|PMC(3)||FreeBSD Library Functions Manual||PMC(3)|
NAMEpmc — library for accessing hardware performance monitoring counters
LIBRARYPerformance Counters Library (libpmc, -lpmc)
SYNOPSIS#include < pmc.h>
DESCRIPTIONThe Performance Counters Library (libpmc, -lpmc) provides a programming interface that allows applications to use hardware performance counters to gather performance data about specific processes or for the system as a whole. The library is implemented using the lower-level facilities offered by the hwpmc(4) driver.
Key ConceptsPerformance monitoring counters (PMCs) are represented by the library using a software abstraction. These “abstract” PMCs can have two scopes:
- System scope. These PMCs measure events in a whole-system manner, i.e., independent of the currently executing thread. System scope PMCs are allocated on specific CPUs and do not migrate between CPUs. Non-privileged process are allowed to allocate system scope PMCs if the hwpmc(4) sysctl tunable: security.bsd.unprivileged_syspmcs is non-zero.
- Process scope. These PMCs only measure hardware events when the processes they are attached to are executing on a CPU. In an SMP system, process scope PMCs migrate between CPUs along with their target processes.
Orthogonal to PMC scope, PMCs may be allocated in one of two operational modes:
- Counting PMCs measure events according to their scope (system or process). The application needs to explicitly read these counters to retrieve their value.
- Sampling PMCs cause the CPU to be periodically interrupted and information about its state of execution to be collected. Sampling PMCs are used to profile specific processes and kernel threads or to profile the system as a whole.
The scope and operational mode for a software PMC are specified at PMC allocation time. An application is allowed to allocate multiple PMCs subject to availability of hardware resources.
The library uses human-readable strings to name the event being measured by hardware. The syntax used for specifying a hardware event along with additional event specific qualifiers (if any) is described in detail in section EVENT SPECIFIERS below.
PMCs are associated with the process that allocated them and will be automatically reclaimed by the system when the process exits. Additionally, process-scope PMCs have to be attached to one or more target processes before they can perform measurements. A process-scope PMC may be attached to those target processes that its owner process would otherwise be permitted to debug. An owner process may attach PMCs to itself allowing it to measure its own behavior. Additionally, on some machine architectures, such self-attached PMCs may be read cheaply using specialized instructions supported by the processor.
Certain kinds of PMCs require that a log file be configured before they may be started. These include:
- System scope sampling PMCs.
- Process scope sampling PMCs.
- Process scope counting PMCs that have been configured to report PMC readings on process context switches or process exits.
Up to one log file may be configured per owner process. Events logged to a log file may be subsequently analyzed using the pmclog(3) family of functions.
Supported CPUsThe CPUs known to the PMC library are named by the enum pmc_cputype enumeration. Supported CPUs include:
- AMD Athlon CPUs.
- AMD Athlon64 CPUs.
- Intel Atom CPUs and other CPUs conforming to version 3 of the Intel performance measurement architecture.
- Intel Core Solo and Core Duo CPUs, and other CPUs conforming to version 1 of the Intel performance measurement architecture.
- Intel Core2 Solo, Core2 Duo and Core2 Extreme CPUs, and other CPUs conforming to version 2 of the Intel performance measurement architecture.
- Intel Pentium CPUs.
- Intel Pentium Pro CPUs.
- Intel Pentium II CPUs.
- Intel Pentium III CPUs.
- Intel Pentium 4 CPUs.
- Intel Pentium M CPUs.
Supported PMCsPMC supported by this library are named by the enum pmc_class enumeration. Supported PMC kinds include:
- Fixed function hardware counters presents in CPUs conforming to the Intel performance measurement architecture version 2 and later.
- Programmable hardware counters present in CPUs conforming to the Intel performance measurement architecture version 1 and later.
- Programmable hardware counters present in AMD Athlon CPUs.
- Programmable hardware counters present in AMD Athlon64 CPUs.
- Programmable hardware counters present in Intel Pentium 4 CPUs.
- Programmable hardware counters present in Intel Pentium CPUs.
- Programmable hardware counters present in Intel Pentium Pro, Pentium II, Pentium III, Celeron, and Pentium M CPUs.
- The timestamp counter on i386 and amd64 architecture CPUs.
- Software events.
PMC CapabilitiesCapabilities of performance monitoring hardware are denoted using the enum pmc_caps enumeration. Supported capabilities include:
- The ability to cascade counters.
- The ability to count negated to asserted transitions of the hardware conditions being probed for.
- The ability to interrupt the CPU.
- The ability to invert the sense of the hardware conditions being measured.
- The ability to perform precise sampling.
- The hardware allows monitored to be further qualified in some system dependent way.
- The ability to read from performance counters.
- The ability to restrict counting of hardware events to when the CPU is running privileged code.
- The ability to ignore simultaneous hardware events below a programmable threshold.
- The ability to restrict counting of hardware events to those when the CPU is running unprivileged code.
- The ability to write to performance counters.
CPU Naming ConventionsCPUs are named using small integers from zero up to, but excluding, the value returned by function pmc_ncpu(). On platforms supporting sparsely numbered CPUs not all the numbers in this range will denote valid CPUs. Operations on non-existent CPUs will return an error.
Functional Grouping of the APIThis section contains a brief overview of the available functionality in the PMC library. Each function listed here is described further in its own manual page.
- pmc_disable(), pmc_enable()
- Administratively disable (enable) specific performance monitoring counter hardware. Counters that are disabled will not be available to applications to use.
- Convenience Functions
- Returns a list of event names supported by a given PMC type.
- Convert a PMC_CAP_* flag to a human-readable string.
- Convert a PMC_CLASS_* constant to a human-readable string.
- Return a human-readable name for a CPU type.
- Return a human-readable string describing a PMC's disposition.
- Convert a numeric event code to a human-readable string.
- Convert a PMC_MODE_* constant to a human-readable name.
- Return a human-readable string describing a PMC's current state.
- Library Initialization
- Initialize the library. This function must be called before any other library function.
- Log File Handling
- Configure a log file for hwpmc(4) to write logged events to.
- Flush all pending log data in hwpmc(4)'s buffers.
- Flush all pending log data and close hwpmc(4)'s side of the stream.
- Append arbitrary user data to the current log file.
- PMC Management
- pmc_allocate(), pmc_release()
- Allocate (free) a PMC.
- pmc_attach(), pmc_detach()
- Attach (detach) a process scope PMC to a target.
- pmc_read(), pmc_write(), pmc_rw()
- Read (write) a value from (to) a PMC.
- pmc_start(), pmc_stop()
- Start (stop) a software PMC.
- Set the reload value for a sampling PMC.
- Retrieve the capabilities for a given PMC.
- Retrieve information about the CPUs and PMC hardware present in the system.
- Retrieve statistics maintained by hwpmc(4).
- Determine the greatest possible CPU number on the system.
- Return the number of hardware PMCs present in a given CPU.
- Return information about the state of a given CPU's PMCs.
- Determine the width of a hardware counter in bits.
- x86 Architecture Specific API
- Returns the processor model specific register number associated with pmc. Applications may then use the x86 RDPMC instruction to directly read the contents of the PMC.
Signal Handling RequirementsApplications using PMCs are required to handle the following signals:
- When the hwpmc(4) module is unloaded using kldunload(8), processes that have PMCs allocated to them will be sent a SIGBUS signal.
hwpmc(4) driver will send a PMC owning process a
SIGIO signal if:
- If any process-mode PMC allocated by it loses all its target processes.
- If the driver encounters an error when writing log data to a configured log file. This error may be retrieved by a subsequent call to pmc_flush_logfile().
Typical Program Flow
- An application would first invoke function pmc_init() to allow the library to initialize itself.
- Signal handling would then be set up.
- Next the application would allocate the PMCs it desires using function pmc_allocate().
- Initial values for PMCs may be set using function pmc_set().
- If a log file is necessary for the PMCs to work, it would be configured using function pmc_configure_logfile().
- Process scope PMCs would then be attached to their target processes using function pmc_attach().
- The PMCs would then be started using function pmc_start().
- Once started, the values of counting PMCs may be read using function pmc_read(). For PMCs that write events to the log file, this logged data would be read and parsed using the pmclog(3) family of functions.
- PMCs are stopped using function pmc_stop(), and process scope PMCs are detached from their targets using function pmc_detach().
- Before the process exits, its may release its PMCs using function pmc_release(). Any configured log file may be closed using function pmc_configure_logfile().
EVENT SPECIFIERSEvent specifiers are strings comprising of an event name, followed by optional parameters modifying the semantics of the hardware event being probed. Event names are PMC architecture dependent, but the PMC library defines machine independent aliases for commonly used events.
Event specifiers spellings are case-insensitive and space characters, periods, underscores and hyphens are considered equivalent to each other. Thus the event specifiers “Example Event”, “example-event”, and “EXAMPLE_EVENT” are equivalent.
PMC Architecture Dependent EventsPMC architecture dependent event specifiers are described in the following manual pages:
|PMC Class||Manual Page|
||pmc.atom(3), pmc.core(3), pmc.core2(3)|
Event Name AliasesEvent name aliases are PMC-independent names for commonly used events. The following aliases are known to this version of the pmc library:
- Measure the number of branches retired.
- Measure the number of retired branches that were mispredicted.
- Measure processor cycles. This event is implemented using the processor's Time Stamp Counter register.
- Measure the number of data cache misses.
- Measure the number of instruction cache misses.
- Measure the number of instructions retired.
- Measure the number of interrupts seen.
- Measure the number of cycles the processor is not in a halted or sleep state.
COMPATIBILITYThe interface between the pmc library and the hwpmc(4) driver is intended to be private to the implementation and may change. In order to ease forward compatibility with future versions of the hwpmc(4) driver, applications are urged to dynamically link with the pmc library.
The pmc API is currently under development.
SEE ALSOpmc.atom(3), pmc.core(3), pmc.core2(3), pmc.iaf(3), pmc.k7(3), pmc.k8(3), pmc.p4(3), pmc.p5(3), pmc.p6(3), pmc.soft(3), pmc.tsc(3), pmclog(3), hwpmc(4), pmccontrol(8), pmcstat(8)
HISTORYThe pmc library first appeared in FreeBSD 6.0.
AUTHORSThe Performance Counters Library (libpmc, -lpmc) library was written by <jkoshy@FreeBSD.org>.
|November 24, 2008||FreeBSD|