Re: [Qemu-devel] [PATCH V16 09/11] NUMA: set guest numa nodes memory policy

[Wanlong Gao]

On 11/27/2013 10:35 PM, Paolo Bonzini wrote:
> Il 22/11/2013 03:08, Wanlong Gao ha scritto:
>> +static int set_node_mem_policy(int nodeid)
>> +{
>> +#ifdef __linux__
>> +    void *ram_ptr;
>> +    RAMBlock *block;
>> +    ram_addr_t len, ram_offset = 0;
>> +    int bind_mode;
>> +    int i;
>> +
>> +    QTAILQ_FOREACH(block, &ram_list.blocks, next) {
>> +        if (!strcmp(block->mr->name, "pc.ram")) {
> 
> This is not acceptable, "pc.ram" is a board-specific name.
> 
> I think instead set_node_mem_policy could be something like
> 
> int memory_region_set_mem_policy(MemoryRegion *mr,
>                                  uint64_t start, uint64_t length,
>                                  uint64_t offset);
> 
> that applies the NUMA policies specified for [offset, offset+length) to
> the host physical address [ptr+start, ptr+start+length), where ptr is
> memory_region_get_ram_ptr(mr).
> 
> Each board then can call the function after it adds RAM with
> memory_region_add_subregion.

Got it, than you.

Thanks,
Wanlong Gao


> 
> Paolo
> 
>> +            break;
>> +        }
>> +    }
>> +
>> +    if (block->host == NULL) {
>> +        return -1;
>> +    }
>> +
>> +    ram_ptr = block->host;
>> +    for (i = 0; i < nodeid; i++) {
>> +        len = numa_info[i].node_mem;
>> +        ram_offset += len;
>> +    }
>> +
>> +    len = numa_info[nodeid].node_mem;
>> +    bind_mode = node_parse_bind_mode(nodeid);
>> +    unsigned long *nodes = numa_info[nodeid].host_mem;
>> +
>> +    /* This is a workaround for a long standing bug in Linux'
>> +     * mbind implementation, which cuts off the last specified
>> +     * node. To stay compatible should this bug be fixed, we
>> +     * specify one more node and zero this one out.
>> +     */
>> +    unsigned long maxnode = find_last_bit(nodes, MAX_NODES);
>> +    if (syscall(SYS_mbind, ram_ptr + ram_offset, len, bind_mode,
>> +                nodes, maxnode + 2, 0)) {
>> +            perror("mbind");
>> +            return -1;
>> +    }
> 
> Also, it's still not clear to me why we're not using libnuma.
> 
>> +#endif
>> +
>> +    return 0;
>> +}
>> +
>>  void set_numa_modes(void)
>>  {
>>      CPUState *cpu;
>> @@ -240,4 +319,11 @@ void set_numa_modes(void)
>>              }
>>          }
>>      }
>> +
>> +    for (i = 0; i < nb_numa_nodes; i++) {
>> +        if (set_node_mem_policy(i) == -1) {
>> +            fprintf(stderr,
>> +                    "qemu: can not set host memory policy for node%d\n", i);
>> +        }
>> +    }
>>  }
>>
> 
>