* [PATCH 1/5] perf, x86, uncore: Fix section mismatch in split uncore driver
@ 2014-09-04 23:08 Andi Kleen
2014-09-04 23:08 ` [PATCH 2/5] perf, x86, uncore: Add Haswell-EP uncore support Andi Kleen
` (4 more replies)
0 siblings, 5 replies; 17+ messages in thread
From: Andi Kleen @ 2014-09-04 23:08 UTC (permalink / raw)
To: peterz; +Cc: linux-kernel, mingo, eranian, tglx, Andi Kleen
From: Andi Kleen <ak@linux.intel.com>
The new split Intel uncore driver code that recently went
into tip added a section mismatch, which the build process
complains about.
uncore_pmu_register can be called from uncore_pci_probe,
which is not __init and can be called from pci driver ->probe.
I'm not fully sure if it's actually possible to call the probe
function later, but it seems safer to mark uncore_pmu_register
not __init.
This also fixes the warning.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
---
arch/x86/kernel/cpu/perf_event_intel_uncore.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore.c b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
index 4785ee8..812ec5d 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
@@ -661,7 +661,7 @@ static struct attribute_group uncore_pmu_attr_group = {
.attrs = uncore_pmu_attrs,
};
-static int __init uncore_pmu_register(struct intel_uncore_pmu *pmu)
+static int uncore_pmu_register(struct intel_uncore_pmu *pmu)
{
int ret;
--
1.9.3
^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 2/5] perf, x86, uncore: Add Haswell-EP uncore support
2014-09-04 23:08 [PATCH 1/5] perf, x86, uncore: Fix section mismatch in split uncore driver Andi Kleen
@ 2014-09-04 23:08 ` Andi Kleen
2014-09-24 15:00 ` [tip:perf/core] perf/x86/intel/uncore: " tip-bot for Yan, Zheng
2014-10-22 19:07 ` [PATCH 2/5] perf, x86, uncore: " Alexei Starovoitov
2014-09-04 23:08 ` [PATCH 3/5] perf, x86, uncore: Register the PMU only if the uncore pci device exists Andi Kleen
` (3 subsequent siblings)
4 siblings, 2 replies; 17+ messages in thread
From: Andi Kleen @ 2014-09-04 23:08 UTC (permalink / raw)
To: peterz; +Cc: linux-kernel, mingo, eranian, tglx, Yan, Zheng, Andi Kleen
From: "Yan, Zheng" <zheng.z.yan@intel.com>
The uncore subsystem in Haswell-EP is similar to Sandy/Ivy
Bridge-EP. There are some differences in config register
encoding and pci device IDs. The Haswell-EP uncore also
supports a few new events. Add the Haswell-EP driver to
the snbep split driver.
v2: (AK): Add missing break. Add imc events. Add cbox nc/isoc/c6.
Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
---
arch/x86/kernel/cpu/perf_event_intel_uncore.c | 7 +-
arch/x86/kernel/cpu/perf_event_intel_uncore.h | 2 +
.../x86/kernel/cpu/perf_event_intel_uncore_snbep.c | 608 ++++++++++++++++++++-
3 files changed, 615 insertions(+), 2 deletions(-)
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore.c b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
index 812ec5d..2b0fa02 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
@@ -883,6 +883,9 @@ static int __init uncore_pci_init(void)
case 62: /* Ivy Bridge-EP */
ret = ivbep_uncore_pci_init();
break;
+ case 63: /* Haswell-EP */
+ ret = hswep_uncore_pci_init();
+ break;
case 42: /* Sandy Bridge */
ret = snb_uncore_pci_init();
break;
@@ -1190,7 +1193,9 @@ static int __init uncore_cpu_init(void)
case 62: /* Ivy Bridge-EP */
ivbep_uncore_cpu_init();
break;
-
+ case 63: /* Haswell-EP */
+ hswep_uncore_cpu_init();
+ break;
default:
return 0;
}
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore.h b/arch/x86/kernel/cpu/perf_event_intel_uncore.h
index 1d7e894..18eb78b 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore.h
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore.h
@@ -332,6 +332,8 @@ int snbep_uncore_pci_init(void);
void snbep_uncore_cpu_init(void);
int ivbep_uncore_pci_init(void);
void ivbep_uncore_cpu_init(void);
+int hswep_uncore_pci_init(void);
+void hswep_uncore_cpu_init(void);
/* perf_event_intel_uncore_nhmex.c */
void nhmex_uncore_cpu_init(void);
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c b/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
index 6606ed0..a4fe9fa 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
@@ -163,6 +163,53 @@
#define __BITS_VALUE(x, i, n) ((typeof(x))(((x) >> ((i) * (n))) & \
((1ULL << (n)) - 1)))
+/* Haswell-EP Ubox */
+#define HSWEP_U_MSR_PMON_CTR0 0x705
+#define HSWEP_U_MSR_PMON_CTL0 0x709
+#define HSWEP_U_MSR_PMON_FILTER 0x707
+
+#define HSWEP_U_MSR_PMON_UCLK_FIXED_CTL 0x703
+#define HSWEP_U_MSR_PMON_UCLK_FIXED_CTR 0x704
+
+#define HSWEP_U_MSR_PMON_BOX_FILTER_TID (0x1 << 0)
+#define HSWEP_U_MSR_PMON_BOX_FILTER_CID (0x1fULL << 1)
+#define HSWEP_U_MSR_PMON_BOX_FILTER_MASK \
+ (HSWEP_U_MSR_PMON_BOX_FILTER_TID | \
+ HSWEP_U_MSR_PMON_BOX_FILTER_CID)
+
+/* Haswell-EP CBo */
+#define HSWEP_C0_MSR_PMON_CTR0 0xe08
+#define HSWEP_C0_MSR_PMON_CTL0 0xe01
+#define HSWEP_C0_MSR_PMON_BOX_CTL 0xe00
+#define HSWEP_C0_MSR_PMON_BOX_FILTER0 0xe05
+#define HSWEP_CBO_MSR_OFFSET 0x10
+
+
+#define HSWEP_CB0_MSR_PMON_BOX_FILTER_TID (0x3fULL << 0)
+#define HSWEP_CB0_MSR_PMON_BOX_FILTER_LINK (0xfULL << 6)
+#define HSWEP_CB0_MSR_PMON_BOX_FILTER_STATE (0x7fULL << 17)
+#define HSWEP_CB0_MSR_PMON_BOX_FILTER_NID (0xffffULL << 32)
+#define HSWEP_CB0_MSR_PMON_BOX_FILTER_OPC (0x1ffULL << 52)
+#define HSWEP_CB0_MSR_PMON_BOX_FILTER_C6 (0x1ULL << 61)
+#define HSWEP_CB0_MSR_PMON_BOX_FILTER_NC (0x1ULL << 62)
+#define HSWEP_CB0_MSR_PMON_BOX_FILTER_ISOC (0x1ULL << 63)
+
+
+/* Haswell-EP Sbox */
+#define HSWEP_S0_MSR_PMON_CTR0 0x726
+#define HSWEP_S0_MSR_PMON_CTL0 0x721
+#define HSWEP_S0_MSR_PMON_BOX_CTL 0x720
+#define HSWEP_SBOX_MSR_OFFSET 0xa
+#define HSWEP_S_MSR_PMON_RAW_EVENT_MASK (SNBEP_PMON_RAW_EVENT_MASK | \
+ SNBEP_CBO_PMON_CTL_TID_EN)
+
+/* Haswell-EP PCU */
+#define HSWEP_PCU_MSR_PMON_CTR0 0x717
+#define HSWEP_PCU_MSR_PMON_CTL0 0x711
+#define HSWEP_PCU_MSR_PMON_BOX_CTL 0x710
+#define HSWEP_PCU_MSR_PMON_BOX_FILTER 0x715
+
+
DEFINE_UNCORE_FORMAT_ATTR(event, event, "config:0-7");
DEFINE_UNCORE_FORMAT_ATTR(event_ext, event, "config:0-7,21");
DEFINE_UNCORE_FORMAT_ATTR(umask, umask, "config:8-15");
@@ -175,13 +222,21 @@ DEFINE_UNCORE_FORMAT_ATTR(occ_sel, occ_sel, "config:14-15");
DEFINE_UNCORE_FORMAT_ATTR(occ_invert, occ_invert, "config:30");
DEFINE_UNCORE_FORMAT_ATTR(occ_edge, occ_edge, "config:14-51");
DEFINE_UNCORE_FORMAT_ATTR(filter_tid, filter_tid, "config1:0-4");
+DEFINE_UNCORE_FORMAT_ATTR(filter_tid2, filter_tid, "config1:0");
+DEFINE_UNCORE_FORMAT_ATTR(filter_tid3, filter_tid, "config1:0-5");
+DEFINE_UNCORE_FORMAT_ATTR(filter_cid, filter_cid, "config1:5");
DEFINE_UNCORE_FORMAT_ATTR(filter_link, filter_link, "config1:5-8");
+DEFINE_UNCORE_FORMAT_ATTR(filter_link2, filter_link, "config1:6-8");
DEFINE_UNCORE_FORMAT_ATTR(filter_nid, filter_nid, "config1:10-17");
DEFINE_UNCORE_FORMAT_ATTR(filter_nid2, filter_nid, "config1:32-47");
DEFINE_UNCORE_FORMAT_ATTR(filter_state, filter_state, "config1:18-22");
DEFINE_UNCORE_FORMAT_ATTR(filter_state2, filter_state, "config1:17-22");
+DEFINE_UNCORE_FORMAT_ATTR(filter_state3, filter_state, "config1:17-23");
DEFINE_UNCORE_FORMAT_ATTR(filter_opc, filter_opc, "config1:23-31");
DEFINE_UNCORE_FORMAT_ATTR(filter_opc2, filter_opc, "config1:52-60");
+DEFINE_UNCORE_FORMAT_ATTR(filter_nc, filter_nc, "config1:62");
+DEFINE_UNCORE_FORMAT_ATTR(filter_c6, filter_c6, "config1:61");
+DEFINE_UNCORE_FORMAT_ATTR(filter_isoc, filter_isoc, "config1:63");
DEFINE_UNCORE_FORMAT_ATTR(filter_band0, filter_band0, "config1:0-7");
DEFINE_UNCORE_FORMAT_ATTR(filter_band1, filter_band1, "config1:8-15");
DEFINE_UNCORE_FORMAT_ATTR(filter_band2, filter_band2, "config1:16-23");
@@ -857,7 +912,6 @@ static void snbep_qpi_enable_event(struct intel_uncore_box *box, struct perf_eve
if (reg1->idx != EXTRA_REG_NONE) {
int idx = box->pmu->pmu_idx + SNBEP_PCI_QPI_PORT0_FILTER;
struct pci_dev *filter_pdev = uncore_extra_pci_dev[box->phys_id][idx];
- WARN_ON_ONCE(!filter_pdev);
if (filter_pdev) {
pci_write_config_dword(filter_pdev, reg1->reg,
(u32)reg1->config);
@@ -1643,3 +1697,555 @@ int ivbep_uncore_pci_init(void)
return 0;
}
/* end of IvyTown uncore support */
+
+/* Haswell-EP uncore support */
+static struct attribute *hswep_uncore_ubox_formats_attr[] = {
+ &format_attr_event.attr,
+ &format_attr_umask.attr,
+ &format_attr_edge.attr,
+ &format_attr_inv.attr,
+ &format_attr_thresh5.attr,
+ &format_attr_filter_tid2.attr,
+ &format_attr_filter_cid.attr,
+ NULL,
+};
+
+static struct attribute_group hswep_uncore_ubox_format_group = {
+ .name = "format",
+ .attrs = hswep_uncore_ubox_formats_attr,
+};
+
+static int hswep_ubox_hw_config(struct intel_uncore_box *box, struct perf_event *event)
+{
+ struct hw_perf_event_extra *reg1 = &event->hw.extra_reg;
+ reg1->reg = HSWEP_U_MSR_PMON_FILTER;
+ reg1->config = event->attr.config1 & HSWEP_U_MSR_PMON_BOX_FILTER_MASK;
+ reg1->idx = 0;
+ return 0;
+}
+
+static struct intel_uncore_ops hswep_uncore_ubox_ops = {
+ SNBEP_UNCORE_MSR_OPS_COMMON_INIT(),
+ .hw_config = hswep_ubox_hw_config,
+ .get_constraint = uncore_get_constraint,
+ .put_constraint = uncore_put_constraint,
+};
+
+static struct intel_uncore_type hswep_uncore_ubox = {
+ .name = "ubox",
+ .num_counters = 2,
+ .num_boxes = 1,
+ .perf_ctr_bits = 44,
+ .fixed_ctr_bits = 48,
+ .perf_ctr = HSWEP_U_MSR_PMON_CTR0,
+ .event_ctl = HSWEP_U_MSR_PMON_CTL0,
+ .event_mask = SNBEP_U_MSR_PMON_RAW_EVENT_MASK,
+ .fixed_ctr = HSWEP_U_MSR_PMON_UCLK_FIXED_CTR,
+ .fixed_ctl = HSWEP_U_MSR_PMON_UCLK_FIXED_CTL,
+ .num_shared_regs = 1,
+ .ops = &hswep_uncore_ubox_ops,
+ .format_group = &hswep_uncore_ubox_format_group,
+};
+
+static struct attribute *hswep_uncore_cbox_formats_attr[] = {
+ &format_attr_event.attr,
+ &format_attr_umask.attr,
+ &format_attr_edge.attr,
+ &format_attr_tid_en.attr,
+ &format_attr_thresh8.attr,
+ &format_attr_filter_tid3.attr,
+ &format_attr_filter_link2.attr,
+ &format_attr_filter_state3.attr,
+ &format_attr_filter_nid2.attr,
+ &format_attr_filter_opc2.attr,
+ &format_attr_filter_nc.attr,
+ &format_attr_filter_c6.attr,
+ &format_attr_filter_isoc.attr,
+ NULL,
+};
+
+static struct attribute_group hswep_uncore_cbox_format_group = {
+ .name = "format",
+ .attrs = hswep_uncore_cbox_formats_attr,
+};
+
+static struct event_constraint hswep_uncore_cbox_constraints[] = {
+ UNCORE_EVENT_CONSTRAINT(0x01, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x09, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x11, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x36, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x38, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x3b, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x3e, 0x1),
+ EVENT_CONSTRAINT_END
+};
+
+static struct extra_reg hswep_uncore_cbox_extra_regs[] = {
+ SNBEP_CBO_EVENT_EXTRA_REG(SNBEP_CBO_PMON_CTL_TID_EN,
+ SNBEP_CBO_PMON_CTL_TID_EN, 0x1),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x0334, 0xffff, 0x4),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x0534, 0xffff, 0x4),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x0934, 0xffff, 0x4),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x1134, 0xffff, 0x4),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x2134, 0xffff, 0x4),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4134, 0xffff, 0x4),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4037, 0x40ff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4028, 0x40ff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4032, 0x40ff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4029, 0x40ff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4033, 0x40ff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x402A, 0x40ff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x0135, 0xffff, 0x12),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x0335, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4135, 0xffff, 0x18),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4435, 0xffff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4835, 0xffff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x5035, 0xffff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4335, 0xffff, 0x18),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4a35, 0xffff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x2335, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x8335, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x2135, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x8135, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x0136, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x0336, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4136, 0xffff, 0x18),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4436, 0xffff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4836, 0xffff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4336, 0xffff, 0x18),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4a36, 0xffff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x2336, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x8336, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x2136, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x8136, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x5036, 0xffff, 0x8),
+ EVENT_EXTRA_END
+};
+
+static u64 hswep_cbox_filter_mask(int fields)
+{
+ u64 mask = 0;
+ if (fields & 0x1)
+ mask |= HSWEP_CB0_MSR_PMON_BOX_FILTER_TID;
+ if (fields & 0x2)
+ mask |= HSWEP_CB0_MSR_PMON_BOX_FILTER_LINK;
+ if (fields & 0x4)
+ mask |= HSWEP_CB0_MSR_PMON_BOX_FILTER_STATE;
+ if (fields & 0x8)
+ mask |= HSWEP_CB0_MSR_PMON_BOX_FILTER_NID;
+ if (fields & 0x10) {
+ mask |= HSWEP_CB0_MSR_PMON_BOX_FILTER_OPC;
+ mask |= HSWEP_CB0_MSR_PMON_BOX_FILTER_NC;
+ mask |= HSWEP_CB0_MSR_PMON_BOX_FILTER_C6;
+ mask |= HSWEP_CB0_MSR_PMON_BOX_FILTER_ISOC;
+ }
+ return mask;
+}
+
+static struct event_constraint *
+hswep_cbox_get_constraint(struct intel_uncore_box *box, struct perf_event *event)
+{
+ return __snbep_cbox_get_constraint(box, event, hswep_cbox_filter_mask);
+}
+
+static int hswep_cbox_hw_config(struct intel_uncore_box *box, struct perf_event *event)
+{
+ struct hw_perf_event_extra *reg1 = &event->hw.extra_reg;
+ struct extra_reg *er;
+ int idx = 0;
+
+ for (er = hswep_uncore_cbox_extra_regs; er->msr; er++) {
+ if (er->event != (event->hw.config & er->config_mask))
+ continue;
+ idx |= er->idx;
+ }
+
+ if (idx) {
+ reg1->reg = HSWEP_C0_MSR_PMON_BOX_FILTER0 +
+ HSWEP_CBO_MSR_OFFSET * box->pmu->pmu_idx;
+ reg1->config = event->attr.config1 & hswep_cbox_filter_mask(idx);
+ reg1->idx = idx;
+ }
+ return 0;
+}
+
+static void hswep_cbox_enable_event(struct intel_uncore_box *box,
+ struct perf_event *event)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ struct hw_perf_event_extra *reg1 = &hwc->extra_reg;
+
+ if (reg1->idx != EXTRA_REG_NONE) {
+ u64 filter = uncore_shared_reg_config(box, 0);
+ wrmsrl(reg1->reg, filter & 0xffffffff);
+ wrmsrl(reg1->reg + 1, filter >> 32);
+ }
+
+ wrmsrl(hwc->config_base, hwc->config | SNBEP_PMON_CTL_EN);
+}
+
+static struct intel_uncore_ops hswep_uncore_cbox_ops = {
+ .init_box = snbep_uncore_msr_init_box,
+ .disable_box = snbep_uncore_msr_disable_box,
+ .enable_box = snbep_uncore_msr_enable_box,
+ .disable_event = snbep_uncore_msr_disable_event,
+ .enable_event = hswep_cbox_enable_event,
+ .read_counter = uncore_msr_read_counter,
+ .hw_config = hswep_cbox_hw_config,
+ .get_constraint = hswep_cbox_get_constraint,
+ .put_constraint = snbep_cbox_put_constraint,
+};
+
+static struct intel_uncore_type hswep_uncore_cbox = {
+ .name = "cbox",
+ .num_counters = 4,
+ .num_boxes = 18,
+ .perf_ctr_bits = 44,
+ .event_ctl = HSWEP_C0_MSR_PMON_CTL0,
+ .perf_ctr = HSWEP_C0_MSR_PMON_CTR0,
+ .event_mask = SNBEP_CBO_MSR_PMON_RAW_EVENT_MASK,
+ .box_ctl = HSWEP_C0_MSR_PMON_BOX_CTL,
+ .msr_offset = HSWEP_CBO_MSR_OFFSET,
+ .num_shared_regs = 1,
+ .constraints = hswep_uncore_cbox_constraints,
+ .ops = &hswep_uncore_cbox_ops,
+ .format_group = &hswep_uncore_cbox_format_group,
+};
+
+static struct attribute *hswep_uncore_sbox_formats_attr[] = {
+ &format_attr_event.attr,
+ &format_attr_umask.attr,
+ &format_attr_edge.attr,
+ &format_attr_tid_en.attr,
+ &format_attr_inv.attr,
+ &format_attr_thresh8.attr,
+ NULL,
+};
+
+static struct attribute_group hswep_uncore_sbox_format_group = {
+ .name = "format",
+ .attrs = hswep_uncore_sbox_formats_attr,
+};
+
+static struct intel_uncore_type hswep_uncore_sbox = {
+ .name = "sbox",
+ .num_counters = 4,
+ .num_boxes = 4,
+ .perf_ctr_bits = 44,
+ .event_ctl = HSWEP_S0_MSR_PMON_CTL0,
+ .perf_ctr = HSWEP_S0_MSR_PMON_CTR0,
+ .event_mask = HSWEP_S_MSR_PMON_RAW_EVENT_MASK,
+ .box_ctl = HSWEP_S0_MSR_PMON_BOX_CTL,
+ .msr_offset = HSWEP_SBOX_MSR_OFFSET,
+ .ops = &snbep_uncore_msr_ops,
+ .format_group = &hswep_uncore_sbox_format_group,
+};
+
+static int hswep_pcu_hw_config(struct intel_uncore_box *box, struct perf_event *event)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ struct hw_perf_event_extra *reg1 = &hwc->extra_reg;
+ int ev_sel = hwc->config & SNBEP_PMON_CTL_EV_SEL_MASK;
+
+ if (ev_sel >= 0xb && ev_sel <= 0xe) {
+ reg1->reg = HSWEP_PCU_MSR_PMON_BOX_FILTER;
+ reg1->idx = ev_sel - 0xb;
+ reg1->config = event->attr.config1 & (0xff << reg1->idx);
+ }
+ return 0;
+}
+
+static struct intel_uncore_ops hswep_uncore_pcu_ops = {
+ SNBEP_UNCORE_MSR_OPS_COMMON_INIT(),
+ .hw_config = hswep_pcu_hw_config,
+ .get_constraint = snbep_pcu_get_constraint,
+ .put_constraint = snbep_pcu_put_constraint,
+};
+
+static struct intel_uncore_type hswep_uncore_pcu = {
+ .name = "pcu",
+ .num_counters = 4,
+ .num_boxes = 1,
+ .perf_ctr_bits = 48,
+ .perf_ctr = HSWEP_PCU_MSR_PMON_CTR0,
+ .event_ctl = HSWEP_PCU_MSR_PMON_CTL0,
+ .event_mask = SNBEP_PCU_MSR_PMON_RAW_EVENT_MASK,
+ .box_ctl = HSWEP_PCU_MSR_PMON_BOX_CTL,
+ .num_shared_regs = 1,
+ .ops = &hswep_uncore_pcu_ops,
+ .format_group = &snbep_uncore_pcu_format_group,
+};
+
+static struct intel_uncore_type *hswep_msr_uncores[] = {
+ &hswep_uncore_ubox,
+ &hswep_uncore_cbox,
+ &hswep_uncore_sbox,
+ &hswep_uncore_pcu,
+ NULL,
+};
+
+void hswep_uncore_cpu_init(void)
+{
+ if (hswep_uncore_cbox.num_boxes > boot_cpu_data.x86_max_cores)
+ hswep_uncore_cbox.num_boxes = boot_cpu_data.x86_max_cores;
+ uncore_msr_uncores = hswep_msr_uncores;
+}
+
+static struct intel_uncore_type hswep_uncore_ha = {
+ .name = "ha",
+ .num_counters = 5,
+ .num_boxes = 2,
+ .perf_ctr_bits = 48,
+ SNBEP_UNCORE_PCI_COMMON_INIT(),
+};
+
+static struct uncore_event_desc hswep_uncore_imc_events[] = {
+ INTEL_UNCORE_EVENT_DESC(clockticks, "event=0x00,umask=0x00"),
+ INTEL_UNCORE_EVENT_DESC(cas_count_read, "event=0x04,umask=0x03"),
+ INTEL_UNCORE_EVENT_DESC(cas_count_write, "event=0x04,umask=0x0c"),
+ { /* end: all zeroes */ },
+};
+
+static struct intel_uncore_type hswep_uncore_imc = {
+ .name = "imc",
+ .num_counters = 5,
+ .num_boxes = 8,
+ .perf_ctr_bits = 48,
+ .fixed_ctr_bits = 48,
+ .fixed_ctr = SNBEP_MC_CHy_PCI_PMON_FIXED_CTR,
+ .fixed_ctl = SNBEP_MC_CHy_PCI_PMON_FIXED_CTL,
+ .event_descs = hswep_uncore_imc_events,
+ SNBEP_UNCORE_PCI_COMMON_INIT(),
+};
+
+static struct intel_uncore_ops hswep_uncore_irp_ops = {
+ .init_box = snbep_uncore_pci_init_box,
+ .disable_box = snbep_uncore_pci_disable_box,
+ .enable_box = snbep_uncore_pci_enable_box,
+ .disable_event = ivbep_uncore_irp_disable_event,
+ .enable_event = ivbep_uncore_irp_enable_event,
+ .read_counter = ivbep_uncore_irp_read_counter,
+};
+
+static struct intel_uncore_type hswep_uncore_irp = {
+ .name = "irp",
+ .num_counters = 4,
+ .num_boxes = 1,
+ .perf_ctr_bits = 48,
+ .event_mask = SNBEP_PMON_RAW_EVENT_MASK,
+ .box_ctl = SNBEP_PCI_PMON_BOX_CTL,
+ .ops = &hswep_uncore_irp_ops,
+ .format_group = &snbep_uncore_format_group,
+};
+
+static struct intel_uncore_type hswep_uncore_qpi = {
+ .name = "qpi",
+ .num_counters = 5,
+ .num_boxes = 3,
+ .perf_ctr_bits = 48,
+ .perf_ctr = SNBEP_PCI_PMON_CTR0,
+ .event_ctl = SNBEP_PCI_PMON_CTL0,
+ .event_mask = SNBEP_QPI_PCI_PMON_RAW_EVENT_MASK,
+ .box_ctl = SNBEP_PCI_PMON_BOX_CTL,
+ .num_shared_regs = 1,
+ .ops = &snbep_uncore_qpi_ops,
+ .format_group = &snbep_uncore_qpi_format_group,
+};
+
+static struct event_constraint hswep_uncore_r2pcie_constraints[] = {
+ UNCORE_EVENT_CONSTRAINT(0x10, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x11, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x13, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x23, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x24, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x25, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x26, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x27, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x28, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x29, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x2a, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x2b, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x2c, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x2d, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x32, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x33, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x34, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x35, 0x3),
+ EVENT_CONSTRAINT_END
+};
+
+static struct intel_uncore_type hswep_uncore_r2pcie = {
+ .name = "r2pcie",
+ .num_counters = 4,
+ .num_boxes = 1,
+ .perf_ctr_bits = 48,
+ .constraints = hswep_uncore_r2pcie_constraints,
+ SNBEP_UNCORE_PCI_COMMON_INIT(),
+};
+
+static struct event_constraint hswep_uncore_r3qpi_constraints[] = {
+ UNCORE_EVENT_CONSTRAINT(0x01, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x07, 0x7),
+ UNCORE_EVENT_CONSTRAINT(0x08, 0x7),
+ UNCORE_EVENT_CONSTRAINT(0x09, 0x7),
+ UNCORE_EVENT_CONSTRAINT(0x0a, 0x7),
+ UNCORE_EVENT_CONSTRAINT(0x0e, 0x7),
+ UNCORE_EVENT_CONSTRAINT(0x10, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x11, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x12, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x13, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x14, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x15, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x1f, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x20, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x21, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x22, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x23, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x25, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x26, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x28, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x29, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x2c, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x2d, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x2e, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x2f, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x31, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x32, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x33, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x34, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x36, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x37, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x38, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x39, 0x3),
+ EVENT_CONSTRAINT_END
+};
+
+static struct intel_uncore_type hswep_uncore_r3qpi = {
+ .name = "r3qpi",
+ .num_counters = 4,
+ .num_boxes = 3,
+ .perf_ctr_bits = 44,
+ .constraints = hswep_uncore_r3qpi_constraints,
+ SNBEP_UNCORE_PCI_COMMON_INIT(),
+};
+
+enum {
+ HSWEP_PCI_UNCORE_HA,
+ HSWEP_PCI_UNCORE_IMC,
+ HSWEP_PCI_UNCORE_IRP,
+ HSWEP_PCI_UNCORE_QPI,
+ HSWEP_PCI_UNCORE_R2PCIE,
+ HSWEP_PCI_UNCORE_R3QPI,
+};
+
+static struct intel_uncore_type *hswep_pci_uncores[] = {
+ [HSWEP_PCI_UNCORE_HA] = &hswep_uncore_ha,
+ [HSWEP_PCI_UNCORE_IMC] = &hswep_uncore_imc,
+ [HSWEP_PCI_UNCORE_IRP] = &hswep_uncore_irp,
+ [HSWEP_PCI_UNCORE_QPI] = &hswep_uncore_qpi,
+ [HSWEP_PCI_UNCORE_R2PCIE] = &hswep_uncore_r2pcie,
+ [HSWEP_PCI_UNCORE_R3QPI] = &hswep_uncore_r3qpi,
+ NULL,
+};
+
+static DEFINE_PCI_DEVICE_TABLE(hswep_uncore_pci_ids) = {
+ { /* Home Agent 0 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f30),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_HA, 0),
+ },
+ { /* Home Agent 1 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f38),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_HA, 1),
+ },
+ { /* MC0 Channel 0 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2fb0),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IMC, 0),
+ },
+ { /* MC0 Channel 1 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2fb1),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IMC, 1),
+ },
+ { /* MC0 Channel 2 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2fb4),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IMC, 2),
+ },
+ { /* MC0 Channel 3 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2fb5),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IMC, 3),
+ },
+ { /* MC1 Channel 0 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2fd0),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IMC, 4),
+ },
+ { /* MC1 Channel 1 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2fd1),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IMC, 5),
+ },
+ { /* MC1 Channel 2 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2fd4),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IMC, 6),
+ },
+ { /* MC1 Channel 3 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2fd5),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IMC, 7),
+ },
+ { /* IRP */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f39),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IRP, 0),
+ },
+ { /* QPI0 Port 0 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f32),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_QPI, 0),
+ },
+ { /* QPI0 Port 1 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f33),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_QPI, 1),
+ },
+ { /* QPI1 Port 2 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f3a),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_QPI, 2),
+ },
+ { /* R2PCIe */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f34),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_R2PCIE, 0),
+ },
+ { /* R3QPI0 Link 0 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f36),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_R3QPI, 0),
+ },
+ { /* R3QPI0 Link 1 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f37),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_R3QPI, 1),
+ },
+ { /* R3QPI1 Link 2 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f3e),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_R3QPI, 2),
+ },
+ { /* QPI Port 0 filter */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f86),
+ .driver_data = UNCORE_PCI_DEV_DATA(UNCORE_EXTRA_PCI_DEV,
+ SNBEP_PCI_QPI_PORT0_FILTER),
+ },
+ { /* QPI Port 1 filter */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f96),
+ .driver_data = UNCORE_PCI_DEV_DATA(UNCORE_EXTRA_PCI_DEV,
+ SNBEP_PCI_QPI_PORT1_FILTER),
+ },
+ { /* end: all zeroes */ }
+};
+
+static struct pci_driver hswep_uncore_pci_driver = {
+ .name = "hswep_uncore",
+ .id_table = hswep_uncore_pci_ids,
+};
+
+int hswep_uncore_pci_init(void)
+{
+ int ret = snbep_pci2phy_map_init(0x2f1e);
+ if (ret)
+ return ret;
+ uncore_pci_uncores = hswep_pci_uncores;
+ uncore_pci_driver = &hswep_uncore_pci_driver;
+ return 0;
+}
+/* end of Haswell-EP uncore support */
--
1.9.3
^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 3/5] perf, x86, uncore: Register the PMU only if the uncore pci device exists
2014-09-04 23:08 [PATCH 1/5] perf, x86, uncore: Fix section mismatch in split uncore driver Andi Kleen
2014-09-04 23:08 ` [PATCH 2/5] perf, x86, uncore: Add Haswell-EP uncore support Andi Kleen
@ 2014-09-04 23:08 ` Andi Kleen
2014-09-24 15:00 ` [tip:perf/core] perf/x86/intel/uncore: " tip-bot for Yan, Zheng
2014-09-04 23:08 ` [PATCH 4/5] perf, x86, uncore: Add missing cbox filter flags on IvyBridge-EP uncore driver Andi Kleen
` (2 subsequent siblings)
4 siblings, 1 reply; 17+ messages in thread
From: Andi Kleen @ 2014-09-04 23:08 UTC (permalink / raw)
To: peterz; +Cc: linux-kernel, mingo, eranian, tglx, Yan, Zheng, Andi Kleen
From: "Yan, Zheng" <zheng.z.yan@intel.com>
Current code registers PMUs for all possible uncore pci devices.
This is not good because, on some machines, one or more uncore pci
devices can be missing. The missing pci device make corresponding
PMU unusable. Register the PMU only if the uncore device exists.
Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
---
arch/x86/kernel/cpu/perf_event_intel_uncore.c | 19 +++++++++++--------
1 file changed, 11 insertions(+), 8 deletions(-)
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore.c b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
index 2b0fa02..42d00e5 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
@@ -793,6 +793,7 @@ static int uncore_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id
struct intel_uncore_box *box;
struct intel_uncore_type *type;
int phys_id;
+ bool first_box = false;
phys_id = uncore_pcibus_to_physid[pdev->bus->number];
if (phys_id < 0)
@@ -827,9 +828,13 @@ static int uncore_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id
pci_set_drvdata(pdev, box);
raw_spin_lock(&uncore_box_lock);
+ if (list_empty(&pmu->box_list))
+ first_box = true;
list_add_tail(&box->list, &pmu->box_list);
raw_spin_unlock(&uncore_box_lock);
+ if (first_box)
+ uncore_pmu_register(pmu);
return 0;
}
@@ -838,6 +843,7 @@ static void uncore_pci_remove(struct pci_dev *pdev)
struct intel_uncore_box *box = pci_get_drvdata(pdev);
struct intel_uncore_pmu *pmu;
int i, cpu, phys_id = uncore_pcibus_to_physid[pdev->bus->number];
+ bool last_box = false;
box = pci_get_drvdata(pdev);
if (!box) {
@@ -859,6 +865,8 @@ static void uncore_pci_remove(struct pci_dev *pdev)
raw_spin_lock(&uncore_box_lock);
list_del(&box->list);
+ if (list_empty(&pmu->box_list))
+ last_box = true;
raw_spin_unlock(&uncore_box_lock);
for_each_possible_cpu(cpu) {
@@ -870,6 +878,9 @@ static void uncore_pci_remove(struct pci_dev *pdev)
WARN_ON_ONCE(atomic_read(&box->refcnt) != 1);
kfree(box);
+
+ if (last_box)
+ perf_pmu_unregister(&pmu->pmu);
}
static int __init uncore_pci_init(void)
@@ -1221,14 +1232,6 @@ static int __init uncore_pmus_register(void)
}
}
- for (i = 0; uncore_pci_uncores[i]; i++) {
- type = uncore_pci_uncores[i];
- for (j = 0; j < type->num_boxes; j++) {
- pmu = &type->pmus[j];
- uncore_pmu_register(pmu);
- }
- }
-
return 0;
}
--
1.9.3
^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 4/5] perf, x86, uncore: Add missing cbox filter flags on IvyBridge-EP uncore driver
2014-09-04 23:08 [PATCH 1/5] perf, x86, uncore: Fix section mismatch in split uncore driver Andi Kleen
2014-09-04 23:08 ` [PATCH 2/5] perf, x86, uncore: Add Haswell-EP uncore support Andi Kleen
2014-09-04 23:08 ` [PATCH 3/5] perf, x86, uncore: Register the PMU only if the uncore pci device exists Andi Kleen
@ 2014-09-04 23:08 ` Andi Kleen
2014-09-24 15:00 ` [tip:perf/core] perf/x86/intel/uncore: " tip-bot for Andi Kleen
2014-09-04 23:08 ` [PATCH 5/5] perf, x86, uncore: Fix PCU filter setup for Sandy/Ivy/Haswell EP Andi Kleen
2014-09-19 0:44 ` [PATCH 1/5] perf, x86, uncore: Fix section mismatch in split uncore driver Peter Zijlstra
4 siblings, 1 reply; 17+ messages in thread
From: Andi Kleen @ 2014-09-04 23:08 UTC (permalink / raw)
To: peterz; +Cc: linux-kernel, mingo, eranian, tglx, Andi Kleen
From: Andi Kleen <ak@linux.intel.com>
The IvyBridge-EP uncore driver was missing three filter flags:
NC, ISOC, C6 which are useful in some cases. Support them in the same way
as the Haswell EP driver, by allowing to set them and exposing
them in the sysfs formats.
Also fix a typo in a define.
Relies on the Haswell EP driver to be applied earlier.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
---
arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c | 11 +++++++++--
1 file changed, 9 insertions(+), 2 deletions(-)
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c b/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
index a4fe9fa..b237abc2 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
@@ -139,7 +139,7 @@
#define IVBEP_CB0_MSR_PMON_BOX_FILTER_OPC (0x1ffULL << 52)
#define IVBEP_CB0_MSR_PMON_BOX_FILTER_C6 (0x1ULL << 61)
#define IVBEP_CB0_MSR_PMON_BOX_FILTER_NC (0x1ULL << 62)
-#define IVBEP_CB0_MSR_PMON_BOX_FILTER_IOSC (0x1ULL << 63)
+#define IVBEP_CB0_MSR_PMON_BOX_FILTER_ISOC (0x1ULL << 63)
/* IVBEP home agent */
#define IVBEP_HA_PCI_PMON_CTL_Q_OCC_RST (1 << 16)
@@ -1214,6 +1214,9 @@ static struct attribute *ivbep_uncore_cbox_formats_attr[] = {
&format_attr_filter_state2.attr,
&format_attr_filter_nid2.attr,
&format_attr_filter_opc2.attr,
+ &format_attr_filter_nc.attr,
+ &format_attr_filter_c6.attr,
+ &format_attr_filter_isoc.attr,
NULL,
};
@@ -1350,8 +1353,12 @@ static u64 ivbep_cbox_filter_mask(int fields)
mask |= IVBEP_CB0_MSR_PMON_BOX_FILTER_STATE;
if (fields & 0x8)
mask |= IVBEP_CB0_MSR_PMON_BOX_FILTER_NID;
- if (fields & 0x10)
+ if (fields & 0x10) {
mask |= IVBEP_CB0_MSR_PMON_BOX_FILTER_OPC;
+ mask |= IVBEP_CB0_MSR_PMON_BOX_FILTER_NC;
+ mask |= IVBEP_CB0_MSR_PMON_BOX_FILTER_C6;
+ mask |= IVBEP_CB0_MSR_PMON_BOX_FILTER_ISOC;
+ }
return mask;
}
--
1.9.3
^ permalink raw reply related [flat|nested] 17+ messages in thread
* [PATCH 5/5] perf, x86, uncore: Fix PCU filter setup for Sandy/Ivy/Haswell EP
2014-09-04 23:08 [PATCH 1/5] perf, x86, uncore: Fix section mismatch in split uncore driver Andi Kleen
` (2 preceding siblings ...)
2014-09-04 23:08 ` [PATCH 4/5] perf, x86, uncore: Add missing cbox filter flags on IvyBridge-EP uncore driver Andi Kleen
@ 2014-09-04 23:08 ` Andi Kleen
2014-09-24 15:00 ` [tip:perf/core] perf/x86/intel/uncore: " tip-bot for Andi Kleen
2014-09-19 0:44 ` [PATCH 1/5] perf, x86, uncore: Fix section mismatch in split uncore driver Peter Zijlstra
4 siblings, 1 reply; 17+ messages in thread
From: Andi Kleen @ 2014-09-04 23:08 UTC (permalink / raw)
To: peterz; +Cc: linux-kernel, mingo, eranian, tglx, Andi Kleen
From: Andi Kleen <ak@linux.intel.com>
The PCU frequency band filters use 8 bit each in a register.
When setting up the value the shift value was not correctly
scaled, which resulted in all filters except for band 0 to
be zero. Fix the scaling.
This allows to correctly monitor multiple uncore frequency bands.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
---
arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c b/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
index b237abc2..adf138e 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
@@ -841,7 +841,7 @@ static int snbep_pcu_hw_config(struct intel_uncore_box *box, struct perf_event *
if (ev_sel >= 0xb && ev_sel <= 0xe) {
reg1->reg = SNBEP_PCU_MSR_PMON_BOX_FILTER;
reg1->idx = ev_sel - 0xb;
- reg1->config = event->attr.config1 & (0xff << reg1->idx);
+ reg1->config = event->attr.config1 & (0xff << (reg1->idx * 8));
}
return 0;
}
--
1.9.3
^ permalink raw reply related [flat|nested] 17+ messages in thread
* Re: [PATCH 1/5] perf, x86, uncore: Fix section mismatch in split uncore driver
2014-09-04 23:08 [PATCH 1/5] perf, x86, uncore: Fix section mismatch in split uncore driver Andi Kleen
` (3 preceding siblings ...)
2014-09-04 23:08 ` [PATCH 5/5] perf, x86, uncore: Fix PCU filter setup for Sandy/Ivy/Haswell EP Andi Kleen
@ 2014-09-19 0:44 ` Peter Zijlstra
4 siblings, 0 replies; 17+ messages in thread
From: Peter Zijlstra @ 2014-09-19 0:44 UTC (permalink / raw)
To: Andi Kleen; +Cc: linux-kernel, mingo, eranian, tglx, Andi Kleen
Thanks Andi
^ permalink raw reply [flat|nested] 17+ messages in thread
* [tip:perf/core] perf/x86/intel/uncore: Add Haswell-EP uncore support
2014-09-04 23:08 ` [PATCH 2/5] perf, x86, uncore: Add Haswell-EP uncore support Andi Kleen
@ 2014-09-24 15:00 ` tip-bot for Yan, Zheng
2014-10-22 19:07 ` [PATCH 2/5] perf, x86, uncore: " Alexei Starovoitov
1 sibling, 0 replies; 17+ messages in thread
From: tip-bot for Yan, Zheng @ 2014-09-24 15:00 UTC (permalink / raw)
To: linux-tip-commits; +Cc: linux-kernel, hpa, mingo, peterz, zheng.z.yan, ak, tglx
Commit-ID: e735b9db12d76d45f74aee78bd63bbd2f8f480e1
Gitweb: http://git.kernel.org/tip/e735b9db12d76d45f74aee78bd63bbd2f8f480e1
Author: Yan, Zheng <zheng.z.yan@intel.com>
AuthorDate: Thu, 4 Sep 2014 16:08:26 -0700
Committer: Ingo Molnar <mingo@kernel.org>
CommitDate: Wed, 24 Sep 2014 14:48:21 +0200
perf/x86/intel/uncore: Add Haswell-EP uncore support
The uncore subsystem in Haswell-EP is similar to Sandy/Ivy
Bridge-EP. There are some differences in config register
encoding and pci device IDs. The Haswell-EP uncore also
supports a few new events. Add the Haswell-EP driver to
the snbep split driver.
Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
[ Add missing break. Add imc events. Add cbox nc/isoc/c6. ]
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: eranian@google.com
Link: http://lkml.kernel.org/r/1409872109-31645-2-git-send-email-andi@firstfloor.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
arch/x86/kernel/cpu/perf_event_intel_uncore.c | 7 +-
arch/x86/kernel/cpu/perf_event_intel_uncore.h | 2 +
.../x86/kernel/cpu/perf_event_intel_uncore_snbep.c | 608 ++++++++++++++++++++-
3 files changed, 615 insertions(+), 2 deletions(-)
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore.c b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
index 812ec5d..2b0fa02 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
@@ -883,6 +883,9 @@ static int __init uncore_pci_init(void)
case 62: /* Ivy Bridge-EP */
ret = ivbep_uncore_pci_init();
break;
+ case 63: /* Haswell-EP */
+ ret = hswep_uncore_pci_init();
+ break;
case 42: /* Sandy Bridge */
ret = snb_uncore_pci_init();
break;
@@ -1190,7 +1193,9 @@ static int __init uncore_cpu_init(void)
case 62: /* Ivy Bridge-EP */
ivbep_uncore_cpu_init();
break;
-
+ case 63: /* Haswell-EP */
+ hswep_uncore_cpu_init();
+ break;
default:
return 0;
}
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore.h b/arch/x86/kernel/cpu/perf_event_intel_uncore.h
index 1d7e894..18eb78b 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore.h
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore.h
@@ -332,6 +332,8 @@ int snbep_uncore_pci_init(void);
void snbep_uncore_cpu_init(void);
int ivbep_uncore_pci_init(void);
void ivbep_uncore_cpu_init(void);
+int hswep_uncore_pci_init(void);
+void hswep_uncore_cpu_init(void);
/* perf_event_intel_uncore_nhmex.c */
void nhmex_uncore_cpu_init(void);
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c b/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
index 6606ed0..a4fe9fa 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
@@ -163,6 +163,53 @@
#define __BITS_VALUE(x, i, n) ((typeof(x))(((x) >> ((i) * (n))) & \
((1ULL << (n)) - 1)))
+/* Haswell-EP Ubox */
+#define HSWEP_U_MSR_PMON_CTR0 0x705
+#define HSWEP_U_MSR_PMON_CTL0 0x709
+#define HSWEP_U_MSR_PMON_FILTER 0x707
+
+#define HSWEP_U_MSR_PMON_UCLK_FIXED_CTL 0x703
+#define HSWEP_U_MSR_PMON_UCLK_FIXED_CTR 0x704
+
+#define HSWEP_U_MSR_PMON_BOX_FILTER_TID (0x1 << 0)
+#define HSWEP_U_MSR_PMON_BOX_FILTER_CID (0x1fULL << 1)
+#define HSWEP_U_MSR_PMON_BOX_FILTER_MASK \
+ (HSWEP_U_MSR_PMON_BOX_FILTER_TID | \
+ HSWEP_U_MSR_PMON_BOX_FILTER_CID)
+
+/* Haswell-EP CBo */
+#define HSWEP_C0_MSR_PMON_CTR0 0xe08
+#define HSWEP_C0_MSR_PMON_CTL0 0xe01
+#define HSWEP_C0_MSR_PMON_BOX_CTL 0xe00
+#define HSWEP_C0_MSR_PMON_BOX_FILTER0 0xe05
+#define HSWEP_CBO_MSR_OFFSET 0x10
+
+
+#define HSWEP_CB0_MSR_PMON_BOX_FILTER_TID (0x3fULL << 0)
+#define HSWEP_CB0_MSR_PMON_BOX_FILTER_LINK (0xfULL << 6)
+#define HSWEP_CB0_MSR_PMON_BOX_FILTER_STATE (0x7fULL << 17)
+#define HSWEP_CB0_MSR_PMON_BOX_FILTER_NID (0xffffULL << 32)
+#define HSWEP_CB0_MSR_PMON_BOX_FILTER_OPC (0x1ffULL << 52)
+#define HSWEP_CB0_MSR_PMON_BOX_FILTER_C6 (0x1ULL << 61)
+#define HSWEP_CB0_MSR_PMON_BOX_FILTER_NC (0x1ULL << 62)
+#define HSWEP_CB0_MSR_PMON_BOX_FILTER_ISOC (0x1ULL << 63)
+
+
+/* Haswell-EP Sbox */
+#define HSWEP_S0_MSR_PMON_CTR0 0x726
+#define HSWEP_S0_MSR_PMON_CTL0 0x721
+#define HSWEP_S0_MSR_PMON_BOX_CTL 0x720
+#define HSWEP_SBOX_MSR_OFFSET 0xa
+#define HSWEP_S_MSR_PMON_RAW_EVENT_MASK (SNBEP_PMON_RAW_EVENT_MASK | \
+ SNBEP_CBO_PMON_CTL_TID_EN)
+
+/* Haswell-EP PCU */
+#define HSWEP_PCU_MSR_PMON_CTR0 0x717
+#define HSWEP_PCU_MSR_PMON_CTL0 0x711
+#define HSWEP_PCU_MSR_PMON_BOX_CTL 0x710
+#define HSWEP_PCU_MSR_PMON_BOX_FILTER 0x715
+
+
DEFINE_UNCORE_FORMAT_ATTR(event, event, "config:0-7");
DEFINE_UNCORE_FORMAT_ATTR(event_ext, event, "config:0-7,21");
DEFINE_UNCORE_FORMAT_ATTR(umask, umask, "config:8-15");
@@ -175,13 +222,21 @@ DEFINE_UNCORE_FORMAT_ATTR(occ_sel, occ_sel, "config:14-15");
DEFINE_UNCORE_FORMAT_ATTR(occ_invert, occ_invert, "config:30");
DEFINE_UNCORE_FORMAT_ATTR(occ_edge, occ_edge, "config:14-51");
DEFINE_UNCORE_FORMAT_ATTR(filter_tid, filter_tid, "config1:0-4");
+DEFINE_UNCORE_FORMAT_ATTR(filter_tid2, filter_tid, "config1:0");
+DEFINE_UNCORE_FORMAT_ATTR(filter_tid3, filter_tid, "config1:0-5");
+DEFINE_UNCORE_FORMAT_ATTR(filter_cid, filter_cid, "config1:5");
DEFINE_UNCORE_FORMAT_ATTR(filter_link, filter_link, "config1:5-8");
+DEFINE_UNCORE_FORMAT_ATTR(filter_link2, filter_link, "config1:6-8");
DEFINE_UNCORE_FORMAT_ATTR(filter_nid, filter_nid, "config1:10-17");
DEFINE_UNCORE_FORMAT_ATTR(filter_nid2, filter_nid, "config1:32-47");
DEFINE_UNCORE_FORMAT_ATTR(filter_state, filter_state, "config1:18-22");
DEFINE_UNCORE_FORMAT_ATTR(filter_state2, filter_state, "config1:17-22");
+DEFINE_UNCORE_FORMAT_ATTR(filter_state3, filter_state, "config1:17-23");
DEFINE_UNCORE_FORMAT_ATTR(filter_opc, filter_opc, "config1:23-31");
DEFINE_UNCORE_FORMAT_ATTR(filter_opc2, filter_opc, "config1:52-60");
+DEFINE_UNCORE_FORMAT_ATTR(filter_nc, filter_nc, "config1:62");
+DEFINE_UNCORE_FORMAT_ATTR(filter_c6, filter_c6, "config1:61");
+DEFINE_UNCORE_FORMAT_ATTR(filter_isoc, filter_isoc, "config1:63");
DEFINE_UNCORE_FORMAT_ATTR(filter_band0, filter_band0, "config1:0-7");
DEFINE_UNCORE_FORMAT_ATTR(filter_band1, filter_band1, "config1:8-15");
DEFINE_UNCORE_FORMAT_ATTR(filter_band2, filter_band2, "config1:16-23");
@@ -857,7 +912,6 @@ static void snbep_qpi_enable_event(struct intel_uncore_box *box, struct perf_eve
if (reg1->idx != EXTRA_REG_NONE) {
int idx = box->pmu->pmu_idx + SNBEP_PCI_QPI_PORT0_FILTER;
struct pci_dev *filter_pdev = uncore_extra_pci_dev[box->phys_id][idx];
- WARN_ON_ONCE(!filter_pdev);
if (filter_pdev) {
pci_write_config_dword(filter_pdev, reg1->reg,
(u32)reg1->config);
@@ -1643,3 +1697,555 @@ int ivbep_uncore_pci_init(void)
return 0;
}
/* end of IvyTown uncore support */
+
+/* Haswell-EP uncore support */
+static struct attribute *hswep_uncore_ubox_formats_attr[] = {
+ &format_attr_event.attr,
+ &format_attr_umask.attr,
+ &format_attr_edge.attr,
+ &format_attr_inv.attr,
+ &format_attr_thresh5.attr,
+ &format_attr_filter_tid2.attr,
+ &format_attr_filter_cid.attr,
+ NULL,
+};
+
+static struct attribute_group hswep_uncore_ubox_format_group = {
+ .name = "format",
+ .attrs = hswep_uncore_ubox_formats_attr,
+};
+
+static int hswep_ubox_hw_config(struct intel_uncore_box *box, struct perf_event *event)
+{
+ struct hw_perf_event_extra *reg1 = &event->hw.extra_reg;
+ reg1->reg = HSWEP_U_MSR_PMON_FILTER;
+ reg1->config = event->attr.config1 & HSWEP_U_MSR_PMON_BOX_FILTER_MASK;
+ reg1->idx = 0;
+ return 0;
+}
+
+static struct intel_uncore_ops hswep_uncore_ubox_ops = {
+ SNBEP_UNCORE_MSR_OPS_COMMON_INIT(),
+ .hw_config = hswep_ubox_hw_config,
+ .get_constraint = uncore_get_constraint,
+ .put_constraint = uncore_put_constraint,
+};
+
+static struct intel_uncore_type hswep_uncore_ubox = {
+ .name = "ubox",
+ .num_counters = 2,
+ .num_boxes = 1,
+ .perf_ctr_bits = 44,
+ .fixed_ctr_bits = 48,
+ .perf_ctr = HSWEP_U_MSR_PMON_CTR0,
+ .event_ctl = HSWEP_U_MSR_PMON_CTL0,
+ .event_mask = SNBEP_U_MSR_PMON_RAW_EVENT_MASK,
+ .fixed_ctr = HSWEP_U_MSR_PMON_UCLK_FIXED_CTR,
+ .fixed_ctl = HSWEP_U_MSR_PMON_UCLK_FIXED_CTL,
+ .num_shared_regs = 1,
+ .ops = &hswep_uncore_ubox_ops,
+ .format_group = &hswep_uncore_ubox_format_group,
+};
+
+static struct attribute *hswep_uncore_cbox_formats_attr[] = {
+ &format_attr_event.attr,
+ &format_attr_umask.attr,
+ &format_attr_edge.attr,
+ &format_attr_tid_en.attr,
+ &format_attr_thresh8.attr,
+ &format_attr_filter_tid3.attr,
+ &format_attr_filter_link2.attr,
+ &format_attr_filter_state3.attr,
+ &format_attr_filter_nid2.attr,
+ &format_attr_filter_opc2.attr,
+ &format_attr_filter_nc.attr,
+ &format_attr_filter_c6.attr,
+ &format_attr_filter_isoc.attr,
+ NULL,
+};
+
+static struct attribute_group hswep_uncore_cbox_format_group = {
+ .name = "format",
+ .attrs = hswep_uncore_cbox_formats_attr,
+};
+
+static struct event_constraint hswep_uncore_cbox_constraints[] = {
+ UNCORE_EVENT_CONSTRAINT(0x01, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x09, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x11, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x36, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x38, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x3b, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x3e, 0x1),
+ EVENT_CONSTRAINT_END
+};
+
+static struct extra_reg hswep_uncore_cbox_extra_regs[] = {
+ SNBEP_CBO_EVENT_EXTRA_REG(SNBEP_CBO_PMON_CTL_TID_EN,
+ SNBEP_CBO_PMON_CTL_TID_EN, 0x1),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x0334, 0xffff, 0x4),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x0534, 0xffff, 0x4),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x0934, 0xffff, 0x4),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x1134, 0xffff, 0x4),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x2134, 0xffff, 0x4),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4134, 0xffff, 0x4),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4037, 0x40ff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4028, 0x40ff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4032, 0x40ff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4029, 0x40ff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4033, 0x40ff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x402A, 0x40ff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x0135, 0xffff, 0x12),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x0335, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4135, 0xffff, 0x18),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4435, 0xffff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4835, 0xffff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x5035, 0xffff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4335, 0xffff, 0x18),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4a35, 0xffff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x2335, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x8335, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x2135, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x8135, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x0136, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x0336, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4136, 0xffff, 0x18),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4436, 0xffff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4836, 0xffff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4336, 0xffff, 0x18),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x4a36, 0xffff, 0x8),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x2336, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x8336, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x2136, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x8136, 0xffff, 0x10),
+ SNBEP_CBO_EVENT_EXTRA_REG(0x5036, 0xffff, 0x8),
+ EVENT_EXTRA_END
+};
+
+static u64 hswep_cbox_filter_mask(int fields)
+{
+ u64 mask = 0;
+ if (fields & 0x1)
+ mask |= HSWEP_CB0_MSR_PMON_BOX_FILTER_TID;
+ if (fields & 0x2)
+ mask |= HSWEP_CB0_MSR_PMON_BOX_FILTER_LINK;
+ if (fields & 0x4)
+ mask |= HSWEP_CB0_MSR_PMON_BOX_FILTER_STATE;
+ if (fields & 0x8)
+ mask |= HSWEP_CB0_MSR_PMON_BOX_FILTER_NID;
+ if (fields & 0x10) {
+ mask |= HSWEP_CB0_MSR_PMON_BOX_FILTER_OPC;
+ mask |= HSWEP_CB0_MSR_PMON_BOX_FILTER_NC;
+ mask |= HSWEP_CB0_MSR_PMON_BOX_FILTER_C6;
+ mask |= HSWEP_CB0_MSR_PMON_BOX_FILTER_ISOC;
+ }
+ return mask;
+}
+
+static struct event_constraint *
+hswep_cbox_get_constraint(struct intel_uncore_box *box, struct perf_event *event)
+{
+ return __snbep_cbox_get_constraint(box, event, hswep_cbox_filter_mask);
+}
+
+static int hswep_cbox_hw_config(struct intel_uncore_box *box, struct perf_event *event)
+{
+ struct hw_perf_event_extra *reg1 = &event->hw.extra_reg;
+ struct extra_reg *er;
+ int idx = 0;
+
+ for (er = hswep_uncore_cbox_extra_regs; er->msr; er++) {
+ if (er->event != (event->hw.config & er->config_mask))
+ continue;
+ idx |= er->idx;
+ }
+
+ if (idx) {
+ reg1->reg = HSWEP_C0_MSR_PMON_BOX_FILTER0 +
+ HSWEP_CBO_MSR_OFFSET * box->pmu->pmu_idx;
+ reg1->config = event->attr.config1 & hswep_cbox_filter_mask(idx);
+ reg1->idx = idx;
+ }
+ return 0;
+}
+
+static void hswep_cbox_enable_event(struct intel_uncore_box *box,
+ struct perf_event *event)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ struct hw_perf_event_extra *reg1 = &hwc->extra_reg;
+
+ if (reg1->idx != EXTRA_REG_NONE) {
+ u64 filter = uncore_shared_reg_config(box, 0);
+ wrmsrl(reg1->reg, filter & 0xffffffff);
+ wrmsrl(reg1->reg + 1, filter >> 32);
+ }
+
+ wrmsrl(hwc->config_base, hwc->config | SNBEP_PMON_CTL_EN);
+}
+
+static struct intel_uncore_ops hswep_uncore_cbox_ops = {
+ .init_box = snbep_uncore_msr_init_box,
+ .disable_box = snbep_uncore_msr_disable_box,
+ .enable_box = snbep_uncore_msr_enable_box,
+ .disable_event = snbep_uncore_msr_disable_event,
+ .enable_event = hswep_cbox_enable_event,
+ .read_counter = uncore_msr_read_counter,
+ .hw_config = hswep_cbox_hw_config,
+ .get_constraint = hswep_cbox_get_constraint,
+ .put_constraint = snbep_cbox_put_constraint,
+};
+
+static struct intel_uncore_type hswep_uncore_cbox = {
+ .name = "cbox",
+ .num_counters = 4,
+ .num_boxes = 18,
+ .perf_ctr_bits = 44,
+ .event_ctl = HSWEP_C0_MSR_PMON_CTL0,
+ .perf_ctr = HSWEP_C0_MSR_PMON_CTR0,
+ .event_mask = SNBEP_CBO_MSR_PMON_RAW_EVENT_MASK,
+ .box_ctl = HSWEP_C0_MSR_PMON_BOX_CTL,
+ .msr_offset = HSWEP_CBO_MSR_OFFSET,
+ .num_shared_regs = 1,
+ .constraints = hswep_uncore_cbox_constraints,
+ .ops = &hswep_uncore_cbox_ops,
+ .format_group = &hswep_uncore_cbox_format_group,
+};
+
+static struct attribute *hswep_uncore_sbox_formats_attr[] = {
+ &format_attr_event.attr,
+ &format_attr_umask.attr,
+ &format_attr_edge.attr,
+ &format_attr_tid_en.attr,
+ &format_attr_inv.attr,
+ &format_attr_thresh8.attr,
+ NULL,
+};
+
+static struct attribute_group hswep_uncore_sbox_format_group = {
+ .name = "format",
+ .attrs = hswep_uncore_sbox_formats_attr,
+};
+
+static struct intel_uncore_type hswep_uncore_sbox = {
+ .name = "sbox",
+ .num_counters = 4,
+ .num_boxes = 4,
+ .perf_ctr_bits = 44,
+ .event_ctl = HSWEP_S0_MSR_PMON_CTL0,
+ .perf_ctr = HSWEP_S0_MSR_PMON_CTR0,
+ .event_mask = HSWEP_S_MSR_PMON_RAW_EVENT_MASK,
+ .box_ctl = HSWEP_S0_MSR_PMON_BOX_CTL,
+ .msr_offset = HSWEP_SBOX_MSR_OFFSET,
+ .ops = &snbep_uncore_msr_ops,
+ .format_group = &hswep_uncore_sbox_format_group,
+};
+
+static int hswep_pcu_hw_config(struct intel_uncore_box *box, struct perf_event *event)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ struct hw_perf_event_extra *reg1 = &hwc->extra_reg;
+ int ev_sel = hwc->config & SNBEP_PMON_CTL_EV_SEL_MASK;
+
+ if (ev_sel >= 0xb && ev_sel <= 0xe) {
+ reg1->reg = HSWEP_PCU_MSR_PMON_BOX_FILTER;
+ reg1->idx = ev_sel - 0xb;
+ reg1->config = event->attr.config1 & (0xff << reg1->idx);
+ }
+ return 0;
+}
+
+static struct intel_uncore_ops hswep_uncore_pcu_ops = {
+ SNBEP_UNCORE_MSR_OPS_COMMON_INIT(),
+ .hw_config = hswep_pcu_hw_config,
+ .get_constraint = snbep_pcu_get_constraint,
+ .put_constraint = snbep_pcu_put_constraint,
+};
+
+static struct intel_uncore_type hswep_uncore_pcu = {
+ .name = "pcu",
+ .num_counters = 4,
+ .num_boxes = 1,
+ .perf_ctr_bits = 48,
+ .perf_ctr = HSWEP_PCU_MSR_PMON_CTR0,
+ .event_ctl = HSWEP_PCU_MSR_PMON_CTL0,
+ .event_mask = SNBEP_PCU_MSR_PMON_RAW_EVENT_MASK,
+ .box_ctl = HSWEP_PCU_MSR_PMON_BOX_CTL,
+ .num_shared_regs = 1,
+ .ops = &hswep_uncore_pcu_ops,
+ .format_group = &snbep_uncore_pcu_format_group,
+};
+
+static struct intel_uncore_type *hswep_msr_uncores[] = {
+ &hswep_uncore_ubox,
+ &hswep_uncore_cbox,
+ &hswep_uncore_sbox,
+ &hswep_uncore_pcu,
+ NULL,
+};
+
+void hswep_uncore_cpu_init(void)
+{
+ if (hswep_uncore_cbox.num_boxes > boot_cpu_data.x86_max_cores)
+ hswep_uncore_cbox.num_boxes = boot_cpu_data.x86_max_cores;
+ uncore_msr_uncores = hswep_msr_uncores;
+}
+
+static struct intel_uncore_type hswep_uncore_ha = {
+ .name = "ha",
+ .num_counters = 5,
+ .num_boxes = 2,
+ .perf_ctr_bits = 48,
+ SNBEP_UNCORE_PCI_COMMON_INIT(),
+};
+
+static struct uncore_event_desc hswep_uncore_imc_events[] = {
+ INTEL_UNCORE_EVENT_DESC(clockticks, "event=0x00,umask=0x00"),
+ INTEL_UNCORE_EVENT_DESC(cas_count_read, "event=0x04,umask=0x03"),
+ INTEL_UNCORE_EVENT_DESC(cas_count_write, "event=0x04,umask=0x0c"),
+ { /* end: all zeroes */ },
+};
+
+static struct intel_uncore_type hswep_uncore_imc = {
+ .name = "imc",
+ .num_counters = 5,
+ .num_boxes = 8,
+ .perf_ctr_bits = 48,
+ .fixed_ctr_bits = 48,
+ .fixed_ctr = SNBEP_MC_CHy_PCI_PMON_FIXED_CTR,
+ .fixed_ctl = SNBEP_MC_CHy_PCI_PMON_FIXED_CTL,
+ .event_descs = hswep_uncore_imc_events,
+ SNBEP_UNCORE_PCI_COMMON_INIT(),
+};
+
+static struct intel_uncore_ops hswep_uncore_irp_ops = {
+ .init_box = snbep_uncore_pci_init_box,
+ .disable_box = snbep_uncore_pci_disable_box,
+ .enable_box = snbep_uncore_pci_enable_box,
+ .disable_event = ivbep_uncore_irp_disable_event,
+ .enable_event = ivbep_uncore_irp_enable_event,
+ .read_counter = ivbep_uncore_irp_read_counter,
+};
+
+static struct intel_uncore_type hswep_uncore_irp = {
+ .name = "irp",
+ .num_counters = 4,
+ .num_boxes = 1,
+ .perf_ctr_bits = 48,
+ .event_mask = SNBEP_PMON_RAW_EVENT_MASK,
+ .box_ctl = SNBEP_PCI_PMON_BOX_CTL,
+ .ops = &hswep_uncore_irp_ops,
+ .format_group = &snbep_uncore_format_group,
+};
+
+static struct intel_uncore_type hswep_uncore_qpi = {
+ .name = "qpi",
+ .num_counters = 5,
+ .num_boxes = 3,
+ .perf_ctr_bits = 48,
+ .perf_ctr = SNBEP_PCI_PMON_CTR0,
+ .event_ctl = SNBEP_PCI_PMON_CTL0,
+ .event_mask = SNBEP_QPI_PCI_PMON_RAW_EVENT_MASK,
+ .box_ctl = SNBEP_PCI_PMON_BOX_CTL,
+ .num_shared_regs = 1,
+ .ops = &snbep_uncore_qpi_ops,
+ .format_group = &snbep_uncore_qpi_format_group,
+};
+
+static struct event_constraint hswep_uncore_r2pcie_constraints[] = {
+ UNCORE_EVENT_CONSTRAINT(0x10, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x11, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x13, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x23, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x24, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x25, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x26, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x27, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x28, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x29, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x2a, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x2b, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x2c, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x2d, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x32, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x33, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x34, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x35, 0x3),
+ EVENT_CONSTRAINT_END
+};
+
+static struct intel_uncore_type hswep_uncore_r2pcie = {
+ .name = "r2pcie",
+ .num_counters = 4,
+ .num_boxes = 1,
+ .perf_ctr_bits = 48,
+ .constraints = hswep_uncore_r2pcie_constraints,
+ SNBEP_UNCORE_PCI_COMMON_INIT(),
+};
+
+static struct event_constraint hswep_uncore_r3qpi_constraints[] = {
+ UNCORE_EVENT_CONSTRAINT(0x01, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x07, 0x7),
+ UNCORE_EVENT_CONSTRAINT(0x08, 0x7),
+ UNCORE_EVENT_CONSTRAINT(0x09, 0x7),
+ UNCORE_EVENT_CONSTRAINT(0x0a, 0x7),
+ UNCORE_EVENT_CONSTRAINT(0x0e, 0x7),
+ UNCORE_EVENT_CONSTRAINT(0x10, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x11, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x12, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x13, 0x1),
+ UNCORE_EVENT_CONSTRAINT(0x14, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x15, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x1f, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x20, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x21, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x22, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x23, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x25, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x26, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x28, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x29, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x2c, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x2d, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x2e, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x2f, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x31, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x32, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x33, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x34, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x36, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x37, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x38, 0x3),
+ UNCORE_EVENT_CONSTRAINT(0x39, 0x3),
+ EVENT_CONSTRAINT_END
+};
+
+static struct intel_uncore_type hswep_uncore_r3qpi = {
+ .name = "r3qpi",
+ .num_counters = 4,
+ .num_boxes = 3,
+ .perf_ctr_bits = 44,
+ .constraints = hswep_uncore_r3qpi_constraints,
+ SNBEP_UNCORE_PCI_COMMON_INIT(),
+};
+
+enum {
+ HSWEP_PCI_UNCORE_HA,
+ HSWEP_PCI_UNCORE_IMC,
+ HSWEP_PCI_UNCORE_IRP,
+ HSWEP_PCI_UNCORE_QPI,
+ HSWEP_PCI_UNCORE_R2PCIE,
+ HSWEP_PCI_UNCORE_R3QPI,
+};
+
+static struct intel_uncore_type *hswep_pci_uncores[] = {
+ [HSWEP_PCI_UNCORE_HA] = &hswep_uncore_ha,
+ [HSWEP_PCI_UNCORE_IMC] = &hswep_uncore_imc,
+ [HSWEP_PCI_UNCORE_IRP] = &hswep_uncore_irp,
+ [HSWEP_PCI_UNCORE_QPI] = &hswep_uncore_qpi,
+ [HSWEP_PCI_UNCORE_R2PCIE] = &hswep_uncore_r2pcie,
+ [HSWEP_PCI_UNCORE_R3QPI] = &hswep_uncore_r3qpi,
+ NULL,
+};
+
+static DEFINE_PCI_DEVICE_TABLE(hswep_uncore_pci_ids) = {
+ { /* Home Agent 0 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f30),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_HA, 0),
+ },
+ { /* Home Agent 1 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f38),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_HA, 1),
+ },
+ { /* MC0 Channel 0 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2fb0),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IMC, 0),
+ },
+ { /* MC0 Channel 1 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2fb1),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IMC, 1),
+ },
+ { /* MC0 Channel 2 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2fb4),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IMC, 2),
+ },
+ { /* MC0 Channel 3 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2fb5),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IMC, 3),
+ },
+ { /* MC1 Channel 0 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2fd0),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IMC, 4),
+ },
+ { /* MC1 Channel 1 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2fd1),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IMC, 5),
+ },
+ { /* MC1 Channel 2 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2fd4),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IMC, 6),
+ },
+ { /* MC1 Channel 3 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2fd5),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IMC, 7),
+ },
+ { /* IRP */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f39),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_IRP, 0),
+ },
+ { /* QPI0 Port 0 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f32),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_QPI, 0),
+ },
+ { /* QPI0 Port 1 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f33),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_QPI, 1),
+ },
+ { /* QPI1 Port 2 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f3a),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_QPI, 2),
+ },
+ { /* R2PCIe */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f34),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_R2PCIE, 0),
+ },
+ { /* R3QPI0 Link 0 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f36),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_R3QPI, 0),
+ },
+ { /* R3QPI0 Link 1 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f37),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_R3QPI, 1),
+ },
+ { /* R3QPI1 Link 2 */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f3e),
+ .driver_data = UNCORE_PCI_DEV_DATA(HSWEP_PCI_UNCORE_R3QPI, 2),
+ },
+ { /* QPI Port 0 filter */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f86),
+ .driver_data = UNCORE_PCI_DEV_DATA(UNCORE_EXTRA_PCI_DEV,
+ SNBEP_PCI_QPI_PORT0_FILTER),
+ },
+ { /* QPI Port 1 filter */
+ PCI_DEVICE(PCI_VENDOR_ID_INTEL, 0x2f96),
+ .driver_data = UNCORE_PCI_DEV_DATA(UNCORE_EXTRA_PCI_DEV,
+ SNBEP_PCI_QPI_PORT1_FILTER),
+ },
+ { /* end: all zeroes */ }
+};
+
+static struct pci_driver hswep_uncore_pci_driver = {
+ .name = "hswep_uncore",
+ .id_table = hswep_uncore_pci_ids,
+};
+
+int hswep_uncore_pci_init(void)
+{
+ int ret = snbep_pci2phy_map_init(0x2f1e);
+ if (ret)
+ return ret;
+ uncore_pci_uncores = hswep_pci_uncores;
+ uncore_pci_driver = &hswep_uncore_pci_driver;
+ return 0;
+}
+/* end of Haswell-EP uncore support */
^ permalink raw reply related [flat|nested] 17+ messages in thread
* [tip:perf/core] perf/x86/intel/uncore: Register the PMU only if the uncore pci device exists
2014-09-04 23:08 ` [PATCH 3/5] perf, x86, uncore: Register the PMU only if the uncore pci device exists Andi Kleen
@ 2014-09-24 15:00 ` tip-bot for Yan, Zheng
0 siblings, 0 replies; 17+ messages in thread
From: tip-bot for Yan, Zheng @ 2014-09-24 15:00 UTC (permalink / raw)
To: linux-tip-commits; +Cc: linux-kernel, hpa, mingo, peterz, zheng.z.yan, ak, tglx
Commit-ID: 513d793e5fe138d779ad6c532cd20c289a6897fc
Gitweb: http://git.kernel.org/tip/513d793e5fe138d779ad6c532cd20c289a6897fc
Author: Yan, Zheng <zheng.z.yan@intel.com>
AuthorDate: Thu, 4 Sep 2014 16:08:27 -0700
Committer: Ingo Molnar <mingo@kernel.org>
CommitDate: Wed, 24 Sep 2014 14:48:22 +0200
perf/x86/intel/uncore: Register the PMU only if the uncore pci device exists
Current code registers PMUs for all possible uncore pci devices.
This is not good because, on some machines, one or more uncore pci
devices can be missing. The missing pci device make corresponding
PMU unusable. Register the PMU only if the uncore device exists.
Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: eranian@google.com
Link: http://lkml.kernel.org/r/1409872109-31645-3-git-send-email-andi@firstfloor.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
arch/x86/kernel/cpu/perf_event_intel_uncore.c | 19 +++++++++++--------
1 file changed, 11 insertions(+), 8 deletions(-)
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore.c b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
index 2b0fa02..42d00e5 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
@@ -793,6 +793,7 @@ static int uncore_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id
struct intel_uncore_box *box;
struct intel_uncore_type *type;
int phys_id;
+ bool first_box = false;
phys_id = uncore_pcibus_to_physid[pdev->bus->number];
if (phys_id < 0)
@@ -827,9 +828,13 @@ static int uncore_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id
pci_set_drvdata(pdev, box);
raw_spin_lock(&uncore_box_lock);
+ if (list_empty(&pmu->box_list))
+ first_box = true;
list_add_tail(&box->list, &pmu->box_list);
raw_spin_unlock(&uncore_box_lock);
+ if (first_box)
+ uncore_pmu_register(pmu);
return 0;
}
@@ -838,6 +843,7 @@ static void uncore_pci_remove(struct pci_dev *pdev)
struct intel_uncore_box *box = pci_get_drvdata(pdev);
struct intel_uncore_pmu *pmu;
int i, cpu, phys_id = uncore_pcibus_to_physid[pdev->bus->number];
+ bool last_box = false;
box = pci_get_drvdata(pdev);
if (!box) {
@@ -859,6 +865,8 @@ static void uncore_pci_remove(struct pci_dev *pdev)
raw_spin_lock(&uncore_box_lock);
list_del(&box->list);
+ if (list_empty(&pmu->box_list))
+ last_box = true;
raw_spin_unlock(&uncore_box_lock);
for_each_possible_cpu(cpu) {
@@ -870,6 +878,9 @@ static void uncore_pci_remove(struct pci_dev *pdev)
WARN_ON_ONCE(atomic_read(&box->refcnt) != 1);
kfree(box);
+
+ if (last_box)
+ perf_pmu_unregister(&pmu->pmu);
}
static int __init uncore_pci_init(void)
@@ -1221,14 +1232,6 @@ static int __init uncore_pmus_register(void)
}
}
- for (i = 0; uncore_pci_uncores[i]; i++) {
- type = uncore_pci_uncores[i];
- for (j = 0; j < type->num_boxes; j++) {
- pmu = &type->pmus[j];
- uncore_pmu_register(pmu);
- }
- }
-
return 0;
}
^ permalink raw reply related [flat|nested] 17+ messages in thread
* [tip:perf/core] perf/x86/intel/uncore: Add missing cbox filter flags on IvyBridge-EP uncore driver
2014-09-04 23:08 ` [PATCH 4/5] perf, x86, uncore: Add missing cbox filter flags on IvyBridge-EP uncore driver Andi Kleen
@ 2014-09-24 15:00 ` tip-bot for Andi Kleen
0 siblings, 0 replies; 17+ messages in thread
From: tip-bot for Andi Kleen @ 2014-09-24 15:00 UTC (permalink / raw)
To: linux-tip-commits; +Cc: linux-kernel, hpa, mingo, peterz, ak, tglx
Commit-ID: 7e96ae1a8915e7064091ab75857f8c63334ea28d
Gitweb: http://git.kernel.org/tip/7e96ae1a8915e7064091ab75857f8c63334ea28d
Author: Andi Kleen <ak@linux.intel.com>
AuthorDate: Thu, 4 Sep 2014 16:08:28 -0700
Committer: Ingo Molnar <mingo@kernel.org>
CommitDate: Wed, 24 Sep 2014 14:48:23 +0200
perf/x86/intel/uncore: Add missing cbox filter flags on IvyBridge-EP uncore driver
The IvyBridge-EP uncore driver was missing three filter flags:
NC, ISOC, C6 which are useful in some cases. Support them in the same way
as the Haswell EP driver, by allowing to set them and exposing
them in the sysfs formats.
Also fix a typo in a define.
Relies on the Haswell EP driver to be applied earlier.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1409872109-31645-4-git-send-email-andi@firstfloor.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c | 11 +++++++++--
1 file changed, 9 insertions(+), 2 deletions(-)
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c b/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
index a4fe9fa..b237abc 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
@@ -139,7 +139,7 @@
#define IVBEP_CB0_MSR_PMON_BOX_FILTER_OPC (0x1ffULL << 52)
#define IVBEP_CB0_MSR_PMON_BOX_FILTER_C6 (0x1ULL << 61)
#define IVBEP_CB0_MSR_PMON_BOX_FILTER_NC (0x1ULL << 62)
-#define IVBEP_CB0_MSR_PMON_BOX_FILTER_IOSC (0x1ULL << 63)
+#define IVBEP_CB0_MSR_PMON_BOX_FILTER_ISOC (0x1ULL << 63)
/* IVBEP home agent */
#define IVBEP_HA_PCI_PMON_CTL_Q_OCC_RST (1 << 16)
@@ -1214,6 +1214,9 @@ static struct attribute *ivbep_uncore_cbox_formats_attr[] = {
&format_attr_filter_state2.attr,
&format_attr_filter_nid2.attr,
&format_attr_filter_opc2.attr,
+ &format_attr_filter_nc.attr,
+ &format_attr_filter_c6.attr,
+ &format_attr_filter_isoc.attr,
NULL,
};
@@ -1350,8 +1353,12 @@ static u64 ivbep_cbox_filter_mask(int fields)
mask |= IVBEP_CB0_MSR_PMON_BOX_FILTER_STATE;
if (fields & 0x8)
mask |= IVBEP_CB0_MSR_PMON_BOX_FILTER_NID;
- if (fields & 0x10)
+ if (fields & 0x10) {
mask |= IVBEP_CB0_MSR_PMON_BOX_FILTER_OPC;
+ mask |= IVBEP_CB0_MSR_PMON_BOX_FILTER_NC;
+ mask |= IVBEP_CB0_MSR_PMON_BOX_FILTER_C6;
+ mask |= IVBEP_CB0_MSR_PMON_BOX_FILTER_ISOC;
+ }
return mask;
}
^ permalink raw reply related [flat|nested] 17+ messages in thread
* [tip:perf/core] perf/x86/intel/uncore: Fix PCU filter setup for Sandy/Ivy/Haswell EP
2014-09-04 23:08 ` [PATCH 5/5] perf, x86, uncore: Fix PCU filter setup for Sandy/Ivy/Haswell EP Andi Kleen
@ 2014-09-24 15:00 ` tip-bot for Andi Kleen
0 siblings, 0 replies; 17+ messages in thread
From: tip-bot for Andi Kleen @ 2014-09-24 15:00 UTC (permalink / raw)
To: linux-tip-commits; +Cc: linux-kernel, hpa, mingo, peterz, ak, tglx
Commit-ID: b10fc1c3e30c44033d1cb1d2900cc2ab06dff342
Gitweb: http://git.kernel.org/tip/b10fc1c3e30c44033d1cb1d2900cc2ab06dff342
Author: Andi Kleen <ak@linux.intel.com>
AuthorDate: Thu, 4 Sep 2014 16:08:29 -0700
Committer: Ingo Molnar <mingo@kernel.org>
CommitDate: Wed, 24 Sep 2014 14:48:24 +0200
perf/x86/intel/uncore: Fix PCU filter setup for Sandy/Ivy/Haswell EP
The PCU frequency band filters use 8 bit each in a register.
When setting up the value the shift value was not correctly
scaled, which resulted in all filters except for band 0 to
be zero. Fix the scaling.
This allows to correctly monitor multiple uncore frequency bands.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: eranian@google.com
Link: http://lkml.kernel.org/r/1409872109-31645-5-git-send-email-andi@firstfloor.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c b/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
index b237abc..adf138e 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
@@ -841,7 +841,7 @@ static int snbep_pcu_hw_config(struct intel_uncore_box *box, struct perf_event *
if (ev_sel >= 0xb && ev_sel <= 0xe) {
reg1->reg = SNBEP_PCU_MSR_PMON_BOX_FILTER;
reg1->idx = ev_sel - 0xb;
- reg1->config = event->attr.config1 & (0xff << reg1->idx);
+ reg1->config = event->attr.config1 & (0xff << (reg1->idx * 8));
}
return 0;
}
^ permalink raw reply related [flat|nested] 17+ messages in thread
* Re: [PATCH 2/5] perf, x86, uncore: Add Haswell-EP uncore support
2014-09-04 23:08 ` [PATCH 2/5] perf, x86, uncore: Add Haswell-EP uncore support Andi Kleen
2014-09-24 15:00 ` [tip:perf/core] perf/x86/intel/uncore: " tip-bot for Yan, Zheng
@ 2014-10-22 19:07 ` Alexei Starovoitov
2014-10-27 11:57 ` Andi Kleen
1 sibling, 1 reply; 17+ messages in thread
From: Alexei Starovoitov @ 2014-10-22 19:07 UTC (permalink / raw)
To: Andi Kleen
Cc: Peter Zijlstra, linux-kernel, Ingo Molnar, eranian,
Thomas Gleixner, Yan, Zheng, Andi Kleen
On Thu, Sep 4, 2014 at 4:08 PM, Andi Kleen <andi@firstfloor.org> wrote:
> From: "Yan, Zheng" <zheng.z.yan@intel.com>
>
> The uncore subsystem in Haswell-EP is similar to Sandy/Ivy
> Bridge-EP. There are some differences in config register
> encoding and pci device IDs. The Haswell-EP uncore also
> supports a few new events. Add the Haswell-EP driver to
> the snbep split driver.
>
> v2: (AK): Add missing break. Add imc events. Add cbox nc/isoc/c6.
> Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
> Signed-off-by: Andi Kleen <ak@linux.intel.com>
> ---
> arch/x86/kernel/cpu/perf_event_intel_uncore.c | 7 +-
> arch/x86/kernel/cpu/perf_event_intel_uncore.h | 2 +
> .../x86/kernel/cpu/perf_event_intel_uncore_snbep.c | 608 ++++++++++++++++++++-
> 3 files changed, 615 insertions(+), 2 deletions(-)
>
> diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore.c b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
> index 812ec5d..2b0fa02 100644
> --- a/arch/x86/kernel/cpu/perf_event_intel_uncore.c
> +++ b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
> @@ -883,6 +883,9 @@ static int __init uncore_pci_init(void)
> case 62: /* Ivy Bridge-EP */
> ret = ivbep_uncore_pci_init();
> break;
> + case 63: /* Haswell-EP */
> + ret = hswep_uncore_pci_init();
> + break;
my E5-1630 v3 crashes hard during boot in
snbep_uncore_msr_init_box()+0x3c
I don't have a setup to capture the screen so early in the boot.
So this is just heads up.
I replaced the call above with 'return -ENODEV' as a workaround.
btw, E5-16xx v3 became available just a month ago.
^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH 2/5] perf, x86, uncore: Add Haswell-EP uncore support
2014-10-22 19:07 ` [PATCH 2/5] perf, x86, uncore: " Alexei Starovoitov
@ 2014-10-27 11:57 ` Andi Kleen
0 siblings, 0 replies; 17+ messages in thread
From: Andi Kleen @ 2014-10-27 11:57 UTC (permalink / raw)
To: Alexei Starovoitov
Cc: Peter Zijlstra, linux-kernel, Ingo Molnar, eranian,
Thomas Gleixner, Yan, Zheng
Alexei Starovoitov <alexei.starovoitov@gmail.com> writes:
> my E5-1630 v3 crashes hard during boot in
> snbep_uncore_msr_init_box()+0x3c
> I don't have a setup to capture the screen so early in the boot.
> So this is just heads up.
Is it a general general protection fault or a page fault?
If it's a page fault please post the disassembled code:
gdb vmlinux
disassemble snbep_uncore_msr_init_box
> I replaced the call above with 'return -ENODEV' as a workaround.
> btw, E5-16xx v3 became available just a month ago.
-Andi
--
ak@linux.intel.com -- Speaking for myself only
^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH 2/5] perf, x86, uncore: Add Haswell-EP uncore support
2014-10-29 3:11 ` Andi Kleen
@ 2014-10-29 14:49 ` Patrick Lu
0 siblings, 0 replies; 17+ messages in thread
From: Patrick Lu @ 2014-10-29 14:49 UTC (permalink / raw)
To: Andi Kleen
Cc: Alexei Starovoitov, Peter Zijlstra, linux-kernel, Ingo Molnar,
Stephane Eranian, Thomas Gleixner, Yan, Zheng
On Wed, Oct 29, 2014 at 04:11:15AM +0100, Andi Kleen wrote:
> >
> > yes. removing line 'hswep_uncore_sbox' also worked as workaround.
> > You have an idea what's going on?
>
> Not yet. I don't see it on my own systems, so it's something that's
> different on yours.
Hi Andi, I am experiencing the same crash using E5-2658v3 CPU in our lab in Chandler. I can
provide remote access for you to take a look.
Also, I used your ucevents.py to check out IRP counter, and it looks like
it's reporting half of what I'm expecting. Do you know if the script/PMU
program for both IRP0 and IRP1?
Patrick
>
> > I suspect other folks will see the same crash when E5 v3
> > become more widespread.
>
> I guess we can disable the sbox for now.
>
> -Andi
> >
>
> --
> ak@linux.intel.com -- Speaking for myself only.
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH 2/5] perf, x86, uncore: Add Haswell-EP uncore support
2014-10-28 23:15 Alexei Starovoitov
@ 2014-10-29 3:11 ` Andi Kleen
2014-10-29 14:49 ` Patrick Lu
0 siblings, 1 reply; 17+ messages in thread
From: Andi Kleen @ 2014-10-29 3:11 UTC (permalink / raw)
To: Alexei Starovoitov
Cc: Andi Kleen, Peter Zijlstra, linux-kernel, Ingo Molnar,
Stephane Eranian, Thomas Gleixner, Yan, Zheng
>
> yes. removing line 'hswep_uncore_sbox' also worked as workaround.
> You have an idea what's going on?
Not yet. I don't see it on my own systems, so it's something that's
different on yours.
> I suspect other folks will see the same crash when E5 v3
> become more widespread.
I guess we can disable the sbox for now.
-Andi
>
--
ak@linux.intel.com -- Speaking for myself only.
^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH 2/5] perf, x86, uncore: Add Haswell-EP uncore support
@ 2014-10-28 23:15 Alexei Starovoitov
2014-10-29 3:11 ` Andi Kleen
0 siblings, 1 reply; 17+ messages in thread
From: Alexei Starovoitov @ 2014-10-28 23:15 UTC (permalink / raw)
To: Andi Kleen
Cc: Peter Zijlstra, linux-kernel, Ingo Molnar, Stephane Eranian,
Thomas Gleixner, Yan, Zheng
On Tue, Oct 28, 2014 at 8:30 AM, Andi Kleen <andi@firstfloor.org> wrote:
>> snbep_uncore_msr_init_box() is called with:
>> [ 1.055678] msr 0
>> [ 1.056027] msr 3584
>> [ 1.056376] msr 3600
>> [ 1.056725] msr 3616
>> [ 1.057074] msr 3632
>> [ 1.057424] msr 1824
>
> So it would normally crash here, right? That's the SBOX.
>
> When you remove the sbox in hswep_msr_uncores the crash goes away,
> correct?
yes. removing line 'hswep_uncore_sbox' also worked as workaround.
You have an idea what's going on?
I suspect other folks will see the same crash when E5 v3
become more widespread.
^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH 2/5] perf, x86, uncore: Add Haswell-EP uncore support
2014-10-27 20:13 [PATCH 2/5] perf, x86, uncore: Add Haswell-EP uncore support Alexei Starovoitov
@ 2014-10-28 15:30 ` Andi Kleen
0 siblings, 0 replies; 17+ messages in thread
From: Andi Kleen @ 2014-10-28 15:30 UTC (permalink / raw)
To: Alexei Starovoitov
Cc: Andi Kleen, Peter Zijlstra, linux-kernel, Ingo Molnar,
Stephane Eranian, Thomas Gleixner, Yan, Zheng
> snbep_uncore_msr_init_box() is called with:
> [ 1.055678] msr 0
> [ 1.056027] msr 3584
> [ 1.056376] msr 3600
> [ 1.056725] msr 3616
> [ 1.057074] msr 3632
> [ 1.057424] msr 1824
So it would normally crash here, right? That's the SBOX.
When you remove the sbox in hswep_msr_uncores the crash goes away,
correct?
-Andi
^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH 2/5] perf, x86, uncore: Add Haswell-EP uncore support
@ 2014-10-27 20:13 Alexei Starovoitov
2014-10-28 15:30 ` Andi Kleen
0 siblings, 1 reply; 17+ messages in thread
From: Alexei Starovoitov @ 2014-10-27 20:13 UTC (permalink / raw)
To: Andi Kleen
Cc: Peter Zijlstra, linux-kernel, Ingo Molnar, Stephane Eranian,
Thomas Gleixner, Yan, Zheng
On Mon, Oct 27, 2014 at 4:57 AM, Andi Kleen <andi@firstfloor.org> wrote:
> Alexei Starovoitov <alexei.starovoitov@gmail.com> writes:
>> my E5-1630 v3 crashes hard during boot in
>> snbep_uncore_msr_init_box()+0x3c
>
>> I don't have a setup to capture the screen so early in the boot.
>> So this is just heads up.
>
> Is it a general general protection fault or a page fault?
>
> If it's a page fault please post the disassembled code:
>
> gdb vmlinux
> disassemble snbep_uncore_msr_init_box
>
snbep_uncore_msr_init_box() is called with:
[ 1.055678] msr 0
[ 1.056027] msr 3584
[ 1.056376] msr 3600
[ 1.056725] msr 3616
[ 1.057074] msr 3632
[ 1.057424] msr 1824
[ 1.057773] msr 1834
[ 1.058123] msr 1844
[ 1.058473] msr 1854
[ 1.058822] msr 1808
since msr 3584 is HSWEP_C0_MSR_PMON_BOX_CTL
I've tried the following workaround which helped as well:
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
b/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
index adf138eac85c..1ca46523a541 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore_snbep.c
@@ -368,7 +368,7 @@ static void snbep_uncore_msr_init_box(struct
intel_uncore_box *box)
{
unsigned msr = uncore_msr_box_ctl(box);
- if (msr)
+ if (msr && msr >= 3584)
wrmsrl(msr, SNBEP_PMON_BOX_CTL_INT);
}
^ permalink raw reply related [flat|nested] 17+ messages in thread
end of thread, other threads:[~2014-10-29 15:06 UTC | newest]
Thread overview: 17+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2014-09-04 23:08 [PATCH 1/5] perf, x86, uncore: Fix section mismatch in split uncore driver Andi Kleen
2014-09-04 23:08 ` [PATCH 2/5] perf, x86, uncore: Add Haswell-EP uncore support Andi Kleen
2014-09-24 15:00 ` [tip:perf/core] perf/x86/intel/uncore: " tip-bot for Yan, Zheng
2014-10-22 19:07 ` [PATCH 2/5] perf, x86, uncore: " Alexei Starovoitov
2014-10-27 11:57 ` Andi Kleen
2014-09-04 23:08 ` [PATCH 3/5] perf, x86, uncore: Register the PMU only if the uncore pci device exists Andi Kleen
2014-09-24 15:00 ` [tip:perf/core] perf/x86/intel/uncore: " tip-bot for Yan, Zheng
2014-09-04 23:08 ` [PATCH 4/5] perf, x86, uncore: Add missing cbox filter flags on IvyBridge-EP uncore driver Andi Kleen
2014-09-24 15:00 ` [tip:perf/core] perf/x86/intel/uncore: " tip-bot for Andi Kleen
2014-09-04 23:08 ` [PATCH 5/5] perf, x86, uncore: Fix PCU filter setup for Sandy/Ivy/Haswell EP Andi Kleen
2014-09-24 15:00 ` [tip:perf/core] perf/x86/intel/uncore: " tip-bot for Andi Kleen
2014-09-19 0:44 ` [PATCH 1/5] perf, x86, uncore: Fix section mismatch in split uncore driver Peter Zijlstra
2014-10-27 20:13 [PATCH 2/5] perf, x86, uncore: Add Haswell-EP uncore support Alexei Starovoitov
2014-10-28 15:30 ` Andi Kleen
2014-10-28 23:15 Alexei Starovoitov
2014-10-29 3:11 ` Andi Kleen
2014-10-29 14:49 ` Patrick Lu
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).