Struct hwloc::Topology
[−]
[src]
pub struct Topology {
// some fields omitted
}Methods
impl Topology
fn new() -> Topology
Creates a new Topology.
If no further customization is needed on init, this method represents the main entry point. A topology is returned which contains the logical representation of the physical hardware.
Examples
use hwloc::Topology; let topology = Topology::new();
Note that the topology implements the Drop trait, so when it goes out of scope no further cleanup is necessary.
fn with_flags(flags: Vec<TopologyFlag>) -> Topology
Creates a new Topology with custom flags.
This method works like new, but allows to provide a vector
of flags which customize the topology discovery process.
Examples
use hwloc::{Topology, TopologyFlag}; let topology = Topology::with_flags(vec![TopologyFlag::IoDevices]);
Note that the topology implements the Drop trait, so when it goes out of scope no further cleanup is necessary.
fn support(&self) -> &TopologySupport
fn flags(&self) -> Vec<TopologyFlag>
Returns the flags currently set for this topology.
Note that the flags are only used during initialization, so this method can just be used for debugging purposes.
Examples
use hwloc::{Topology,TopologyFlag}; let default_topology = Topology::new(); assert_eq!(0, default_topology.flags().len()); let topology_with_flags = Topology::with_flags(vec![TopologyFlag::IoDevices]); assert_eq!(vec![TopologyFlag::IoDevices], topology_with_flags.flags());
fn depth(&self) -> u32
Returns the full depth of the topology.
In practice, the full depth of the topology equals the depth of the ObjectType::PU
plus one.
The full topology depth is useful to know if one needs to manually traverse the complete topology.
Examples
use hwloc::Topology; let topology = Topology::new(); assert!(topology.depth() > 0);
fn depth_for_type(&self, object_type: &ObjectType) -> Result<u32, TypeDepthError>
Returns the depth for the given ObjectType.
Examples
use hwloc::{Topology,ObjectType}; let topology = Topology::new(); let machine_depth = topology.depth_for_type(&ObjectType::Machine).unwrap(); let pu_depth = topology.depth_for_type(&ObjectType::PU).unwrap(); assert!(machine_depth < pu_depth);
Failures
If hwloc can't find the depth for the given ObjectType, this method will
return an error from the TypeDepthError enum. See this one for more info
on each specific error.
Note that for ObjectType::Bridge, ObjectType::PCIDevice and ObjectType::OSDevice,
always an error will be returned which signals their virtual depth.
fn depth_or_below_for_type(&self, object_type: &ObjectType) -> Result<u32, TypeDepthError>
fn depth_or_above_for_type(&self, object_type: &ObjectType) -> Result<u32, TypeDepthError>
fn type_at_depth(&self, depth: u32) -> ObjectType
Returns the corresponding ObjectType for the given depth.
Examples
use hwloc::{Topology,ObjectType}; let topology = Topology::new(); // Load depth for PU to assert against let pu_depth = topology.depth_for_type(&ObjectType::PU).unwrap(); // Retrieve the type for the given depth assert_eq!(ObjectType::PU, topology.type_at_depth(pu_depth));
Panics
This method will panic if the given depth is larger than the full depth minus one. It can't be negative since its an unsigned integer, but be careful with the depth provided in general.
fn size_at_depth(&self, depth: u32) -> u32
Returns the number of objects at the given depth.
Examples
use hwloc::Topology; let topology = Topology::new(); let topo_depth = topology.depth(); assert!(topology.size_at_depth(topo_depth - 1) > 0);
Panics
This method will panic if the given depth is larger than the full depth minus one. It can't be negative since its an unsigned integer, but be careful with the depth provided in general.
fn object_at_root(&self) -> &TopologyObject
Returns the TopologyObject at the root of the topology.
Examples
use hwloc::{Topology,TopologyObject}; let topology = Topology::new(); assert_eq!(topology.type_at_root(), topology.object_at_root().object_type());
fn type_at_root(&self) -> ObjectType
Returns the ObjectType at the root of the topology.
This method is a convenient shorthand for type_at_depth(0).
Examples
use hwloc::{Topology,ObjectType}; let topology = Topology::new(); let root_type = topology.type_at_root(); let depth_type = topology.type_at_depth(0); assert_eq!(root_type, depth_type);
fn objects_with_type(&self, object_type: &ObjectType) -> Result<Vec<&TopologyObject>, TypeDepthError>
Returns all TopologyObjects with the given ObjectType.
fn objects_at_depth(&self, depth: u32) -> Vec<&TopologyObject>
Returns all TopologyObjects at the given depth.
fn set_cpubind(&mut self, set: CpuSet, flags: CpuBindFlags) -> Result<(), CpuBindError>
Binds the current process or thread on CPUs given in the CpuSet.
fn get_cpubind(&self, flags: CpuBindFlags) -> Option<CpuSet>
Get current process or thread binding.
fn set_cpubind_for_process(&mut self, pid: pid_t, set: CpuSet, flags: CpuBindFlags) -> Result<(), CpuBindError>
Binds a process (identified by its pid) on CPUs identified by the given CpuSet.
fn get_cpubind_for_process(&self, pid: pid_t, flags: CpuBindFlags) -> Option<CpuSet>
Get the current physical binding of a process, identified by its pid.
fn set_cpubind_for_thread(&mut self, tid: pthread_t, set: CpuSet, flags: CpuBindFlags) -> Result<(), CpuBindError>
Bind a thread (by its tid) on CPUs given in through the CpuSet.
fn get_cpubind_for_thread(&self, tid: pthread_t, flags: CpuBindFlags) -> Option<CpuSet>
Get the current physical binding of thread tid.
fn get_cpu_location(&self, flags: CpuBindFlags) -> Option<CpuSet>
Get the last physical CPU where the current process or thread ran.
The operating system may move some tasks from one processor to another at any time according to their binding, so this function may return something that is already outdated.
Flags can include either CPUBIND_PROCESS or CPUBIND_THREAD to
specify whether the query should be for the whole process (union of all CPUs
on which all threads are running), or only the current thread. If the
process is single-threaded, flags can be set to zero to let hwloc use
whichever method is available on the underlying OS.
fn get_cpu_location_for_process(&self, pid: pid_t, flags: CpuBindFlags) -> Option<CpuSet>
Get the last physical CPU where a process ran.
The operating system may move some tasks from one processor to another at any time according to their binding, so this function may return something that is already outdated.