From 8c895a440ff67d4248db4d667c12c9fa4f835942 Mon Sep 17 00:00:00 2001 From: rstone Date: Mon, 25 Nov 2013 18:31:13 +0000 Subject: pmc(3) does not document a dependency on but requires it, so include it explicitly from . MFC after: 1 month --- lib/libpmc/pmc.h | 1 + 1 file changed, 1 insertion(+) (limited to 'lib/libpmc') diff --git a/lib/libpmc/pmc.h b/lib/libpmc/pmc.h index e17a584..81672d2 100644 --- a/lib/libpmc/pmc.h +++ b/lib/libpmc/pmc.h @@ -30,6 +30,7 @@ #define _PMC_H_ #include +#include #include /* -- cgit v1.1 From 859fb3bb22aee35c1a8558c3f1cdb245976f80e5 Mon Sep 17 00:00:00 2001 From: jhibbits Date: Sat, 1 Feb 2014 02:03:50 +0000 Subject: Add hwpmc(4) support for the PowerPC 970 class processors, direct events. This also fixes asserts on removal of the module for the mpc74xx. The PowerPC 970 processors have two different types of events: direct events and indirect events. Thus far only direct events are supported. I included some documentation in the driver on how indirect events work, but support is for the future. MFC after: 1 month --- lib/libpmc/libpmc.c | 42 ++++++++++++++++++++++++++++++++---------- 1 file changed, 32 insertions(+), 10 deletions(-) (limited to 'lib/libpmc') diff --git a/lib/libpmc/libpmc.c b/lib/libpmc/libpmc.c index 74956a9..209ae56 100644 --- a/lib/libpmc/libpmc.c +++ b/lib/libpmc/libpmc.c @@ -28,6 +28,7 @@ __FBSDID("$FreeBSD$"); #include +#include #include #include #include @@ -85,7 +86,7 @@ static int soft_allocate_pmc(enum pmc_event _pe, char *_ctrspec, struct pmc_op_pmcallocate *_pmc_config); #if defined(__powerpc__) -static int ppc7450_allocate_pmc(enum pmc_event _pe, char* ctrspec, +static int powerpc_allocate_pmc(enum pmc_event _pe, char* ctrspec, struct pmc_op_pmcallocate *_pmc_config); #endif /* __powerpc__ */ @@ -156,6 +157,7 @@ PMC_CLASSDEP_TABLE(mips24k, MIPS24K); PMC_CLASSDEP_TABLE(octeon, OCTEON); PMC_CLASSDEP_TABLE(ucf, UCF); PMC_CLASSDEP_TABLE(ppc7450, PPC7450); +PMC_CLASSDEP_TABLE(ppc970, PPC970); static struct pmc_event_descr soft_event_table[PMC_EV_DYN_COUNT]; @@ -262,6 +264,7 @@ PMC_MDEP_TABLE(xscale, XSCALE, PMC_CLASS_SOFT, PMC_CLASS_XSCALE); PMC_MDEP_TABLE(mips24k, MIPS24K, PMC_CLASS_SOFT, PMC_CLASS_MIPS24K); PMC_MDEP_TABLE(octeon, OCTEON, PMC_CLASS_SOFT, PMC_CLASS_OCTEON); PMC_MDEP_TABLE(ppc7450, PPC7450, PMC_CLASS_SOFT, PMC_CLASS_PPC7450); +PMC_MDEP_TABLE(ppc970, PPC970, PMC_CLASS_SOFT, PMC_CLASS_PPC970); PMC_MDEP_TABLE(generic, SOFT, PMC_CLASS_SOFT); static const struct pmc_event_descr tsc_event_table[] = @@ -322,7 +325,8 @@ PMC_CLASS_TABLE_DESC(mips24k, MIPS24K, mips24k, mips); PMC_CLASS_TABLE_DESC(octeon, OCTEON, octeon, mips); #endif /* __mips__ */ #if defined(__powerpc__) -PMC_CLASS_TABLE_DESC(ppc7450, PPC7450, ppc7450, ppc7450); +PMC_CLASS_TABLE_DESC(ppc7450, PPC7450, ppc7450, powerpc); +PMC_CLASS_TABLE_DESC(ppc970, PPC970, ppc970, powerpc); #endif static struct pmc_class_descr soft_class_table_descr = @@ -2404,13 +2408,19 @@ static struct pmc_event_alias ppc7450_aliases[] = { EV_ALIAS(NULL, NULL) }; -#define PPC7450_KW_OS "os" -#define PPC7450_KW_USR "usr" -#define PPC7450_KW_ANYTHREAD "anythread" +static struct pmc_event_alias ppc970_aliases[] = { + EV_ALIAS("instructions", "INSTR_COMPLETED"), + EV_ALIAS("cycles", "CYCLES"), + EV_ALIAS(NULL, NULL) +}; + +#define POWERPC_KW_OS "os" +#define POWERPC_KW_USR "usr" +#define POWERPC_KW_ANYTHREAD "anythread" static int -ppc7450_allocate_pmc(enum pmc_event pe, char *ctrspec __unused, - struct pmc_op_pmcallocate *pmc_config __unused) +powerpc_allocate_pmc(enum pmc_event pe, char *ctrspec __unused, + struct pmc_op_pmcallocate *pmc_config __unused) { char *p; @@ -2419,11 +2429,11 @@ ppc7450_allocate_pmc(enum pmc_event pe, char *ctrspec __unused, pmc_config->pm_caps |= (PMC_CAP_READ | PMC_CAP_WRITE); while ((p = strsep(&ctrspec, ",")) != NULL) { - if (KWMATCH(p, PPC7450_KW_OS)) + if (KWMATCH(p, POWERPC_KW_OS)) pmc_config->pm_caps |= PMC_CAP_SYSTEM; - else if (KWMATCH(p, PPC7450_KW_USR)) + else if (KWMATCH(p, POWERPC_KW_USR)) pmc_config->pm_caps |= PMC_CAP_USER; - else if (KWMATCH(p, PPC7450_KW_ANYTHREAD)) + else if (KWMATCH(p, POWERPC_KW_ANYTHREAD)) pmc_config->pm_caps |= (PMC_CAP_USER | PMC_CAP_SYSTEM); else return (-1); @@ -2431,6 +2441,7 @@ ppc7450_allocate_pmc(enum pmc_event pe, char *ctrspec __unused, return (0); } + #endif /* __powerpc__ */ @@ -2830,6 +2841,10 @@ pmc_event_names_of_class(enum pmc_class cl, const char ***eventnames, ev = ppc7450_event_table; count = PMC_EVENT_TABLE_SIZE(ppc7450); break; + case PMC_CLASS_PPC970: + ev = ppc970_event_table; + count = PMC_EVENT_TABLE_SIZE(ppc970); + break; case PMC_CLASS_SOFT: ev = soft_event_table; count = soft_event_info.pm_nevent; @@ -3100,6 +3115,10 @@ pmc_init(void) PMC_MDEP_INIT(ppc7450); pmc_class_table[n] = &ppc7450_class_table_descr; break; + case PMC_CPU_PPC_970: + PMC_MDEP_INIT(ppc970); + pmc_class_table[n] = &ppc970_class_table_descr; + break; #endif default: /* @@ -3270,6 +3289,9 @@ _pmc_name_of_event(enum pmc_event pe, enum pmc_cputype cpu) } else if (pe >= PMC_EV_PPC7450_FIRST && pe <= PMC_EV_PPC7450_LAST) { ev = ppc7450_event_table; evfence = ppc7450_event_table + PMC_EVENT_TABLE_SIZE(ppc7450); + } else if (pe >= PMC_EV_PPC970_FIRST && pe <= PMC_EV_PPC970_LAST) { + ev = ppc970_event_table; + evfence = ppc970_event_table + PMC_EVENT_TABLE_SIZE(ppc970); } else if (pe == PMC_EV_TSC_TSC) { ev = tsc_event_table; evfence = tsc_event_table + PMC_EVENT_TABLE_SIZE(tsc); -- cgit v1.1 From 2babaa36d7395945084bafdde10218682e306b93 Mon Sep 17 00:00:00 2001 From: hiren Date: Thu, 20 Mar 2014 20:51:08 +0000 Subject: Update hwpmc to support core events for Atom Silvermont microarchitecture. (Model 0x4D as per Intel document 330061-001 01/2014) Tested by: Olivier Cochard-Labbe MFC after: 4 weeks --- lib/libpmc/Makefile | 1 + lib/libpmc/libpmc.c | 22 ++ lib/libpmc/pmc.atom.3 | 3 +- lib/libpmc/pmc.atomsilvermont.3 | 535 ++++++++++++++++++++++++++++++++++++++++ 4 files changed, 560 insertions(+), 1 deletion(-) create mode 100644 lib/libpmc/pmc.atomsilvermont.3 (limited to 'lib/libpmc') diff --git a/lib/libpmc/Makefile b/lib/libpmc/Makefile index d9ac0b9..4227794 100644 --- a/lib/libpmc/Makefile +++ b/lib/libpmc/Makefile @@ -24,6 +24,7 @@ MAN+= pmc.soft.3 # PMC-dependent manual pages MAN+= pmc.atom.3 +MAN+= pmc.atomsilvermont.3 MAN+= pmc.core.3 MAN+= pmc.core2.3 MAN+= pmc.corei7.3 diff --git a/lib/libpmc/libpmc.c b/lib/libpmc/libpmc.c index 209ae56..2b7b61b2 100644 --- a/lib/libpmc/libpmc.c +++ b/lib/libpmc/libpmc.c @@ -169,6 +169,11 @@ static const struct pmc_event_descr atom_event_table[] = __PMC_EV_ALIAS_ATOM() }; +static const struct pmc_event_descr atom_silvermont_event_table[] = +{ + __PMC_EV_ALIAS_ATOM_SILVERMONT() +}; + static const struct pmc_event_descr core_event_table[] = { __PMC_EV_ALIAS_CORE() @@ -246,6 +251,7 @@ static const struct pmc_event_descr westmereuc_event_table[] = } PMC_MDEP_TABLE(atom, IAP, PMC_CLASS_SOFT, PMC_CLASS_IAF, PMC_CLASS_TSC); +PMC_MDEP_TABLE(atom_silvermont, IAP, PMC_CLASS_SOFT, PMC_CLASS_IAF, PMC_CLASS_TSC); PMC_MDEP_TABLE(core, IAP, PMC_CLASS_SOFT, PMC_CLASS_TSC); PMC_MDEP_TABLE(core2, IAP, PMC_CLASS_SOFT, PMC_CLASS_IAF, PMC_CLASS_TSC); PMC_MDEP_TABLE(corei7, IAP, PMC_CLASS_SOFT, PMC_CLASS_IAF, PMC_CLASS_TSC, PMC_CLASS_UCF, PMC_CLASS_UCP); @@ -288,6 +294,7 @@ static const struct pmc_class_descr NAME##_class_table_descr = \ #if defined(__i386__) || defined(__amd64__) PMC_CLASS_TABLE_DESC(iaf, IAF, iaf, iaf); PMC_CLASS_TABLE_DESC(atom, IAP, atom, iap); +PMC_CLASS_TABLE_DESC(atom_silvermont, IAP, atom_silvermont, iap); PMC_CLASS_TABLE_DESC(core, IAP, core, iap); PMC_CLASS_TABLE_DESC(core2, IAP, core2, iap); PMC_CLASS_TABLE_DESC(corei7, IAP, corei7, iap); @@ -597,6 +604,8 @@ static struct pmc_event_alias core2_aliases_without_iaf[] = { #define atom_aliases core2_aliases #define atom_aliases_without_iaf core2_aliases_without_iaf +#define atom_silvermont_aliases core2_aliases +#define atom_silvermont_aliases_without_iaf core2_aliases_without_iaf #define corei7_aliases core2_aliases #define corei7_aliases_without_iaf core2_aliases_without_iaf #define haswell_aliases core2_aliases @@ -842,6 +851,7 @@ iap_allocate_pmc(enum pmc_event pe, char *ctrspec, if (n != 1) return (-1); } else if (cpu_info.pm_cputype == PMC_CPU_INTEL_ATOM || + cpu_info.pm_cputype == PMC_CPU_INTEL_ATOM_SILVERMONT || cpu_info.pm_cputype == PMC_CPU_INTEL_CORE2 || cpu_info.pm_cputype == PMC_CPU_INTEL_CORE2EXTREME) { if (KWPREFIXMATCH(p, IAP_KW_SNOOPRESPONSE "=")) { @@ -2733,6 +2743,10 @@ pmc_event_names_of_class(enum pmc_class cl, const char ***eventnames, ev = atom_event_table; count = PMC_EVENT_TABLE_SIZE(atom); break; + case PMC_CPU_INTEL_ATOM_SILVERMONT: + ev = atom_silvermont_event_table; + count = PMC_EVENT_TABLE_SIZE(atom_silvermont); + break; case PMC_CPU_INTEL_CORE: ev = core_event_table; count = PMC_EVENT_TABLE_SIZE(core); @@ -3049,6 +3063,9 @@ pmc_init(void) case PMC_CPU_INTEL_ATOM: PMC_MDEP_INIT_INTEL_V2(atom); break; + case PMC_CPU_INTEL_ATOM_SILVERMONT: + PMC_MDEP_INIT_INTEL_V2(atom_silvermont); + break; case PMC_CPU_INTEL_CORE: PMC_MDEP_INIT(core); pmc_class_table[n] = &core_class_table_descr; @@ -3202,6 +3219,11 @@ _pmc_name_of_event(enum pmc_event pe, enum pmc_cputype cpu) ev = atom_event_table; evfence = atom_event_table + PMC_EVENT_TABLE_SIZE(atom); break; + case PMC_CPU_INTEL_ATOM_SILVERMONT: + ev = atom_silvermont_event_table; + evfence = atom_silvermont_event_table + + PMC_EVENT_TABLE_SIZE(atom_silvermont); + break; case PMC_CPU_INTEL_CORE: ev = core_event_table; evfence = core_event_table + PMC_EVENT_TABLE_SIZE(core); diff --git a/lib/libpmc/pmc.atom.3 b/lib/libpmc/pmc.atom.3 index f61a141..9b9a58f 100644 --- a/lib/libpmc/pmc.atom.3 +++ b/lib/libpmc/pmc.atom.3 @@ -23,7 +23,7 @@ .\" .\" $FreeBSD$ .\" -.Dd November 12, 2008 +.Dd March 20, 2014 .Dt PMC.ATOM 3 .Os .Sh NAME @@ -1168,6 +1168,7 @@ and the underlying hardware events used on these CPUs. .El .Sh SEE ALSO .Xr pmc 3 , +.Xr pmc.atomsilvermont 3 , .Xr pmc.core 3 , .Xr pmc.core2 3 , .Xr pmc.iaf 3 , diff --git a/lib/libpmc/pmc.atomsilvermont.3 b/lib/libpmc/pmc.atomsilvermont.3 new file mode 100644 index 0000000..91e22e9 --- /dev/null +++ b/lib/libpmc/pmc.atomsilvermont.3 @@ -0,0 +1,535 @@ +.\" Copyright (c) 2014 Hiren Panchasara +.\" 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 March 20, 2014 +.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 +The +.Lb libpmc +library was written by +.An "Joseph Koshy" +.Aq jkoshy@FreeBSD.org . +The support for the Atom Silvermont +microarchitecture was written by +.An "Hiren Panchasara" +.Aq hiren@FreeBSD.org . -- cgit v1.1