracle.com>]

On 07/08/2015 11:38 AM, Jan Beulich wrote:
>>>> On 08.07.15 at 17:11, <dario.faggioli@citrix.com> wrote:
>> On Wed, 2015-07-08 at 13:38 +0100, Jan Beulich wrote:
>>>>>> On 08.07.15 at 11:36, <chao.p.peng@linux.intel.com> wrote:
>>>> @@ -84,11 +85,21 @@ void *stack_base[NR_CPUS];
>>>>   static void smp_store_cpu_info(int id)
>>>>   {
>>>>       struct cpuinfo_x86 *c = cpu_data + id;
>>>> +    unsigned int socket;
>>>>   
>>>>       *c = boot_cpu_data;
>>>>       if ( id != 0 )
>>>> +    {
>>>>           identify_cpu(c);
>>>>   
>>>> +        socket = cpu_to_socket(id);
>>>> +        if ( !socket_cpumask[socket] )
>>>> +        {
>>>> +            socket_cpumask[socket] = secondary_socket_cpumask;
>>>> +            secondary_socket_cpumask = NULL;
>>> I don't think this will build with small enough NR_CPUS.
>>>
>> And it *does* *not* fix the issue on my box.
> I.e. bad analysis (albeit it seemed correct to me) _and_ new code
> not tested. Chao, one more try, or we'll have to revert the whole
> series (closing the window for it for 4.6).

FWIW, this fixes it for me:

diff --git a/xen/arch/x86/smpboot.c b/xen/arch/x86/smpboot.c
index c73aa1b..a93c723 100644
--- a/xen/arch/x86/smpboot.c
+++ b/xen/arch/x86/smpboot.c
@@ -707,6 +707,8 @@ static int cpu_smpboot_alloc(unsigned int cpu)
      unsigned long stub_page;
      unsigned int socket = cpu_to_socket(cpu);

+    socket = cpu >> 4;
+
      if ( node != NUMA_NO_NODE )
          memflags = MEMF_node(node);



But I suspect this is not what you want ;-)

-boris