facts: fix SPARC cpu count on linux (#30261)

On sparc64, /proc/cpuinfo has no usual 'model name', 'Processor', 'vendor_id', 'Vendor',
as a result "ansible_processor_vcpus" is always 1.
Add check element "ncpus active" to fix the issue.

(cherry picked from commit e93ecac0da)

CHANGELOG.md

@@ -39,6 +39,7 @@ Ansible Changes By Release
 * Fix encoding error when there are nonascii values in the path to the ssh binary
 * removed YAML inventory group name validation, broke existing setups and should be global in any case, and configurable
 * performance improvment for inventory, had slown down considerably from 2.3
+* Fix cpu facts on sparc64 (https://github.com/ansible/ansible/pull/30261)
 
 <a id="2.4"></a>
 

lib/ansible/module_utils/facts/hardware/linux.py

@@ -193,6 +193,7 @@ class LinuxHardware(Hardware):
 
             # model name is for Intel arch, Processor (mind the uppercase P)
             # works for some ARM devices, like the Sheevaplug.
+            # 'ncpus active' is SPARC attribute
             if key in ['model name', 'Processor', 'vendor_id', 'cpu', 'Vendor', 'processor']:
                 if 'processor' not in cpu_facts:
                     cpu_facts['processor'] = []
@@ -216,6 +217,8 @@ class LinuxHardware(Hardware):
                 cores[coreid] = int(data[1].strip())
             elif key == '# processors':
                 cpu_facts['processor_cores'] = int(data[1].strip())
+            elif key == 'ncpus active':
+                i = int(data[1].strip())
 
         # Skip for platforms without vendor_id/model_name in cpuinfo (e.g ppc64le)
         if vendor_id_occurrence > 0: