1 2Export CPU topology info via sysfs. Items (attributes) are similar 3to /proc/cpuinfo. 4 51) /sys/devices/system/cpu/cpuX/topology/physical_package_id: 6 7 physical package id of cpuX. Typically corresponds to a physical 8 socket number, but the actual value is architecture and platform 9 dependent. 10 112) /sys/devices/system/cpu/cpuX/topology/core_id: 12 13 the CPU core ID of cpuX. Typically it is the hardware platform's 14 identifier (rather than the kernel's). The actual value is 15 architecture and platform dependent. 16 173) /sys/devices/system/cpu/cpuX/topology/book_id: 18 19 the book ID of cpuX. Typically it is the hardware platform's 20 identifier (rather than the kernel's). The actual value is 21 architecture and platform dependent. 22 234) /sys/devices/system/cpu/cpuX/topology/thread_siblings: 24 25 internal kernel map of cpuX's hardware threads within the same 26 core as cpuX 27 285) /sys/devices/system/cpu/cpuX/topology/core_siblings: 29 30 internal kernel map of cpuX's hardware threads within the same 31 physical_package_id. 32 336) /sys/devices/system/cpu/cpuX/topology/book_siblings: 34 35 internal kernel map of cpuX's hardware threads within the same 36 book_id. 37 38To implement it in an architecture-neutral way, a new source file, 39drivers/base/topology.c, is to export the 4 or 6 attributes. The two book 40related sysfs files will only be created if CONFIG_SCHED_BOOK is selected. 41 42For an architecture to support this feature, it must define some of 43these macros in include/asm-XXX/topology.h: 44#define topology_physical_package_id(cpu) 45#define topology_core_id(cpu) 46#define topology_book_id(cpu) 47#define topology_thread_cpumask(cpu) 48#define topology_core_cpumask(cpu) 49#define topology_book_cpumask(cpu) 50 51The type of **_id is int. 52The type of siblings is (const) struct cpumask *. 53 54To be consistent on all architectures, include/linux/topology.h 55provides default definitions for any of the above macros that are 56not defined by include/asm-XXX/topology.h: 571) physical_package_id: -1 582) core_id: 0 593) thread_siblings: just the given CPU 604) core_siblings: just the given CPU 61 62For architectures that don't support books (CONFIG_SCHED_BOOK) there are no 63default definitions for topology_book_id() and topology_book_cpumask(). 64 65Additionally, CPU topology information is provided under 66/sys/devices/system/cpu and includes these files. The internal 67source for the output is in brackets ("[]"). 68 69 kernel_max: the maximum CPU index allowed by the kernel configuration. 70 [NR_CPUS-1] 71 72 offline: CPUs that are not online because they have been 73 HOTPLUGGED off (see cpu-hotplug.txt) or exceed the limit 74 of CPUs allowed by the kernel configuration (kernel_max 75 above). [~cpu_online_mask + cpus >= NR_CPUS] 76 77 online: CPUs that are online and being scheduled [cpu_online_mask] 78 79 possible: CPUs that have been allocated resources and can be 80 brought online if they are present. [cpu_possible_mask] 81 82 present: CPUs that have been identified as being present in the 83 system. [cpu_present_mask] 84 85The format for the above output is compatible with cpulist_parse() 86[see <linux/cpumask.h>]. Some examples follow. 87 88In this example, there are 64 CPUs in the system but cpus 32-63 exceed 89the kernel max which is limited to 0..31 by the NR_CPUS config option 90being 32. Note also that CPUs 2 and 4-31 are not online but could be 91brought online as they are both present and possible. 92 93 kernel_max: 31 94 offline: 2,4-31,32-63 95 online: 0-1,3 96 possible: 0-31 97 present: 0-31 98 99In this example, the NR_CPUS config option is 128, but the kernel was 100started with possible_cpus=144. There are 4 CPUs in the system and cpu2 101was manually taken offline (and is the only CPU that can be brought 102online.) 103 104 kernel_max: 127 105 offline: 2,4-127,128-143 106 online: 0-1,3 107 possible: 0-127 108 present: 0-3 109 110See cpu-hotplug.txt for the possible_cpus=NUM kernel start parameter 111as well as more information on the various cpumasks. 112