[PATCH] Add support for ARM Cortex A17 microarchitecture

[Hyeongsik Min]

Added the event list of the ARM Cortex A17 based on
ARM DDI 0535C.

Signed-off-by: Hyeongsik Min <***@gmail.com>
---
events/Makefile.am | 1 +
events/arm/armv7-ca17/events | 77 ++++++++++++++++++++++++++++++++++++++++
events/arm/armv7-ca17/unit_masks | 4 +++
libop/op_cpu_type.c | 3 ++
libop/op_cpu_type.h | 1 +
libop/op_events.c | 1 +
utils/ophelp.c | 6 ++++
7 files changed, 93 insertions(+)
create mode 100644 events/arm/armv7-ca17/events
create mode 100644 events/arm/armv7-ca17/unit_masks

diff --git a/events/Makefile.am b/events/Makefile.am
index 29d4b5f..db43550 100644
--- a/events/Makefile.am
+++ b/events/Makefile.am
@@ -50,6 +50,7 @@ event_files = \
arm/armv7-ca5/events arm/armv7-ca5/unit_masks \
arm/armv7-ca7/events arm/armv7-ca7/unit_masks \
arm/armv7-ca15/events arm/armv7-ca15/unit_masks \
+ arm/armv7-ca17/events arm/armv7-ca17/unit_masks \
arm/mpcore/events arm/mpcore/unit_masks \
arm/armv8-pmuv3-common/events arm/armv8-pmuv3-common/unit_masks \
arm/armv8-xgene/events arm/armv8-xgene/unit_masks \
diff --git a/events/arm/armv7-ca17/events b/events/arm/armv7-ca17/events
new file mode 100644
index 0000000..259852c
--- /dev/null
+++ b/events/arm/armv7-ca17/events
@@ -0,0 +1,77 @@
+# ARM Cortex A17 events
+# From Cortex A17 TRM
+#
+include:arm/armv7-common
+
+event:0x40 counters:1,2,3,4,5,6 um:zero minimum:500 name:L1D_CACHE_LD : Level 1 data cache access, read
+event:0x41 counters:1,2,3,4,5,6 um:zero minimum:500 name:L1D_CACHE_ST : Level 1 data cache access, write
+
+event:0x50 counters:1,2,3,4,5,6 um:zero minimum:500 name:L2D_CACHE_LD : Level 2 data cache access, read
+event:0x51 counters:1,2,3,4,5,6 um:zero minimum:500 name:L2D_CACHE_ST : Level 2 data cache access, write
+
+event:0x56 counters:1,2,3,4,5,6 um:zero minimum:500 name:L2D_CACHE_WB_VICTIM : Level 2 data cache write-back, victim
+event:0x57 counters:1,2,3,4,5,6 um:zero minimum:500 name:L2D_CACHE_WB_CLEAN : Level 2 data cache write-back, cleaning and coherency
+event:0x58 counters:1,2,3,4,5,6 um:zero minimum:500 name:L2D_CACHE_INVAL : Level 2 data cache invalidate
+
+event:0x62 counters:1,2,3,4,5,6 um:zero minimum:500 name:BUS_ACCESS_SHARED : Bus access, normal, cacheable, shareable
+event:0x63 counters:1,2,3,4,5,6 um:zero minimum:500 name:BUS_ACCESS_NOT_SHARED : Bus access, not normal, cacheable, shareable
+event:0x64 counters:1,2,3,4,5,6 um:zero minimum:500 name:BUS_ACCESS_NORMAL : Bus access, normal
+event:0x65 counters:1,2,3,4,5,6 um:zero minimum:500 name:BUS_ACCESS_PERIPH : Bus access, peripheral
+event:0x66 counters:1,2,3,4,5,6 um:zero minimum:500 name:MEM_ACCESS_LD : Data memory access, read
+event:0x67 counters:1,2,3,4,5,6 um:zero minimum:500 name:MEM_ACCESS_ST : Data memory access, write
+event:0x68 counters:1,2,3,4,5,6 um:zero minimum:500 name:UNALIGNED_LD_SPEC : Unaligned access, read
+event:0x69 counters:1,2,3,4,5,6 um:zero minimum:500 name:UNALIGNED_ST_SPEC : Unaligned access, write
+event:0x6A counters:1,2,3,4,5,6 um:zero minimum:500 name:UNALIGNED_LDST_SPEC : Unaligned access
+
+event:0x6C counters:1,2,3,4,5,6 um:zero minimum:500 name:LDREX_SPEC : ldrex instruction speculatively executed
+event:0x6E counters:1,2,3,4,5,6 um:zero minimum:500 name:STREX_FAIL_SPEC : strex instruction speculatively executed, fail
+event:0x6F counters:1,2,3,4,5,6 um:zero minimum:500 name:STREX_SPEC : strex instruction speculatively executed
+
+event:0x70 counters:1,2,3,4,5,6 um:zero minimum:500 name:LD_SPEC : Load instruction speculatively executed
+event:0x71 counters:1,2,3,4,5,6 um:zero minimum:500 name:ST_SPEC : Store instruction speculatively executed
+event:0x72 counters:1,2,3,4,5,6 um:zero minimum:500 name:LDST_SPEC : Load or store instruction speculatively executed
+event:0x73 counters:1,2,3,4,5,6 um:zero minimum:500 name:DP_SPEC : Data processing instruction speculatively executed
+event:0x74 counters:1,2,3,4,5,6 um:zero minimum:500 name:ASE_SPEC : Advanced SIMD extension instruction speculatively executed
+event:0x75 counters:1,2,3,4,5,6 um:zero minimum:500 name:VFP_SPEC : Floating-point extension instruction speculatively executed
+event:0x76 counters:1,2,3,4,5,6 um:zero minimum:500 name:PC_WRITE_SPEC : Software change of the PC instruction speculatively executed
+
+event:0x78 counters:1,2,3,4,5,6 um:zero minimum:500 name:BR_IMMED_SPEC : Immediate branch instruction speculatively executed
+event:0x79 counters:1,2,3,4,5,6 um:zero minimum:500 name:BR_RETURN_SPEC : Procedure return instruction speculatively executed
+event:0x7A counters:1,2,3,4,5,6 um:zero minimum:500 name:BR_INDIRECT_SPEC : Indirect branch instruction speculatively executed
+
+event:0x7C counters:1,2,3,4,5,6 um:zero minimum:500 name:ISB_SPEC : ISB barrier instruction speculatively executed
+event:0x7D counters:1,2,3,4,5,6 um:zero minimum:500 name:DSB_SPEC : DSB barrier instruction speculatively executed
+event:0x7E counters:1,2,3,4,5,6 um:zero minimum:500 name:DMB_SPEC : DMB barrier instruction speculatively executed
+
+event:0x81 counters:1,2,3,4,5,6 um:zero minimum:500 name:EXC_UNDEF : Exception taken, Undefined Instruction
+event:0x8A counters:1,2,3,4,5,6 um:zero minimum:500 name:EXC_HVC : Exception taken, Hypervisor Call
+
+event:0xC0 counters:1,2,3,4,5,6 um:zero minimum:500 name:LF_STALL : Linefill caused an instruction side stall
+event:0xC1 counters:1,2,3,4,5,6 um:zero minimum:500 name:PTW_STALL : Translation table walk caused an instruction side stall
+event:0xC2 counters:1,2,3,4,5,6 um:zero minimum:500 name:I_TAG_RAM_RD : Number of set of 4 ways read in the instruction cache - Tag RAM
+event:0xC3 counters:1,2,3,4,5,6 um:zero minimum:500 name:I_DATA_RAM_RD : Number of ways read in the instruction cache - Data RAM
+event:0xC4 counters:1,2,3,4,5,6 um:zero minimum:500 name:I_BTAC_RAM_RD : Number of ways read in the instruction BTAC RAM
+
+event:0xCA counters:1,2,3,4,5,6 um:zero minimum:500 name:DATA_SNOOP : Data snooped from other processor
+
+event:0xD3 counters:1,2,3,4,5,6 um:zero minimum:500 name:D_LSU_SLOT_FULL : Duration during which all slots in the Load-Store Unit are busy
+event:0xD8 counters:1,2,3,4,5,6 um:zero minimum:500 name:LS_IQ_FULL : Duration during which all slots in the Load-Store Issue queue are busy
+event:0xD9 counters:1,2,3,4,5,6 um:zero minimum:500 name:DP_IQ_FULL : Duration during which all slots in the Data Processing issue queue are busy.
+event:0xDA counters:1,2,3,4,5,6 um:zero minimum:500 name:DE_IQ_FULL : Duration during which all slots in the Data Engine issue queue are busy
+event:0xDB counters:1,2,3,4,5,6 um:zero minimum:500 name:FLUSH_DE_NEON : Number of NEON instruction which fail their condition code and lead to a flush of the DE pipe
+event:0xDC counters:1,2,3,4,5,6 um:zero minimum:500 name:EXC_TRAP_HYP : Number of Trap to hypervisor
+event:0xDE counters:1,2,3,4,5,6 um:zero minimum:500 name:ETM_EXT_OUT0 : PTM EXTOUT 0
+event:0xDF counters:1,2,3,4,5,6 um:zero minimum:500 name:ETM_EXT_OUT1 : PTM EXTOUT 1
+
+event:0xE0 counters:1,2,3,4,5,6 um:zero minimum:500 name:MMU_PTW : Duration during which the MMU handle a translation table walk.
+event:0xE1 counters:1,2,3,4,5,6 um:zero minimum:500 name:MMU_PTW_ST1 : Duration during which the MMU handle a Stage1 translation table walk
+event:0xE2 counters:1,2,3,4,5,6 um:zero minimum:500 name:MMU_PTW_ST2 : Duration during which the MMU handle a Stage2 translation table walk
+event:0xE3 counters:1,2,3,4,5,6 um:zero minimum:500 name:MMU_PTW_LSU : Duration during which the MMU handle a translation table walk requested by the Load Store Unit
+event:0xE4 counters:1,2,3,4,5,6 um:zero minimum:500 name:MMU_PTW_ISIDE : Duration during which the MMU handle a translation table walk requested by the Instruction side
+event:0xE5 counters:1,2,3,4,5,6 um:zero minimum:500 name:MMU_PTW_PLD : Duration during which the MMU handle a translation table walk requested by a Preload instruction or Prefetch request
+event:0xE6 counters:1,2,3,4,5,6 um:zero minimum:500 name:MMU_PTW_CP15 : Duration during which the MMU handle a translation table walk requested by a CP15 operation
+event:0xE7 counters:1,2,3,4,5,6 um:zero minimum:500 name:PLD_UTLB_REFILL : Level 1 PLD TLB refill
+event:0xE8 counters:1,2,3,4,5,6 um:zero minimum:500 name:CP15_UTLB_REFILL : Level 1 CP15 TLB refil
+event:0xE9 counters:1,2,3,4,5,6 um:zero minimum:500 name:UTLB_FLUSH : Level 1 TLB flus
+event:0xEA counters:1,2,3,4,5,6 um:zero minimum:500 name:TLB_ACCESS : Level 2 TLB access
+event:0xEB counters:1,2,3,4,5,6 um:zero minimum:500 name:TLB_MISS : Level 2 TLB miss
diff --git a/events/arm/armv7-ca17/unit_masks b/events/arm/armv7-ca17/unit_masks
new file mode 100644
index 0000000..4027469
--- /dev/null
+++ b/events/arm/armv7-ca17/unit_masks
@@ -0,0 +1,4 @@
+# ARM V7 PMNC possible unit masks
+#
+name:zero type:mandatory default:0x00
+ 0x00 No unit mask
diff --git a/libop/op_cpu_type.c b/libop/op_cpu_type.c
index ea5ff76..d03f9d6 100644
--- a/libop/op_cpu_type.c
+++ b/libop/op_cpu_type.c
@@ -124,6 +124,7 @@ static struct cpu_descr const cpu_descrs[MAX_CPU_TYPE] = {
{ "Intel Skylake microarchitecture", "i386/skylake", CPU_SKYLAKE, 4 },
{ "Intel Goldmont microarchitecture", "i386/goldmont", CPU_GOLDMONT, 4 },
{ "IBM z13", "s390/z13", CPU_S390_Z13, 1 },
+ { "ARM Cortex-A17", "arm/armv7-ca17", CPU_ARM_V7_CA17, 7 },
};

static size_t const nr_cpu_descrs = sizeof(cpu_descrs) / sizeof(struct cpu_descr);
@@ -413,6 +414,8 @@ static op_cpu _get_arm_cpu_type(void)
return op_get_cpu_number("arm/armv7-ca9");
case 0xc0f:
return op_get_cpu_number("arm/armv7-ca15");
+ case 0xc0e:
+ return op_get_cpu_number("arm/armv7-ca17");
case 0xd07:
return op_get_cpu_number("arm/armv8-ca57");
case 0xd03:
diff --git a/libop/op_cpu_type.h b/libop/op_cpu_type.h
index 4f896a0..8d2badd 100644
--- a/libop/op_cpu_type.h
+++ b/libop/op_cpu_type.h
@@ -104,6 +104,7 @@ typedef enum {
CPU_SKYLAKE, /** < Intel Skylake microarchitecture */
CPU_GOLDMONT, /** < Intel Goldmont microarchitecture */
CPU_S390_Z13, /** < IBM z13 */
+ CPU_ARM_V7_CA17, /* ARM Cortex-A17 */
MAX_CPU_TYPE
} op_cpu;

diff --git a/libop/op_events.c b/libop/op_events.c
index ea6ced3..0c7e9bc 100644
--- a/libop/op_events.c
+++ b/libop/op_events.c
@@ -1238,6 +1238,7 @@ void op_default_event(op_cpu cpu_type, struct op_default_event_descr * descr)
case CPU_ARM_V7_CA7:
case CPU_ARM_V7_CA9:
case CPU_ARM_V7_CA15:
+ case CPU_ARM_V7_CA17:
case CPU_ARM_SCORPION:
case CPU_ARM_SCORPIONMP:
case CPU_ARM_KRAIT:
diff --git a/utils/ophelp.c b/utils/ophelp.c
index 3cb1c08..ff4e3f1 100644
--- a/utils/ophelp.c
+++ b/utils/ophelp.c
@@ -630,6 +630,12 @@ int main(int argc, char const * argv[])
"Cortex A15 DDI (ARM DDI 0438F, revision r3p1)\n";
break;

+ case CPU_ARM_V7_CA17:
+ event_doc =
+ "See Cortex-A17 MPCore Technical Reference Manual\n"
+ "Cortex A17 DDI (ARM DDI 0535C, revision r1p1)\n";
+ break;
+
case CPU_ARM_V8_APM_XGENE:
event_doc =
"See ARM Architecture Reference Manual \n"

[William Cohen]

Post by Hyeongsik Min
Added the event list of the ARM Cortex A17 based on
ARM DDI 0535C.
---
events/Makefile.am | 1 +
events/arm/armv7-ca17/events | 77 ++++++++++++++++++++++++++++++++++++++++
events/arm/armv7-ca17/unit_masks | 4 +++
libop/op_cpu_type.c | 3 ++
libop/op_cpu_type.h | 1 +
libop/op_events.c | 1 +
utils/ophelp.c | 6 ++++
7 files changed, 93 insertions(+)
create mode 100644 events/arm/armv7-ca17/events
create mode 100644 events/arm/armv7-ca17/unit_masks

Hi,

Thanks for the patch. The patch looks good and has been applied to the git repository. It would be good to have a matching patch for the oprofile-tests ( https://sourceforge.net/p/oprofile/oprofile-tests/ci/master/tree/), so that the OProfile testsuite can be run on ARM Cortex A17 machines.

-Will