Re: [PATCH v3 1/3] xen/device-tree: Parse 'cpu-map' node for CPU topology exploration

[Jan Beulich <jbeulich@xxxxxxxx>]

On 01.07.2026 22:28, Hirokazu Takahashi wrote:
>> On 01.07.2026 13:11, Hirokazu Takahashi wrote:
>>>>> +#else /* CONFIG_CPU_TOPOLOGY */
>>>>> +
>>>>> +#define cpu_topology ((struct cpu_topology *)NULL)
>>>>
>>>> Why exactly is this needed? It very much looks as if its presence may hide
>> bugs.
>>>
>>> In the current code, the condition if ( cpu_topology ) is used to cover both
>> cases:
>>> when CONFIG_CPU_TOPOLOGY is disabled, and when memory allocation for
>>> topology table fails.
>>>
>>> The next patch uses if ( cpu_topology ).
>>>
>>> static int setup_cpu_sibling_map(int cpu)
>>> {
>>>     if ( !zalloc_cpumask_var(&per_cpu(cpu_sibling_mask, cpu)) ||
>>>          !zalloc_cpumask_var(&per_cpu(cpu_core_mask, cpu)) )
>>>         return -ENOMEM;
>>>
>>>     if ( cpu_topology )
>>
>>     if ( IS_ENABLED(CONFIG_GENERIC_CPU_TOPOLOGY) )
> 
> This line should be if ( IS_ENABLED(CONFIG_GENERIC_CPU_TOPOLOGY) && 
> cpu_topology )
> because 'cpu_topology' can be NULL even if CONFIG_GENERIC_CPU_TOPOLOGY is on.
> 
> void __init init_cpu_topology(void)
> {
>     unsigned int cpu;
> 
>     cpu_topology = xzalloc_array(struct cpu_topology, nr_cpu_ids);
>     if ( !cpu_topology )
>     {
>         printk(XENLOG_ERR "Failed to allocate memory for cpu_topology 
> table\n");
>         return;
>     }

Ah yes. The odd #define can nevertheless go away, which is the important part.

Jan