freebsd-dev/lib/libpmc/pmc.atomsilvermont.3

537 lines
17 KiB
Groff
Raw Normal View History

.\" Copyright (c) 2014 Hiren Panchasara <hiren@FreeBSD.org>
.\" All rights reserved.
.\"
.\" Redistribution and use in source and binary forms, with or without
.\" modification, are permitted provided that the following conditions
.\" are met:
.\" 1. Redistributions of source code must retain the above copyright
.\" notice, this list of conditions and the following disclaimer.
.\" 2. Redistributions in binary form must reproduce the above copyright
.\" notice, this list of conditions and the following disclaimer in the
.\" documentation and/or other materials provided with the distribution.
.\"
.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
.\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
.\" SUCH DAMAGE.
.\"
.\" $FreeBSD$
.\"
.Dd April 6, 2017
.Dt PMC.ATOMSILVERMONT 3
.Os
.Sh NAME
.Nm pmc.atomsilvermont
.Nd measurement events for
.Tn Intel
.Tn Atom Silvermont
family CPUs
.Sh LIBRARY
.Lb libpmc
.Sh SYNOPSIS
.In pmc.h
.Sh DESCRIPTION
.Tn Intel
.Tn Atom Silvermont
CPUs contain PMCs conforming to version 3 of the
.Tn Intel
performance measurement architecture.
These CPUs contains two classes of PMCs:
.Bl -tag -width "Li PMC_CLASS_IAP"
.It Li PMC_CLASS_IAF
Fixed-function counters that count only one hardware event per counter.
.It Li PMC_CLASS_IAP
Programmable counters that may be configured to count one of a defined
set of hardware events.
.El
.Pp
The number of PMCs available in each class and their widths need to be
determined at run time by calling
.Xr pmc_cpuinfo 3 .
.Pp
Intel Atom Silvermont PMCs are documented in
.Rs
.%B "Intel 64 and IA-32 Intel(R) Architecture Software Developer's Manual"
.%T "Combined Volumes"
.%N "Order Number 325462-050US"
.%D February 2014
.%Q "Intel Corporation"
.Re
.Ss ATOM SILVERMONT FIXED FUNCTION PMCS
These PMCs and their supported events are documented in
.Xr pmc.iaf 3 .
.Ss ATOM SILVERMONT PROGRAMMABLE PMCS
The programmable PMCs support the following capabilities:
.Bl -column "PMC_CAP_INTERRUPT" "Support"
.It Em Capability Ta Em Support
.It PMC_CAP_CASCADE Ta \&No
.It PMC_CAP_EDGE Ta Yes
.It PMC_CAP_INTERRUPT Ta Yes
.It PMC_CAP_INVERT Ta Yes
.It PMC_CAP_READ Ta Yes
.It PMC_CAP_PRECISE Ta \&No
.It PMC_CAP_SYSTEM Ta Yes
.It PMC_CAP_TAGGING Ta \&No
.It PMC_CAP_THRESHOLD Ta Yes
.It PMC_CAP_USER Ta Yes
.It PMC_CAP_WRITE Ta Yes
.El
.Ss Event Qualifiers
Event specifiers for these PMCs support the following common
qualifiers:
.Bl -tag -width indent
.It Li any
Count matching events seen on any logical processor in a package.
.It Li cmask= Ns Ar value
Configure the PMC to increment only if the number of configured
events measured in a cycle is greater than or equal to
.Ar value .
.It Li edge
Configure the PMC to count the number of de-asserted to asserted
transitions of the conditions expressed by the other qualifiers.
If specified, the counter will increment only once whenever a
condition becomes true, irrespective of the number of clocks during
which the condition remains true.
.It Li inv
Invert the sense of comparison when the
.Dq Li cmask
qualifier is present, making the counter increment when the number of
events per cycle is less than the value specified by the
.Dq Li cmask
qualifier.
.It Li os
Configure the PMC to count events happening at processor privilege
level 0.
.It Li usr
Configure the PMC to count events occurring at privilege levels 1, 2
or 3.
.El
.Pp
If neither of the
.Dq Li os
or
.Dq Li usr
qualifiers are specified, the default is to enable both.
.Pp
Events that require core-specificity to be specified use a
additional qualifier
.Dq Li core= Ns Ar core ,
where argument
.Ar core
is one of:
.Bl -tag -width indent
.It Li all
Measure event conditions on all cores.
.It Li this
Measure event conditions on this core.
.El
.Pp
The default is
.Dq Li this .
.Pp
Events that require an agent qualifier to be specified use an
additional qualifier
.Dq Li agent= Ns agent ,
where argument
.Ar agent
is one of:
.Bl -tag -width indent
.It Li this
Measure events associated with this bus agent.
.It Li any
Measure events caused by any bus agent.
.El
.Pp
The default is
.Dq Li this .
.Pp
Events that require a hardware prefetch qualifier to be specified use an
additional qualifier
.Dq Li prefetch= Ns Ar prefetch ,
where argument
.Ar prefetch
is one of:
.Bl -tag -width "exclude"
.It Li both
Include all prefetches.
.It Li only
Only count hardware prefetches.
.It Li exclude
Exclude hardware prefetches.
.El
.Pp
The default is
.Dq Li both .
.Pp
Events that require a cache coherence qualifier to be specified use an
additional qualifier
.Dq Li cachestate= Ns Ar state ,
where argument
.Ar state
contains one or more of the following letters:
.Bl -tag -width indent
.It Li e
Count cache lines in the exclusive state.
.It Li i
Count cache lines in the invalid state.
.It Li m
Count cache lines in the modified state.
.It Li s
Count cache lines in the shared state.
.El
.Pp
The default is
.Dq Li eims .
.Pp
Events that require a snoop response qualifier to be specified use an
additional qualifier
.Dq Li snoopresponse= Ns Ar response ,
where argument
.Ar response
comprises of the following keywords separated by
.Dq +
signs:
.Bl -tag -width indent
.It Li clean
Measure CLEAN responses.
.It Li hit
Measure HIT responses.
.It Li hitm
Measure HITM responses.
.El
.Pp
The default is to measure all the above responses.
.Pp
Events that require a snoop type qualifier use an additional qualifier
.Dq Li snooptype= Ns Ar type ,
where argument
.Ar type
comprises the one of the following keywords:
.Bl -tag -width indent
.It Li cmp2i
Measure CMP2I snoops.
.It Li cmp2s
Measure CMP2S snoops.
.El
.Pp
The default is to measure both snoops.
.Ss Event Specifiers (Programmable PMCs)
Atom Silvermont programmable PMCs support the following events:
.Bl -tag -width indent
.It Li REHABQ.LD_BLOCK_ST_FORWARD
.Pq Event 03H , Umask 01H
The number of retired loads that were
prohibited from receiving forwarded data from the store
because of address mismatch.
.It Li REHABQ.LD_BLOCK_STD_NOTREADY
.Pq Event 03H , Umask 02H
The cases where a forward was technically possible,
but did not occur because the store data was not available
at the right time.
.It Li REHABQ.ST_SPLITS
.Pq Event 03H , Umask 04H
The number of retire stores that experienced.
cache line boundary splits.
.It Li REHABQ.LD_SPLITS
.Pq Event 03H , Umask 08H
The number of retire loads that experienced.
cache line boundary splits.
.It Li REHABQ.LOCK
.Pq Event 03H , Umask 10H
The number of retired memory operations with lock semantics.
These are either implicit locked instructions such as the
XCHG instruction or instructions with an explicit LOCK
prefix (0xF0).
.It Li REHABQ.STA_FULL
.Pq Event 03H , Umask 20H
The number of retired stores that are delayed
because there is not a store address buffer available.
.It Li REHABQ.ANY_LD
.Pq Event 03H , Umask 40H
The number of load uops reissued from Rehabq.
.It Li REHABQ.ANY_ST
.Pq Event 03H , Umask 80H
The number of store uops reissued from Rehabq.
.It Li MEM_UOPS_RETIRED.L1_MISS_LOADS
.Pq Event 04H , Umask 01H
The number of load ops retired that miss in L1
Data cache.
Note that prefetch misses will not be counted.
.It Li MEM_UOPS_RETIRED.L2_HIT_LOADS
.Pq Event 04H , Umask 02H
The number of load micro-ops retired that hit L2.
.It Li MEM_UOPS_RETIRED.L2_MISS_LOADS
.Pq Event 04H , Umask 04H
The number of load micro-ops retired that missed L2.
.It Li MEM_UOPS_RETIRED.DTLB_MISS_LOADS
.Pq Event 04H , Umask 08H
The number of load ops retired that had DTLB miss.
.It Li MEM_UOPS_RETIRED.UTLB_MISS
.Pq Event 04H , Umask 10H
The number of load ops retired that had UTLB miss.
.It Li MEM_UOPS_RETIRED.HITM
.Pq Event 04H , Umask 20H
The number of load ops retired that got data
from the other core or from the other module.
.It Li MEM_UOPS_RETIRED.ALL_LOADS
.Pq Event 04H , Umask 40H
The number of load ops retired.
.It Li MEM_UOP_RETIRED.ALL_STORES
.Pq Event 04H , Umask 80H
The number of store ops retired.
.It Li PAGE_WALKS.D_SIDE_CYCLES
.Pq Event 05H , Umask 01H
Every cycle when a D-side (walks due to a load) page walk is in progress.
Page walk duration divided by number of page walks is the average duration of
page-walks.
Edge trigger bit must be cleared.
Set Edge to count the number of page walks.
.It Li PAGE_WALKS.I_SIDE_CYCLES
.Pq Event 05H , Umask 02H
Every cycle when a I-side (walks due to an instruction fetch) page walk is in
progress.
Page walk duration divided by number of page walks is the average duration of
page-walks.
.It Li PAGE_WALKS.WALKS
.Pq Event 05H , Umask 03H
The number of times a data (D) page walk or an instruction (I) page walk is
completed or started.
Since a page walk implies a TLB miss, the number of TLB misses can be counted
by counting the number of pagewalks.
.It Li LONGEST_LAT_CACHE.MISS
.Pq Event 2EH , Umask 41H
the total number of L2 cache references and the number of L2 cache misses
respectively.
L3 is not supported in Silvermont microarchitecture.
.It Li LONGEST_LAT_CACHE.REFERENCE
.Pq Event 2EH , Umask 4FH
The number of requests originating from the core that
references a cache line in the L2 cache.
L3 is not supported in Silvermont microarchitecture.
.It Li L2_REJECT_XQ.ALL
.Pq Event 30H , Umask 00H
The number of demand and prefetch
transactions that the L2 XQ rejects due to a full or near full
condition which likely indicates back pressure from the IDI link.
The XQ may reject transactions from the L2Q (non-cacheable
requests), BBS (L2 misses) and WOB (L2 write-back victims)
.It Li CORE_REJECT_L2Q.ALL
.Pq Event 31H , Umask 00H
The number of demand and L1 prefetcher
requests rejected by the L2Q due to a full or nearly full condition which
likely indicates back pressure from L2Q.
It also counts requests that would have gone directly to the XQ, but are
rejected due to a full or nearly full condition, indicating back pressure from
the IDI link.
The L2Q may also reject transactions from a core to insure fairness between
cores, or to delay a core's dirty eviction when the address conflicts incoming
external snoops.
(Note that L2 prefetcher requests that are dropped are not counted by this
event).
.It Li CPU_CLK_UNHALTED.CORE_P
.Pq Event 3CH , Umask 00H
The number of core cycles while the core is not in a halt state.
The core enters the halt state when it is running the HLT instruction.
In mobile systems the core frequency may change from time to time.
For this reason this event may have a changing ratio with regards to time.
.It Li CPU_CLK_UNHALTED.REF_P
.Pq Event 3CH , Umask 01H
The number of reference cycles that the core is not in a halt state.
The core enters the halt state when it is running the HLT instruction.
In mobile systems the core frequency may change from time.
This event is not affected by core frequency changes but counts as if the core
is running at the maximum frequency all the time.
.It Li ICACHE.HIT
.Pq Event 80H , Umask 01H
The number of instruction fetches from the instruction cache.
.It Li ICACHE.MISSES
.Pq Event 80H , Umask 02H
The number of instruction fetches that miss the Instruction cache or produce
memory requests.
This includes uncacheable fetches.
An instruction fetch miss is counted only once and not once for every cycle
it is outstanding.
.It Li ICACHE.ACCESSES
.Pq Event 80H , Umask 03H
The number of instruction fetches, including uncacheable fetches.
.It Li NIP_STALL.ICACHE_MISS
.Pq Event B6H , Umask 04H
The number of cycles the NIP stalls because of an icache miss.
This is a cumulative count of cycles the NIP stalled for all
icache misses.
.It Li OFFCORE_RESPONSE_0
.Pq Event B7H , Umask 01H
Requires MSR_OFFCORE_RESP0 to specify request type and response.
.It Li OFFCORE_RESPONSE_1
.Pq Event B7H , Umask 02H
Requires MSR_OFFCORE_RESP to specify request type and response.
.It Li INST_RETIRED.ANY_P
.Pq Event C0H , Umask 00H
The number of instructions that retire execution.
For instructions that consist of multiple micro-ops, this event counts the
retirement of the last micro-op of the instruction.
The counter continues counting during hardware interrupts, traps, and inside
interrupt handlers.
.It Li UOPS_RETIRED.MS
.Pq Event C2H , Umask 01H
The number of micro-ops retired that were supplied from MSROM.
.It Li UOPS_RETIRED.ALL
.Pq Event C2H , Umask 10H
The number of micro-ops retired.
.It Li MACHINE_CLEARS.SMC
.Pq Event C3H , Umask 01H
The number of times that a program writes to a code section.
Self-modifying code causes a severe penalty in all Intel
architecture processors.
.It Li MACHINE_CLEARS.MEMORY_ORDERING
.Pq Event C3H , Umask 02H
The number of times that pipeline was cleared due to memory
ordering issues.
.It Li MACHINE_CLEARS.FP_ASSIST
.Pq Event C3H , Umask 04H
The number of times that pipeline stalled due to FP operations
needing assists.
.It Li MACHINE_CLEARS.ALL
.Pq Event C3H , Umask 08H
The number of times that pipeline stalled due to due to any causes
(including SMC, MO, FP assist, etc).
.It Li BR_INST_RETIRED.ALL_BRANCHES
.Pq Event C4H , Umask 00H
The number of branch instructions retired.
.It Li BR_INST_RETIRED.JCC
.Pq Event C4H , Umask 7EH
The number of branch instructions retired that were conditional
jumps.
.It Li BR_INST_RETIRED.FAR_BRANCH
.Pq Event C4H , Umask BFH
The number of far branch instructions retired.
.It Li BR_INST_RETIRED.NON_RETURN_IND
.Pq Event C4H , Umask EBH
The number of branch instructions retired that were near indirect
call or near indirect jmp.
.It Li BR_INST_RETIRED.RETURN
.Pq Event C4H , Umask F7H
The number of near RET branch instructions retired.
.It Li BR_INST_RETIRED.CALL
.Pq Event C4H , Umask F9H
The number of near CALL branch instructions retired.
.It Li BR_INST_RETIRED.IND_CALL
.Pq Event C4H , Umask FBH
The number of near indirect CALL branch instructions retired.
.It Li BR_INST_RETIRED.REL_CALL
.Pq Event C4H , Umask FDH
The number of near relative CALL branch instructions retired.
.It Li BR_INST_RETIRED.TAKEN_JCC
.Pq Event C4H , Umask FEH
The number of branch instructions retired that were conditional
jumps and predicted taken.
.It Li BR_MISP_RETIRED.ALL_BRANCHES
.Pq Event C5H , Umask 00H
The number of mispredicted branch instructions retired.
.It Li BR_MISP_RETIRED.JCC
.Pq Event C5H , Umask 7EH
The number of mispredicted branch instructions retired that were
conditional jumps.
.It Li BR_MISP_RETIRED.FAR
.Pq Event C5H , Umask BFH
The number of mispredicted far branch instructions retired.
.It Li BR_MISP_RETIRED.NON_RETURN_IND
.Pq Event C5H , Umask EBH
The number of mispredicted branch instructions retired that were
near indirect call or near indirect jmp.
.It Li BR_MISP_RETIRED.RETURN
.Pq Event C5H , Umask F7H
The number of mispredicted near RET branch instructions retired.
.It Li BR_MISP_RETIRED.CALL
.Pq Event C5H , Umask F9H
The number of mispredicted near CALL branch instructions retired.
.It Li BR_MISP_RETIRED.IND_CALL
.Pq Event C5H , Umask FBH
The number of mispredicted near indirect CALL branch instructions
retired.
.It Li BR_MISP_RETIRED.REL_CALL
.Pq Event C5H , Umask FDH
The number of mispredicted near relative CALL branch instructions
retired.
.It Li BR_MISP_RETIRED.TAKEN_JCC
.Pq Event C5H , Umask FEH
The number of mispredicted branch instructions retired that were
conditional jumps and predicted taken.
.It Li NO_ALLOC_CYCLES.ROB_FULL
.Pq Event CAH , Umask 01H
The number of cycles when no uops are allocated and the ROB is full
(less than 2 entries available).
.It Li NO_ALLOC_CYCLES.RAT_STALL
.Pq Event CAH , Umask 20H
The number of cycles when no uops are allocated and a RATstall is
asserted.
.It Li NO_ALLOC_CYCLES.ALL
.Pq Event CAH , Umask 3FH
The number of cycles when the front-end does not provide any
instructions to be allocated for any reason.
.It Li NO_ALLOC_CYCLES.NOT_DELIVERED
.Pq Event CAH , Umask 50H
The number of cycles when the front-end does not provide any
instructions to be allocated but the back end is not stalled.
.It Li RS_FULL_STALL.MEC
.Pq Event CBH , Umask 01H
The number of cycles the allocation pipe line stalled due to
the RS for the MEC cluster is full.
.It Li RS_FULL_STALL.ALL
.Pq Event CBH , Umask 1FH
The number of cycles that the allocation pipe line stalled due
to any one of the RS is full.
.It Li CYCLES_DIV_BUSY.ANY
.Pq Event CDH , Umask 01H
The number of cycles the divider is busy.
.It Li BACLEARS.ALL
.Pq Event E6H , Umask 01H
The number of baclears for any type of branch.
.It Li BACLEARS.RETURN
.Pq Event E6H , Umask 08H
The number of baclears for return branches.
.It Li BACLEARS.COND
.Pq Event E6H , Umask 10H
The number of baclears for conditional branches.
.It Li MS_DECODED.MS_ENTRY
.Pq Event E7H , Umask 01H)
The number of times the MSROM starts a flow of UOPS.
.El
.Sh SEE ALSO
.Xr pmc 3 ,
.Xr pmc.atom 3 ,
.Xr pmc.core 3 ,
.Xr pmc.core2 3 ,
.Xr pmc.iaf 3 ,
.Xr pmc.k7 3 ,
.Xr pmc.k8 3 ,
.Xr pmc.p4 3 ,
.Xr pmc.p5 3 ,
.Xr pmc.p6 3 ,
.Xr pmc.soft 3 ,
.Xr pmc.tsc 3 ,
.Xr pmc_cpuinfo 3 ,
.Xr pmclog 3 ,
.Xr hwpmc 4
.Sh HISTORY
The
.Nm pmc
library first appeared in
.Fx 6.0 .
.Sh AUTHORS
.An -nosplit
The
.Lb libpmc
library was written by
.An Joseph Koshy Aq Mt jkoshy@FreeBSD.org .
The support for the Atom Silvermont
microarchitecture was written by
.An Hiren Panchasara Aq Mt hiren@FreeBSD.org .