Module linux_perf_event_reader::consts
source · [−]Constants
skid constraint Specifies how precise the instruction address should be.
generate AUX records instead of events
include bpf events
use build id in mmap2 events
include cgroup events
include comm data
flag comm events that are due to exec
context switch data
off by default
next exec enables
exclude kernel callchains
exclude user callchains
don’t count in guest
don’t count in host
don’t count hypervisor
don’t count when idle
don’t count kernel
don’t count user
only group on PMU
use freq, not period
children inherit it
per task counts
children only inherit if cloned with CLONE_THREAD
include ksymbol events
include mmap data
include mmap with inode data
non-exec mmap data
include namespaces data
must always be on PMU
event is removed from task on exec
sample_type all events
send synchronous SIGTRAP on event
trace fork/exit
include text poke events
use @clockid for time fields
wakeup_watermark
Write ring buffer from end to beginning
Constants from perf_event.h sizeof first published struct
add: config2
add: branch_sample_type
add: sample_regs_user, sample_stack_user, clockid
add: sample_regs_intr
add: aux_watermark
add: aux_sample_size
add: sig_data
The callchain frames following this context marker frame are “guest” frames.
The callchain frames following this context marker frame are “guest kernel” frames.
The callchain frames following this context marker frame are “guest user” frames.
The callchain frames following this context marker frame are “hypervisor” frames.
The callchain frames following this context marker frame are “kernel” frames.
Any callchain frames which are >= PERF_CONTEXT_MAX are not real addresses; instead, they mark the context of the subsequent callchain frames.
The callchain frames following this context marker frame are “user” frames.
Used in header.misc of the HEADER_BUILD_ID event. If set, the length of the buildid is specified in the event (no more than 20).
Used on PERF_RECORD_COMM event.
Indicates that the content of PERF_SAMPLE_IP points to the actual instruction that triggered the event. See also perf_event_attr::precise_ip. Used on PERF_RECORD_SAMPLE of precise events.
Used on PERF_RECORD_FORK events (perf internal).
Indicates that mmap2 event carries build id data. Used on PERF_RECORD_MMAP2 events.
Used on PERF_RECORD_MMAP events to indicate mappings which are not executable. Not used on PERF_RECORD_MMAP2 events - those have the full protection bitset.
Indicates that /proc/PID/maps parsing are truncated by time out.
Used on PERF_RECORD_SWITCH* events.
Indicates that thread was preempted in TASK_RUNNING state. Used on PERF_RECORD_SWITCH* events.
transaction aborts
any call branch
any return branch
any branch types
direct call
call/ret stack
conditional branches
hypervisor branches
save low level index of raw branch records
indirect calls
indirect jumps
in transaction
kernel branches
no cycles
not in transaction
save branch type
user branches