CXL rev 3.0 introduces a standard performance monitoring hardware
block to CXL. Instances are discovered using CXL Register Locator DVSEC
entries. Each CXL component may have multiple PMUs.
This initial driver supports a subset of types of counter.
It supports counters that are either fixed or configurable, but requires
that they support the ability to freeze and write value whilst frozen.
Development done with QEMU model which will be posted shortly.
Example:
$ perf stat -a -e cxl_pmu_mem0.0/h2d_req_snpcur/ -e cxl_pmu_mem0.0/h2d_req_snpdata/ -e cxl_pmu_mem0.0/clock_ticks/ sleep 1
Performance counter stats for 'system wide':
96,757,023,244,321 cxl_pmu_mem0.0/h2d_req_snpcur/
96,757,023,244,365 cxl_pmu_mem0.0/h2d_req_snpdata/
193,514,046,488,653 cxl_pmu_mem0.0/clock_ticks/
1.090539600 seconds time elapsed
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
v7: THanks to Namhyung and Stephane for reviews
- Set the cpu for an event only after validation checks.
- Added -a to make perf default of all CPUs explicit in command line.
- Fix wrong event names in example.
---
MAINTAINERS | 6 +
drivers/cxl/Kconfig | 13 +
drivers/perf/Kconfig | 13 +
drivers/perf/Makefile | 1 +
drivers/perf/cxl_pmu.c | 990 +++++++++++++++++++++++++++++++++++++++++
5 files changed, 1023 insertions(+)
diff --git a/MAINTAINERS b/MAINTAINERS
index e0ad886d3163..976b3e982d3a 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -5191,6 +5191,12 @@ S: Maintained
F: drivers/cxl/
F: include/uapi/linux/cxl_mem.h
+COMPUTE EXPRESS LINK PMU (CPMU)
+M: Jonathan Cameron <jonathan.cameron@huawei.com>
+L: linux-cxl@vger.kernel.org
+S: Maintained
+F: drivers/perf/cxl_pmu.c
+
CONEXANT ACCESSRUNNER USB DRIVER
L: accessrunner-general@lists.sourceforge.net
S: Orphan
diff --git a/drivers/cxl/Kconfig b/drivers/cxl/Kconfig
index ff4e78117b31..d0a4b4cfa50c 100644
--- a/drivers/cxl/Kconfig
+++ b/drivers/cxl/Kconfig
@@ -139,4 +139,17 @@ config CXL_REGION_INVALIDATION_TEST
If unsure, or if this kernel is meant for production environments,
say N.
+config CXL_PMU
+ tristate "CXL Performance Monitoring Unit"
+ default CXL_BUS
+ depends on PERF_EVENTS
+ help
+ Support performance monitoring as defined in CXL rev 3.0
+ section 13.2: Performance Monitoring. CXL components may have
+ one or more CXL Performance Monitoring Units (CPMUs).
+
+ Say 'y/m' to enable a driver that will attach to performance
+ monitoring units and provide standard perf based interfaces.
+
+ If unsure say 'm'.
endif
diff --git a/drivers/perf/Kconfig b/drivers/perf/Kconfig
index 711f82400086..2d2dd400fed2 100644
--- a/drivers/perf/Kconfig
+++ b/drivers/perf/Kconfig
@@ -213,4 +213,17 @@ source "drivers/perf/arm_cspmu/Kconfig"
source "drivers/perf/amlogic/Kconfig"
+config CXL_PMU
+ tristate "CXL Performance Monitoring Unit"
+ depends on CXL_BUS
+ help
+ Support performance monitoring as defined in CXL rev 3.0
+ section 13.2: Performance Monitoring. CXL components may have
+ one or more CXL Performance Monitoring Units (CPMUs).
+
+ Say 'y/m' to enable a driver that will attach to performance
+ monitoring units and provide standard perf based interfaces.
+
+ If unsure say 'm'.
+
endmenu
diff --git a/drivers/perf/Makefile b/drivers/perf/Makefile
index dabc859540ce..f1d7ce9da275 100644
--- a/drivers/perf/Makefile
+++ b/drivers/perf/Makefile
@@ -24,3 +24,4 @@ obj-$(CONFIG_APPLE_M1_CPU_PMU) += apple_m1_cpu_pmu.o
obj-$(CONFIG_ALIBABA_UNCORE_DRW_PMU) += alibaba_uncore_drw_pmu.o
obj-$(CONFIG_ARM_CORESIGHT_PMU_ARCH_SYSTEM_PMU) += arm_cspmu/
obj-$(CONFIG_MESON_DDR_PMU) += amlogic/
+obj-$(CONFIG_CXL_PMU) += cxl_pmu.o
diff --git a/drivers/perf/cxl_pmu.c b/drivers/perf/cxl_pmu.c
new file mode 100644
index 000000000000..0a8f597e695b
--- /dev/null
+++ b/drivers/perf/cxl_pmu.c
@@ -0,0 +1,990 @@
+// SPDX-License-Identifier: GPL-2.0-only
+
+/*
+ * Copyright(c) 2023 Huawei
+ *
+ * The CXL 3.0 specification includes a standard Performance Monitoring Unit,
+ * called the CXL PMU, or CPMU. In order to allow a high degree of
+ * implementation flexibility the specification provides a wide range of
+ * options all of which are self describing.
+ *
+ * Details in CXL rev 3.0 section 8.2.7 CPMU Register Interface
+ */
+
+#include <linux/io-64-nonatomic-lo-hi.h>
+#include <linux/perf_event.h>
+#include <linux/bitops.h>
+#include <linux/device.h>
+#include <linux/bits.h>
+#include <linux/list.h>
+#include <linux/bug.h>
+#include <linux/pci.h>
+
+#include "../cxl/cxlpci.h"
+#include "../cxl/cxl.h"
+#include "../cxl/pmu.h"
+
+#define CXL_PMU_CAP_REG 0x0
+#define CXL_PMU_CAP_NUM_COUNTERS_MSK GENMASK_ULL(4, 0)
+#define CXL_PMU_CAP_COUNTER_WIDTH_MSK GENMASK_ULL(15, 8)
+#define CXL_PMU_CAP_NUM_EVN_CAP_REG_SUP_MSK GENMASK_ULL(24, 20)
+#define CXL_PMU_CAP_FILTERS_SUP_MSK GENMASK_ULL(39, 32)
+#define CXL_PMU_FILTER_HDM BIT(0)
+#define CXL_PMU_FILTER_CHAN_RANK_BANK BIT(1)
+#define CXL_PMU_CAP_MSI_N_MSK GENMASK_ULL(47, 44)
+#define CXL_PMU_CAP_WRITEABLE_WHEN_FROZEN BIT_ULL(48)
+#define CXL_PMU_CAP_FREEZE BIT_ULL(49)
+#define CXL_PMU_CAP_INT BIT_ULL(50)
+#define CXL_PMU_CAP_VERSION_MSK GENMASK_ULL(63, 60)
+
+#define CXL_PMU_OVERFLOW_REG 0x10
+#define CXL_PMU_FREEZE_REG 0x18
+#define CXL_PMU_EVENT_CAP_REG(n) (0x100 + 8 * (n))
+#define CXL_PMU_EVENT_CAP_SUPPORTED_EVENTS_MSK GENMASK_ULL(31, 0)
+#define CXL_PMU_EVENT_CAP_GROUP_ID_MSK GENMASK_ULL(47, 32)
+#define CXL_PMU_EVENT_CAP_VENDOR_ID_MSK GENMASK_ULL(63, 48)
+
+#define CXL_PMU_COUNTER_CFG_REG(n) (0x200 + 8 * (n))
+#define CXL_PMU_COUNTER_CFG_TYPE_MSK GENMASK_ULL(1, 0)
+#define CXL_PMU_COUNTER_CFG_TYPE_FREE_RUN 0
+#define CXL_PMU_COUNTER_CFG_TYPE_FIXED_FUN 1
+#define CXL_PMU_COUNTER_CFG_TYPE_CONFIGURABLE 2
+#define CXL_PMU_COUNTER_CFG_ENABLE BIT_ULL(8)
+#define CXL_PMU_COUNTER_CFG_INT_ON_OVRFLW BIT_ULL(9)
+#define CXL_PMU_COUNTER_CFG_FREEZE_ON_OVRFLW BIT_ULL(10)
+#define CXL_PMU_COUNTER_CFG_EDGE BIT_ULL(11)
+#define CXL_PMU_COUNTER_CFG_INVERT BIT_ULL(12)
+#define CXL_PMU_COUNTER_CFG_THRESHOLD_MSK GENMASK_ULL(23, 16)
+#define CXL_PMU_COUNTER_CFG_EVENTS_MSK GENMASK_ULL(55, 24)
+#define CXL_PMU_COUNTER_CFG_EVENT_GRP_ID_IDX_MSK GENMASK_ULL(63, 59)
+
+#define CXL_PMU_FILTER_CFG_REG(n, f) (0x400 + 4 * ((f) + (n) * 8))
+#define CXL_PMU_FILTER_CFG_VALUE_MSK GENMASK(15, 0)
+
+#define CXL_PMU_COUNTER_REG(n) (0xc00 + 8 * (n))
+
+/* CXL rev 3.0 Table 13-5 Events under CXL Vendor ID */
+#define CXL_PMU_GID_CLOCK_TICKS 0x00
+#define CXL_PMU_GID_D2H_REQ 0x0010
+#define CXL_PMU_GID_D2H_RSP 0x0011
+#define CXL_PMU_GID_H2D_REQ 0x0012
+#define CXL_PMU_GID_H2D_RSP 0x0013
+#define CXL_PMU_GID_CACHE_DATA 0x0014
+#define CXL_PMU_GID_M2S_REQ 0x0020
+#define CXL_PMU_GID_M2S_RWD 0x0021
+#define CXL_PMU_GID_M2S_BIRSP 0x0022
+#define CXL_PMU_GID_S2M_BISNP 0x0023
+#define CXL_PMU_GID_S2M_NDR 0x0024
+#define CXL_PMU_GID_S2M_DRS 0x0025
+#define CXL_PMU_GID_DDR 0x8000
+
+static int cxl_pmu_cpuhp_state_num;
+
+struct cxl_pmu_ev_cap {
+ u16 vid;
+ u16 gid;
+ u32 msk;
+ union {
+ int counter_idx; /* fixed counters */
+ int event_idx; /* configurable counters */
+ };
+ struct list_head node;
+};
+
+#define CXL_PMU_MAX_COUNTERS 64
+struct cxl_pmu_info {
+ struct pmu pmu;
+ void __iomem *base;
+ struct perf_event **hw_events;
+ struct list_head event_caps_configurable;
+ struct list_head event_caps_fixed;
+ DECLARE_BITMAP(used_counter_bm, CXL_PMU_MAX_COUNTERS);
+ DECLARE_BITMAP(conf_counter_bm, CXL_PMU_MAX_COUNTERS);
+ u16 counter_width;
+ u8 num_counters;
+ u8 num_event_capabilities;
+ int on_cpu;
+ struct hlist_node node;
+ bool filter_hdm;
+ int irq;
+};
+
+#define pmu_to_cxl_pmu_info(_pmu) container_of(_pmu, struct cxl_pmu_info, pmu)
+
+/*
+ * All CPMU counters are discoverable via the Event Capabilities Registers.
+ * Each Event Capability register contains a a VID / GroupID.
+ * A counter may then count any combination (by summing) of events in
+ * that group which are in the Supported Events Bitmask.
+ * However, there are some complexities to the scheme.
+ * - Fixed function counters refer to an Event Capabilities register.
+ * That event capability register is not then used for Configurable
+ * counters.
+ */
+static int cxl_pmu_parse_caps(struct device *dev, struct cxl_pmu_info *info)
+{
+ unsigned long fixed_counter_event_cap_bm = 0;
+ void __iomem *base = info->base;
+ bool freeze_for_enable;
+ u64 val, eval;
+ int i;
+
+ val = readq(base + CXL_PMU_CAP_REG);
+ freeze_for_enable = FIELD_GET(CXL_PMU_CAP_WRITEABLE_WHEN_FROZEN, val) &&
+ FIELD_GET(CXL_PMU_CAP_FREEZE, val);
+ if (!freeze_for_enable) {
+ dev_err(dev, "Counters not writable while frozen\n");
+ return -ENODEV;
+ }
+
+ info->num_counters = FIELD_GET(CXL_PMU_CAP_NUM_COUNTERS_MSK, val) + 1;
+ info->counter_width = FIELD_GET(CXL_PMU_CAP_COUNTER_WIDTH_MSK, val);
+ info->num_event_capabilities = FIELD_GET(CXL_PMU_CAP_NUM_EVN_CAP_REG_SUP_MSK, val) + 1;
+
+ info->filter_hdm = FIELD_GET(CXL_PMU_CAP_FILTERS_SUP_MSK, val) & CXL_PMU_FILTER_HDM;
+ if (FIELD_GET(CXL_PMU_CAP_INT, val))
+ info->irq = FIELD_GET(CXL_PMU_CAP_MSI_N_MSK, val);
+ else
+ info->irq = -1;
+
+ /* First handle fixed function counters; note if configurable counters found */
+ for (i = 0; i < info->num_counters; i++) {
+ struct cxl_pmu_ev_cap *pmu_ev;
+ u32 events_msk;
+ u8 group_idx;
+
+ val = readq(base + CXL_PMU_COUNTER_CFG_REG(i));
+
+ if (FIELD_GET(CXL_PMU_COUNTER_CFG_TYPE_MSK, val) ==
+ CXL_PMU_COUNTER_CFG_TYPE_CONFIGURABLE) {
+ set_bit(i, info->conf_counter_bm);
+ }
+
+ if (FIELD_GET(CXL_PMU_COUNTER_CFG_TYPE_MSK, val) !=
+ CXL_PMU_COUNTER_CFG_TYPE_FIXED_FUN)
+ continue;
+
+ /* In this case we know which fields are const */
+ group_idx = FIELD_GET(CXL_PMU_COUNTER_CFG_EVENT_GRP_ID_IDX_MSK, val);
+ events_msk = FIELD_GET(CXL_PMU_COUNTER_CFG_EVENTS_MSK, val);
+ eval = readq(base + CXL_PMU_EVENT_CAP_REG(group_idx));
+ pmu_ev = devm_kzalloc(dev, sizeof(*pmu_ev), GFP_KERNEL);
+ if (!pmu_ev)
+ return -ENOMEM;
+
+ pmu_ev->vid = FIELD_GET(CXL_PMU_EVENT_CAP_VENDOR_ID_MSK, eval);
+ pmu_ev->gid = FIELD_GET(CXL_PMU_EVENT_CAP_GROUP_ID_MSK, eval);
+ /* For a fixed purpose counter use the events mask from the counter CFG */
+ pmu_ev->msk = events_msk;
+ pmu_ev->counter_idx = i;
+ /* This list add is never unwound as all entries deleted on remove */
+ list_add(&pmu_ev->node, &info->event_caps_fixed);
+ /*
+ * Configurable counters must not use an Event Capability registers that
+ * is in use for a Fixed counter
+ */
+ set_bit(group_idx, &fixed_counter_event_cap_bm);
+ }
+
+ if (!bitmap_empty(info->conf_counter_bm, CXL_PMU_MAX_COUNTERS)) {
+ struct cxl_pmu_ev_cap *pmu_ev;
+ int j;
+ /* Walk event capabilities unused by fixed counters */
+ for_each_clear_bit(j, &fixed_counter_event_cap_bm,
+ info->num_event_capabilities) {
+ pmu_ev = devm_kzalloc(dev, sizeof(*pmu_ev), GFP_KERNEL);
+ if (!pmu_ev)
+ return -ENOMEM;
+
+ eval = readq(base + CXL_PMU_EVENT_CAP_REG(j));
+ pmu_ev->vid = FIELD_GET(CXL_PMU_EVENT_CAP_VENDOR_ID_MSK, eval);
+ pmu_ev->gid = FIELD_GET(CXL_PMU_EVENT_CAP_GROUP_ID_MSK, eval);
+ pmu_ev->msk = FIELD_GET(CXL_PMU_EVENT_CAP_SUPPORTED_EVENTS_MSK, eval);
+ pmu_ev->event_idx = j;
+ list_add(&pmu_ev->node, &info->event_caps_configurable);
+ }
+ }
+
+ return 0;
+}
+
+static ssize_t cxl_pmu_format_sysfs_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct dev_ext_attribute *eattr;
+
+ eattr = container_of(attr, struct dev_ext_attribute, attr);
+
+ return sysfs_emit(buf, "%s\n", (char *)eattr->var);
+}
+
+#define CXL_PMU_FORMAT_ATTR(_name, _format)\
+ (&((struct dev_ext_attribute[]) { \
+ { \
+ .attr = __ATTR(_name, 0444, \
+ cxl_pmu_format_sysfs_show, NULL), \
+ .var = (void *)_format \
+ } \
+ })[0].attr.attr)
+
+enum {
+ cxl_pmu_mask_attr,
+ cxl_pmu_gid_attr,
+ cxl_pmu_vid_attr,
+ cxl_pmu_threshold_attr,
+ cxl_pmu_invert_attr,
+ cxl_pmu_edge_attr,
+ cxl_pmu_hdm_filter_en_attr,
+ cxl_pmu_hdm_attr,
+};
+
+static struct attribute *cxl_pmu_format_attr[] = {
+ [cxl_pmu_mask_attr] = CXL_PMU_FORMAT_ATTR(mask, "config:0-31"),
+ [cxl_pmu_gid_attr] = CXL_PMU_FORMAT_ATTR(gid, "config:32-47"),
+ [cxl_pmu_vid_attr] = CXL_PMU_FORMAT_ATTR(vid, "config:48-63"),
+ [cxl_pmu_threshold_attr] = CXL_PMU_FORMAT_ATTR(threshold, "config1:0-15"),
+ [cxl_pmu_invert_attr] = CXL_PMU_FORMAT_ATTR(invert, "config1:16"),
+ [cxl_pmu_edge_attr] = CXL_PMU_FORMAT_ATTR(edge, "config1:17"),
+ [cxl_pmu_hdm_filter_en_attr] = CXL_PMU_FORMAT_ATTR(hdm_filter_en, "config1:18"),
+ [cxl_pmu_hdm_attr] = CXL_PMU_FORMAT_ATTR(hdm, "config2:0-15"),
+ NULL
+};
+
+#define CXL_PMU_ATTR_CONFIG_MASK_MSK GENMASK_ULL(31, 0)
+#define CXL_PMU_ATTR_CONFIG_GID_MSK GENMASK_ULL(47, 32)
+#define CXL_PMU_ATTR_CONFIG_VID_MSK GENMASK_ULL(63, 48)
+#define CXL_PMU_ATTR_CONFIG1_THRESHOLD_MSK GENMASK_ULL(15, 0)
+#define CXL_PMU_ATTR_CONFIG1_INVERT_MSK BIT(16)
+#define CXL_PMU_ATTR_CONFIG1_EDGE_MSK BIT(17)
+#define CXL_PMU_ATTR_CONFIG1_FILTER_EN_MSK BIT(18)
+#define CXL_PMU_ATTR_CONFIG2_HDM_MSK GENMASK(15, 0)
+
+static umode_t cxl_pmu_format_is_visible(struct kobject *kobj,
+ struct attribute *attr, int a)
+{
+ struct device *dev = kobj_to_dev(kobj);
+ struct cxl_pmu_info *info = dev_get_drvdata(dev);
+
+ /*
+ * Filter capability at the CPMU level, so hide the attributes if the particular
+ * filter is not supported.
+ */
+ if (!info->filter_hdm &&
+ (attr == cxl_pmu_format_attr[cxl_pmu_hdm_filter_en_attr] ||
+ attr == cxl_pmu_format_attr[cxl_pmu_hdm_attr]))
+ return 0;
+
+ return attr->mode;
+}
+
+static const struct attribute_group cxl_pmu_format_group = {
+ .name = "format",
+ .attrs = cxl_pmu_format_attr,
+ .is_visible = cxl_pmu_format_is_visible,
+};
+
+static u32 cxl_pmu_config_get_mask(struct perf_event *event)
+{
+ return FIELD_GET(CXL_PMU_ATTR_CONFIG_MASK_MSK, event->attr.config);
+}
+
+static u16 cxl_pmu_config_get_gid(struct perf_event *event)
+{
+ return FIELD_GET(CXL_PMU_ATTR_CONFIG_GID_MSK, event->attr.config);
+}
+
+static u16 cxl_pmu_config_get_vid(struct perf_event *event)
+{
+ return FIELD_GET(CXL_PMU_ATTR_CONFIG_VID_MSK, event->attr.config);
+}
+
+static u8 cxl_pmu_config1_get_threshold(struct perf_event *event)
+{
+ return FIELD_GET(CXL_PMU_ATTR_CONFIG1_THRESHOLD_MSK, event->attr.config1);
+}
+
+static bool cxl_pmu_config1_get_invert(struct perf_event *event)
+{
+ return FIELD_GET(CXL_PMU_ATTR_CONFIG1_INVERT_MSK, event->attr.config1);
+}
+
+static bool cxl_pmu_config1_get_edge(struct perf_event *event)
+{
+ return FIELD_GET(CXL_PMU_ATTR_CONFIG1_EDGE_MSK, event->attr.config1);
+}
+
+/*
+ * CPMU specification allows for 8 filters, each with a 16 bit value...
+ * So we need to find 8x16bits to store it in.
+ * As the value used for disable is 0xffff, a separate enable switch
+ * is needed.
+ */
+
+static bool cxl_pmu_config1_hdm_filter_en(struct perf_event *event)
+{
+ return FIELD_GET(CXL_PMU_ATTR_CONFIG1_FILTER_EN_MSK, event->attr.config1);
+}
+
+static u16 cxl_pmu_config2_get_hdm_decoder(struct perf_event *event)
+{
+ return FIELD_GET(CXL_PMU_ATTR_CONFIG2_HDM_MSK, event->attr.config2);
+}
+
+static ssize_t cxl_pmu_event_sysfs_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct perf_pmu_events_attr *pmu_attr =
+ container_of(attr, struct perf_pmu_events_attr, attr);
+
+ return sysfs_emit(buf, "config=%#llx\n", pmu_attr->id);
+}
+
+#define CXL_PMU_EVENT_ATTR(_name, _vid, _gid, _msk) \
+ PMU_EVENT_ATTR_ID(_name, cxl_pmu_event_sysfs_show, \
+ ((u64)(_vid) << 48) | ((u64)(_gid) << 32) | (u64)(_msk))
+
+/* For CXL spec defined events */
+#define CXL_PMU_EVENT_CXL_ATTR(_name, _gid, _msk) \
+ CXL_PMU_EVENT_ATTR(_name, PCI_DVSEC_VENDOR_ID_CXL, _gid, _msk)
+
+static struct attribute *cxl_pmu_event_attrs[] = {
+ CXL_PMU_EVENT_CXL_ATTR(clock_ticks, CXL_PMU_GID_CLOCK_TICKS, BIT(0)),
+ /* CXL rev 3.0 Table 3-17 - Device to Host Requests */
+ CXL_PMU_EVENT_CXL_ATTR(d2h_req_rdcurr, CXL_PMU_GID_D2H_REQ, BIT(1)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_req_rdown, CXL_PMU_GID_D2H_REQ, BIT(2)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_req_rdshared, CXL_PMU_GID_D2H_REQ, BIT(3)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_req_rdany, CXL_PMU_GID_D2H_REQ, BIT(4)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_req_rdownnodata, CXL_PMU_GID_D2H_REQ, BIT(5)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_req_itomwr, CXL_PMU_GID_D2H_REQ, BIT(6)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_req_wrcurr, CXL_PMU_GID_D2H_REQ, BIT(7)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_req_clflush, CXL_PMU_GID_D2H_REQ, BIT(8)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_req_cleanevict, CXL_PMU_GID_D2H_REQ, BIT(9)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_req_dirtyevict, CXL_PMU_GID_D2H_REQ, BIT(10)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_req_cleanevictnodata, CXL_PMU_GID_D2H_REQ, BIT(11)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_req_wowrinv, CXL_PMU_GID_D2H_REQ, BIT(12)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_req_wowrinvf, CXL_PMU_GID_D2H_REQ, BIT(13)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_req_wrinv, CXL_PMU_GID_D2H_REQ, BIT(14)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_req_cacheflushed, CXL_PMU_GID_D2H_REQ, BIT(16)),
+ /* CXL rev 3.0 Table 3-20 - D2H Repsonse Encodings */
+ CXL_PMU_EVENT_CXL_ATTR(d2h_rsp_rspihiti, CXL_PMU_GID_D2H_RSP, BIT(4)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_rsp_rspvhitv, CXL_PMU_GID_D2H_RSP, BIT(6)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_rsp_rspihitse, CXL_PMU_GID_D2H_RSP, BIT(5)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_rsp_rspshitse, CXL_PMU_GID_D2H_RSP, BIT(1)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_rsp_rspsfwdm, CXL_PMU_GID_D2H_RSP, BIT(7)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_rsp_rspifwdm, CXL_PMU_GID_D2H_RSP, BIT(15)),
+ CXL_PMU_EVENT_CXL_ATTR(d2h_rsp_rspvfwdv, CXL_PMU_GID_D2H_RSP, BIT(22)),
+ /* CXL rev 3.0 Table 3-21 - CXL.cache - Mapping of H2D Requests to D2H Responses */
+ CXL_PMU_EVENT_CXL_ATTR(h2d_req_snpdata, CXL_PMU_GID_H2D_REQ, BIT(1)),
+ CXL_PMU_EVENT_CXL_ATTR(h2d_req_snpinv, CXL_PMU_GID_H2D_REQ, BIT(2)),
+ CXL_PMU_EVENT_CXL_ATTR(h2d_req_snpcur, CXL_PMU_GID_H2D_REQ, BIT(3)),
+ /* CXL rev 3.0 Table 3-22 - H2D Response Opcode Encodings */
+ CXL_PMU_EVENT_CXL_ATTR(h2d_rsp_writepull, CXL_PMU_GID_H2D_RSP, BIT(1)),
+ CXL_PMU_EVENT_CXL_ATTR(h2d_rsp_go, CXL_PMU_GID_H2D_RSP, BIT(4)),
+ CXL_PMU_EVENT_CXL_ATTR(h2d_rsp_gowritepull, CXL_PMU_GID_H2D_RSP, BIT(5)),
+ CXL_PMU_EVENT_CXL_ATTR(h2d_rsp_extcmp, CXL_PMU_GID_H2D_RSP, BIT(6)),
+ CXL_PMU_EVENT_CXL_ATTR(h2d_rsp_gowritepulldrop, CXL_PMU_GID_H2D_RSP, BIT(8)),
+ CXL_PMU_EVENT_CXL_ATTR(h2d_rsp_fastgowritepull, CXL_PMU_GID_H2D_RSP, BIT(13)),
+ CXL_PMU_EVENT_CXL_ATTR(h2d_rsp_goerrwritepull, CXL_PMU_GID_H2D_RSP, BIT(15)),
+ /* CXL rev 3.0 Table 13-5 directly lists these */
+ CXL_PMU_EVENT_CXL_ATTR(cachedata_d2h_data, CXL_PMU_GID_CACHE_DATA, BIT(0)),
+ CXL_PMU_EVENT_CXL_ATTR(cachedata_h2d_data, CXL_PMU_GID_CACHE_DATA, BIT(1)),
+ /* CXL rev 3.0 Table 3-29 M2S Req Memory Opcodes */
+ CXL_PMU_EVENT_CXL_ATTR(m2s_req_meminv, CXL_PMU_GID_M2S_REQ, BIT(0)),
+ CXL_PMU_EVENT_CXL_ATTR(m2s_req_memrd, CXL_PMU_GID_M2S_REQ, BIT(1)),
+ CXL_PMU_EVENT_CXL_ATTR(m2s_req_memrddata, CXL_PMU_GID_M2S_REQ, BIT(2)),
+ CXL_PMU_EVENT_CXL_ATTR(m2s_req_memrdfwd, CXL_PMU_GID_M2S_REQ, BIT(3)),
+ CXL_PMU_EVENT_CXL_ATTR(m2s_req_memwrfwd, CXL_PMU_GID_M2S_REQ, BIT(4)),
+ CXL_PMU_EVENT_CXL_ATTR(m2s_req_memspecrd, CXL_PMU_GID_M2S_REQ, BIT(8)),
+ CXL_PMU_EVENT_CXL_ATTR(m2s_req_meminvnt, CXL_PMU_GID_M2S_REQ, BIT(9)),
+ CXL_PMU_EVENT_CXL_ATTR(m2s_req_memcleanevict, CXL_PMU_GID_M2S_REQ, BIT(10)),
+ /* CXL rev 3.0 Table 3-35 M2S RwD Memory Opcodes */
+ CXL_PMU_EVENT_CXL_ATTR(m2s_rwd_memwr, CXL_PMU_GID_M2S_RWD, BIT(1)),
+ CXL_PMU_EVENT_CXL_ATTR(m2s_rwd_memwrptl, CXL_PMU_GID_M2S_RWD, BIT(2)),
+ CXL_PMU_EVENT_CXL_ATTR(m2s_rwd_biconflict, CXL_PMU_GID_M2S_RWD, BIT(4)),
+ /* CXL rev 3.0 Table 3-38 M2S BIRsp Memory Opcodes */
+ CXL_PMU_EVENT_CXL_ATTR(m2s_birsp_i, CXL_PMU_GID_M2S_BIRSP, BIT(0)),
+ CXL_PMU_EVENT_CXL_ATTR(m2s_birsp_s, CXL_PMU_GID_M2S_BIRSP, BIT(1)),
+ CXL_PMU_EVENT_CXL_ATTR(m2s_birsp_e, CXL_PMU_GID_M2S_BIRSP, BIT(2)),
+ CXL_PMU_EVENT_CXL_ATTR(m2s_birsp_iblk, CXL_PMU_GID_M2S_BIRSP, BIT(4)),
+ CXL_PMU_EVENT_CXL_ATTR(m2s_birsp_sblk, CXL_PMU_GID_M2S_BIRSP, BIT(5)),
+ CXL_PMU_EVENT_CXL_ATTR(m2s_birsp_eblk, CXL_PMU_GID_M2S_BIRSP, BIT(6)),
+ /* CXL rev 3.0 Table 3-40 S2M BISnp Opcodes */
+ CXL_PMU_EVENT_CXL_ATTR(s2m_bisnp_cur, CXL_PMU_GID_S2M_BISNP, BIT(0)),
+ CXL_PMU_EVENT_CXL_ATTR(s2m_bisnp_data, CXL_PMU_GID_S2M_BISNP, BIT(1)),
+ CXL_PMU_EVENT_CXL_ATTR(s2m_bisnp_inv, CXL_PMU_GID_S2M_BISNP, BIT(2)),
+ CXL_PMU_EVENT_CXL_ATTR(s2m_bisnp_curblk, CXL_PMU_GID_S2M_BISNP, BIT(4)),
+ CXL_PMU_EVENT_CXL_ATTR(s2m_bisnp_datblk, CXL_PMU_GID_S2M_BISNP, BIT(5)),
+ CXL_PMU_EVENT_CXL_ATTR(s2m_bisnp_invblk, CXL_PMU_GID_S2M_BISNP, BIT(6)),
+ /* CXL rev 3.0 Table 3-43 S2M NDR Opcopdes */
+ CXL_PMU_EVENT_CXL_ATTR(s2m_ndr_cmp, CXL_PMU_GID_S2M_NDR, BIT(0)),
+ CXL_PMU_EVENT_CXL_ATTR(s2m_ndr_cmps, CXL_PMU_GID_S2M_NDR, BIT(1)),
+ CXL_PMU_EVENT_CXL_ATTR(s2m_ndr_cmpe, CXL_PMU_GID_S2M_NDR, BIT(2)),
+ CXL_PMU_EVENT_CXL_ATTR(s2m_ndr_biconflictack, CXL_PMU_GID_S2M_NDR, BIT(3)),
+ /* CXL rev 3.0 Table 3-46 S2M DRS opcodes */
+ CXL_PMU_EVENT_CXL_ATTR(s2m_drs_memdata, CXL_PMU_GID_S2M_DRS, BIT(0)),
+ CXL_PMU_EVENT_CXL_ATTR(s2m_drs_memdatanxm, CXL_PMU_GID_S2M_DRS, BIT(1)),
+ /* CXL rev 3.0 Table 13-5 directly lists these */
+ CXL_PMU_EVENT_CXL_ATTR(ddr_act, CXL_PMU_GID_DDR, BIT(0)),
+ CXL_PMU_EVENT_CXL_ATTR(ddr_pre, CXL_PMU_GID_DDR, BIT(1)),
+ CXL_PMU_EVENT_CXL_ATTR(ddr_casrd, CXL_PMU_GID_DDR, BIT(2)),
+ CXL_PMU_EVENT_CXL_ATTR(ddr_caswr, CXL_PMU_GID_DDR, BIT(3)),
+ CXL_PMU_EVENT_CXL_ATTR(ddr_refresh, CXL_PMU_GID_DDR, BIT(4)),
+ CXL_PMU_EVENT_CXL_ATTR(ddr_selfrefreshent, CXL_PMU_GID_DDR, BIT(5)),
+ CXL_PMU_EVENT_CXL_ATTR(ddr_rfm, CXL_PMU_GID_DDR, BIT(6)),
+ NULL
+};
+
+static struct cxl_pmu_ev_cap *cxl_pmu_find_fixed_counter_ev_cap(struct cxl_pmu_info *info,
+ int vid, int gid, int msk)
+{
+ struct cxl_pmu_ev_cap *pmu_ev;
+
+ list_for_each_entry(pmu_ev, &info->event_caps_fixed, node) {
+ if (vid != pmu_ev->vid || gid != pmu_ev->gid)
+ continue;
+
+ /* Precise match for fixed counter */
+ if (msk == pmu_ev->msk)
+ return pmu_ev;
+ }
+
+ return ERR_PTR(-EINVAL);
+}
+
+static struct cxl_pmu_ev_cap *cxl_pmu_find_config_counter_ev_cap(struct cxl_pmu_info *info,
+ int vid, int gid, int msk)
+{
+ struct cxl_pmu_ev_cap *pmu_ev;
+
+ list_for_each_entry(pmu_ev, &info->event_caps_configurable, node) {
+ if (vid != pmu_ev->vid || gid != pmu_ev->gid)
+ continue;
+
+ /* Request mask must be subset of supported */
+ if (msk & ~pmu_ev->msk)
+ continue;
+
+ return pmu_ev;
+ }
+
+ return ERR_PTR(-EINVAL);
+}
+
+static umode_t cxl_pmu_event_is_visible(struct kobject *kobj, struct attribute *attr, int a)
+{
+ struct device_attribute *dev_attr = container_of(attr, struct device_attribute, attr);
+ struct perf_pmu_events_attr *pmu_attr =
+ container_of(dev_attr, struct perf_pmu_events_attr, attr);
+ struct device *dev = kobj_to_dev(kobj);
+ struct cxl_pmu_info *info = dev_get_drvdata(dev);
+ int vid = FIELD_GET(CXL_PMU_ATTR_CONFIG_VID_MSK, pmu_attr->id);
+ int gid = FIELD_GET(CXL_PMU_ATTR_CONFIG_GID_MSK, pmu_attr->id);
+ int msk = FIELD_GET(CXL_PMU_ATTR_CONFIG_MASK_MSK, pmu_attr->id);
+
+ if (!IS_ERR(cxl_pmu_find_fixed_counter_ev_cap(info, vid, gid, msk)))
+ return attr->mode;
+
+ if (!IS_ERR(cxl_pmu_find_config_counter_ev_cap(info, vid, gid, msk)))
+ return attr->mode;
+
+ return 0;
+}
+
+static const struct attribute_group cxl_pmu_events = {
+ .name = "events",
+ .attrs = cxl_pmu_event_attrs,
+ .is_visible = cxl_pmu_event_is_visible,
+};
+
+static ssize_t cpumask_show(struct device *dev, struct device_attribute *attr,
+ char *buf)
+{
+ struct cxl_pmu_info *info = dev_get_drvdata(dev);
+
+ return cpumap_print_to_pagebuf(true, buf, cpumask_of(info->on_cpu));
+}
+static DEVICE_ATTR_RO(cpumask);
+
+static struct attribute *cxl_pmu_cpumask_attrs[] = {
+ &dev_attr_cpumask.attr,
+ NULL
+};
+
+static const struct attribute_group cxl_pmu_cpumask_group = {
+ .attrs = cxl_pmu_cpumask_attrs,
+};
+
+static const struct attribute_group *cxl_pmu_attr_groups[] = {
+ &cxl_pmu_events,
+ &cxl_pmu_format_group,
+ &cxl_pmu_cpumask_group,
+ NULL
+};
+
+/* If counter_idx == NULL, don't try to allocate a counter. */
+static int cxl_pmu_get_event_idx(struct perf_event *event, int *counter_idx,
+ int *event_idx)
+{
+ struct cxl_pmu_info *info = pmu_to_cxl_pmu_info(event->pmu);
+ DECLARE_BITMAP(configurable_and_free, CXL_PMU_MAX_COUNTERS);
+ struct cxl_pmu_ev_cap *pmu_ev;
+ u32 mask;
+ u16 gid, vid;
+ int i;
+
+ vid = cxl_pmu_config_get_vid(event);
+ gid = cxl_pmu_config_get_gid(event);
+ mask = cxl_pmu_config_get_mask(event);
+
+ pmu_ev = cxl_pmu_find_fixed_counter_ev_cap(info, vid, gid, mask);
+ if (!IS_ERR(pmu_ev)) {
+ if (!counter_idx)
+ return 0;
+ if (!test_bit(pmu_ev->counter_idx, info->used_counter_bm)) {
+ *counter_idx = pmu_ev->counter_idx;
+ return 0;
+ }
+ /* Fixed counter is in use, but maybe a configurable one? */
+ }
+
+ pmu_ev = cxl_pmu_find_config_counter_ev_cap(info, vid, gid, mask);
+ if (!IS_ERR(pmu_ev)) {
+ if (!counter_idx)
+ return 0;
+
+ bitmap_andnot(configurable_and_free, info->conf_counter_bm,
+ info->used_counter_bm, CXL_PMU_MAX_COUNTERS);
+
+ i = find_first_bit(configurable_and_free, CXL_PMU_MAX_COUNTERS);
+ if (i == CXL_PMU_MAX_COUNTERS)
+ return -EINVAL;
+
+ *counter_idx = i;
+ return 0;
+ }
+
+ return -EINVAL;
+}
+
+static int cxl_pmu_event_init(struct perf_event *event)
+{
+ struct cxl_pmu_info *info = pmu_to_cxl_pmu_info(event->pmu);
+ int rc;
+
+ /* Top level type sanity check - is this a Hardware Event being requested */
+ if (event->attr.type != event->pmu->type)
+ return -ENOENT;
+
+ if (is_sampling_event(event) || event->attach_state & PERF_ATTACH_TASK)
+ return -EOPNOTSUPP;
+ /* TODO: Validation of any filter */
+
+ /*
+ * Verify that it is possible to count what was requested. Either must
+ * be a fixed counter that is a precise match or a configurable counter
+ * where this is a subset.
+ */
+ rc = cxl_pmu_get_event_idx(event, NULL, NULL);
+ if (rc < 0)
+ return rc;
+
+ event->cpu = info->on_cpu;
+
+ return 0;
+}
+
+static void cxl_pmu_enable(struct pmu *pmu)
+{
+ struct cxl_pmu_info *info = pmu_to_cxl_pmu_info(pmu);
+ void __iomem *base = info->base;
+
+ /* Can assume frozen at this stage */
+ writeq(0, base + CXL_PMU_FREEZE_REG);
+}
+
+static void cxl_pmu_disable(struct pmu *pmu)
+{
+ struct cxl_pmu_info *info = pmu_to_cxl_pmu_info(pmu);
+ void __iomem *base = info->base;
+
+ /*
+ * Whilst bits above number of counters are RsvdZ
+ * they are unlikely to be repurposed given
+ * number of counters is allowed to be 64 leaving
+ * no reserved bits. Hence this is only slightly
+ * naughty.
+ */
+ writeq(GENMASK_ULL(63, 0), base + CXL_PMU_FREEZE_REG);
+}
+
+static void cxl_pmu_event_start(struct perf_event *event, int flags)
+{
+ struct cxl_pmu_info *info = pmu_to_cxl_pmu_info(event->pmu);
+ struct hw_perf_event *hwc = &event->hw;
+ void __iomem *base = info->base;
+ u64 cfg;
+
+ /*
+ * All paths to here should either set these flags directly or
+ * call cxl_pmu_event_stop() which will ensure the correct state.
+ */
+ if (WARN_ON_ONCE(!(hwc->state & PERF_HES_STOPPED)))
+ return;
+
+ WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE));
+ hwc->state = 0;
+
+ /*
+ * Currently only hdm filter control is implemnted, this code will
+ * want generalizing when more filters are added.
+ */
+ if (info->filter_hdm) {
+ if (cxl_pmu_config1_hdm_filter_en(event))
+ cfg = cxl_pmu_config2_get_hdm_decoder(event);
+ else
+ cfg = GENMASK(15, 0); /* No filtering if 0xFFFF_FFFF */
+ writeq(cfg, base + CXL_PMU_FILTER_CFG_REG(hwc->idx, 0));
+ }
+
+ cfg = readq(base + CXL_PMU_COUNTER_CFG_REG(hwc->idx));
+ cfg |= FIELD_PREP(CXL_PMU_COUNTER_CFG_INT_ON_OVRFLW, 1);
+ cfg |= FIELD_PREP(CXL_PMU_COUNTER_CFG_FREEZE_ON_OVRFLW, 1);
+ cfg |= FIELD_PREP(CXL_PMU_COUNTER_CFG_ENABLE, 1);
+ cfg |= FIELD_PREP(CXL_PMU_COUNTER_CFG_EDGE,
+ cxl_pmu_config1_get_edge(event) ? 1 : 0);
+ cfg |= FIELD_PREP(CXL_PMU_COUNTER_CFG_INVERT,
+ cxl_pmu_config1_get_invert(event) ? 1 : 0);
+
+ /* Fixed purpose counters have next two fields RO */
+ if (test_bit(hwc->idx, info->conf_counter_bm)) {
+ cfg |= FIELD_PREP(CXL_PMU_COUNTER_CFG_EVENT_GRP_ID_IDX_MSK,
+ hwc->event_base);
+ cfg |= FIELD_PREP(CXL_PMU_COUNTER_CFG_EVENTS_MSK,
+ cxl_pmu_config_get_mask(event));
+ }
+ cfg &= ~CXL_PMU_COUNTER_CFG_THRESHOLD_MSK;
+ /*
+ * For events that generate only 1 count per clock the CXL 3.0 spec
+ * states the threshold shall be set to 1 but if set to 0 it will
+ * count the raw value anwyay?
+ * There is no definition of what events will count multiple per cycle
+ * and hence to which non 1 values of threshold can apply.
+ * (CXL 3.0 8.2.7.2.1 Counter Configuration - threshold field definition)
+ */
+ cfg |= FIELD_PREP(CXL_PMU_COUNTER_CFG_THRESHOLD_MSK,
+ cxl_pmu_config1_get_threshold(event));
+ writeq(cfg, base + CXL_PMU_COUNTER_CFG_REG(hwc->idx));
+
+ local64_set(&hwc->prev_count, 0);
+ writeq(0, base + CXL_PMU_COUNTER_REG(hwc->idx));
+
+ perf_event_update_userpage(event);
+}
+
+static u64 cxl_pmu_read_counter(struct perf_event *event)
+{
+ struct cxl_pmu_info *info = pmu_to_cxl_pmu_info(event->pmu);
+ void __iomem *base = info->base;
+
+ return readq(base + CXL_PMU_COUNTER_REG(event->hw.idx));
+}
+
+static void __cxl_pmu_read(struct perf_event *event, bool overflow)
+{
+ struct cxl_pmu_info *info = pmu_to_cxl_pmu_info(event->pmu);
+ struct hw_perf_event *hwc = &event->hw;
+ u64 new_cnt, prev_cnt, delta;
+
+ do {
+ prev_cnt = local64_read(&hwc->prev_count);
+ new_cnt = cxl_pmu_read_counter(event);
+ } while (local64_cmpxchg(&hwc->prev_count, prev_cnt, new_cnt) != prev_cnt);
+
+ /*
+ * If we know an overflow occur then take that into account.
+ * Note counter is not reset as that would lose events
+ */
+ delta = (new_cnt - prev_cnt) & GENMASK_ULL(info->counter_width - 1, 0);
+ if (overflow && delta < GENMASK_ULL(info->counter_width - 1, 0))
+ delta += (1UL << info->counter_width);
+
+ local64_add(delta, &event->count);
+}
+
+static void cxl_pmu_read(struct perf_event *event)
+{
+ __cxl_pmu_read(event, false);
+}
+
+static void cxl_pmu_event_stop(struct perf_event *event, int flags)
+{
+ struct cxl_pmu_info *info = pmu_to_cxl_pmu_info(event->pmu);
+ void __iomem *base = info->base;
+ struct hw_perf_event *hwc = &event->hw;
+ u64 cfg;
+
+ cxl_pmu_read(event);
+ WARN_ON_ONCE(hwc->state & PERF_HES_STOPPED);
+ hwc->state |= PERF_HES_STOPPED;
+
+ cfg = readq(base + CXL_PMU_COUNTER_CFG_REG(hwc->idx));
+ cfg &= ~(FIELD_PREP(CXL_PMU_COUNTER_CFG_INT_ON_OVRFLW, 1) |
+ FIELD_PREP(CXL_PMU_COUNTER_CFG_ENABLE, 1));
+ writeq(cfg, base + CXL_PMU_COUNTER_CFG_REG(hwc->idx));
+
+ hwc->state |= PERF_HES_UPTODATE;
+}
+
+static int cxl_pmu_event_add(struct perf_event *event, int flags)
+{
+ struct cxl_pmu_info *info = pmu_to_cxl_pmu_info(event->pmu);
+ struct hw_perf_event *hwc = &event->hw;
+ int idx, rc;
+ int event_idx = 0;
+
+ hwc->state = PERF_HES_STOPPED | PERF_HES_UPTODATE;
+
+ rc = cxl_pmu_get_event_idx(event, &idx, &event_idx);
+ if (rc < 0)
+ return rc;
+
+ hwc->idx = idx;
+
+ /* Only set for configurable counters */
+ hwc->event_base = event_idx;
+ info->hw_events[idx] = event;
+ set_bit(idx, info->used_counter_bm);
+
+ if (flags & PERF_EF_START)
+ cxl_pmu_event_start(event, PERF_EF_RELOAD);
+
+ return 0;
+}
+
+static void cxl_pmu_event_del(struct perf_event *event, int flags)
+{
+ struct cxl_pmu_info *info = pmu_to_cxl_pmu_info(event->pmu);
+ struct hw_perf_event *hwc = &event->hw;
+
+ cxl_pmu_event_stop(event, PERF_EF_UPDATE);
+ clear_bit(hwc->idx, info->used_counter_bm);
+ info->hw_events[hwc->idx] = NULL;
+ perf_event_update_userpage(event);
+}
+
+static irqreturn_t cxl_pmu_irq(int irq, void *data)
+{
+ struct cxl_pmu_info *info = data;
+ void __iomem *base = info->base;
+ u64 overflowed;
+ DECLARE_BITMAP(overflowedbm, 64);
+ int i;
+
+ overflowed = readq(base + CXL_PMU_OVERFLOW_REG);
+
+ /* Interrupt may be shared, so maybe it isn't ours */
+ if (!overflowed)
+ return IRQ_NONE;
+
+ bitmap_from_arr64(overflowedbm, &overflowed, 64);
+ for_each_set_bit(i, overflowedbm, info->num_counters) {
+ struct perf_event *event = info->hw_events[i];
+
+ if (!event) {
+ dev_dbg(info->pmu.dev,
+ "overflow but on non enabled counter %d\n", i);
+ continue;
+ }
+
+ __cxl_pmu_read(event, true);
+ }
+
+ writeq(overflowed, base + CXL_PMU_OVERFLOW_REG);
+
+ return IRQ_HANDLED;
+}
+
+static void cxl_pmu_perf_unregister(void *_info)
+{
+ struct cxl_pmu_info *info = _info;
+
+ perf_pmu_unregister(&info->pmu);
+}
+
+static void cxl_pmu_cpuhp_remove(void *_info)
+{
+ struct cxl_pmu_info *info = _info;
+
+ cpuhp_state_remove_instance_nocalls(cxl_pmu_cpuhp_state_num, &info->node);
+}
+
+static int cxl_pmu_probe(struct device *dev)
+{
+ struct cxl_pmu *pmu = to_cxl_pmu(dev);
+ struct pci_dev *pdev = to_pci_dev(dev->parent);
+ struct cxl_pmu_info *info;
+ char *irq_name;
+ char *dev_name;
+ int rc, irq;
+
+ info = devm_kzalloc(dev, sizeof(*info), GFP_KERNEL);
+ if (!info)
+ return -ENOMEM;
+
+ dev_set_drvdata(dev, info);
+ INIT_LIST_HEAD(&info->event_caps_fixed);
+ INIT_LIST_HEAD(&info->event_caps_configurable);
+
+ info->base = pmu->base;
+
+ info->on_cpu = -1;
+ rc = cxl_pmu_parse_caps(dev, info);
+ if (rc)
+ return rc;
+
+ info->hw_events = devm_kcalloc(dev, sizeof(*info->hw_events),
+ info->num_counters, GFP_KERNEL);
+ if (!info->hw_events)
+ return -ENOMEM;
+
+ switch (pmu->type) {
+ case CXL_PMU_MEMDEV:
+ dev_name = devm_kasprintf(dev, GFP_KERNEL, "cxl_pmu_mem%d.%d",
+ pmu->assoc_id, pmu->index);
+ break;
+ }
+ if (!dev_name)
+ return -ENOMEM;
+
+ info->pmu = (struct pmu) {
+ .name = dev_name,
+ .parent = dev,
+ .module = THIS_MODULE,
+ .event_init = cxl_pmu_event_init,
+ .pmu_enable = cxl_pmu_enable,
+ .pmu_disable = cxl_pmu_disable,
+ .add = cxl_pmu_event_add,
+ .del = cxl_pmu_event_del,
+ .start = cxl_pmu_event_start,
+ .stop = cxl_pmu_event_stop,
+ .read = cxl_pmu_read,
+ .task_ctx_nr = perf_invalid_context,
+ .attr_groups = cxl_pmu_attr_groups,
+ .capabilities = PERF_PMU_CAP_NO_EXCLUDE,
+ };
+
+ if (info->irq <= 0)
+ return -EINVAL;
+
+ rc = pci_irq_vector(pdev, info->irq);
+ if (rc < 0)
+ return rc;
+ irq = rc;
+
+ irq_name = devm_kasprintf(dev, GFP_KERNEL, "%s_overflow\n", dev_name);
+ if (!irq_name)
+ return -ENOMEM;
+
+ rc = devm_request_irq(dev, irq, cxl_pmu_irq, IRQF_SHARED | IRQF_ONESHOT,
+ irq_name, info);
+ if (rc)
+ return rc;
+ info->irq = irq;
+
+ rc = cpuhp_state_add_instance(cxl_pmu_cpuhp_state_num, &info->node);
+ if (rc)
+ return rc;
+
+ rc = devm_add_action_or_reset(dev, cxl_pmu_cpuhp_remove, info);
+ if (rc)
+ return rc;
+
+ rc = perf_pmu_register(&info->pmu, info->pmu.name, -1);
+ if (rc)
+ return rc;
+
+ rc = devm_add_action_or_reset(dev, cxl_pmu_perf_unregister, info);
+ if (rc)
+ return rc;
+
+ return 0;
+}
+
+static struct cxl_driver cxl_pmu_driver = {
+ .name = "cxl_pmu",
+ .probe = cxl_pmu_probe,
+ .id = CXL_DEVICE_PMU,
+};
+
+static int cxl_pmu_online_cpu(unsigned int cpu, struct hlist_node *node)
+{
+ struct cxl_pmu_info *info = hlist_entry_safe(node, struct cxl_pmu_info, node);
+
+ if (info->on_cpu != -1)
+ return 0;
+
+ info->on_cpu = cpu;
+ /*
+ * CPU HP lock is held so we should be guaranteed that the CPU hasn't yet
+ * gone away again.
+ */
+ WARN_ON(irq_set_affinity(info->irq, cpumask_of(cpu)));
+
+ return 0;
+}
+
+static int cxl_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node)
+{
+ struct cxl_pmu_info *info = hlist_entry_safe(node, struct cxl_pmu_info, node);
+ unsigned int target;
+
+ if (info->on_cpu != cpu)
+ return 0;
+
+ info->on_cpu = -1;
+ target = cpumask_any_but(cpu_online_mask, cpu);
+ if (target >= nr_cpu_ids) {
+ dev_err(info->pmu.dev, "Unable to find a suitable CPU\n");
+ return 0;
+ }
+
+ perf_pmu_migrate_context(&info->pmu, cpu, target);
+ info->on_cpu = target;
+ /*
+ * CPU HP lock is held so we should be guaranteed that this CPU hasn't yet
+ * gone away.
+ */
+ WARN_ON(irq_set_affinity(info->irq, cpumask_of(target)));
+
+ return 0;
+}
+
+static __init int cxl_pmu_init(void)
+{
+ int rc;
+
+ rc = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN,
+ "AP_PERF_CXL_PMU_ONLINE",
+ cxl_pmu_online_cpu, cxl_pmu_offline_cpu);
+ if (rc < 0)
+ return rc;
+ cxl_pmu_cpuhp_state_num = rc;
+
+ rc = cxl_driver_register(&cxl_pmu_driver);
+ if (rc)
+ cpuhp_remove_multi_state(cxl_pmu_cpuhp_state_num);
+
+ return rc;
+}
+
+static __exit void cxl_pmu_exit(void)
+{
+ cxl_driver_unregister(&cxl_pmu_driver);
+ cpuhp_remove_multi_state(cxl_pmu_cpuhp_state_num);
+}
+
+MODULE_LICENSE("GPL");
+MODULE_IMPORT_NS(CXL);
+module_init(cxl_pmu_init);
+module_exit(cxl_pmu_exit);
+MODULE_ALIAS_CXL(CXL_DEVICE_PMU);
--
2.39.2On Fri, May 26, 2023 at 3:00 AM Jonathan Cameron
<Jonathan.Cameron@huawei.com> wrote:
>
> CXL rev 3.0 introduces a standard performance monitoring hardware
> block to CXL. Instances are discovered using CXL Register Locator DVSEC
> entries. Each CXL component may have multiple PMUs.
>
> This initial driver supports a subset of types of counter.
> It supports counters that are either fixed or configurable, but requires
> that they support the ability to freeze and write value whilst frozen.
>
> Development done with QEMU model which will be posted shortly.
>
> Example:
>
> $ perf stat -a -e cxl_pmu_mem0.0/h2d_req_snpcur/ -e cxl_pmu_mem0.0/h2d_req_snpdata/ -e cxl_pmu_mem0.0/clock_ticks/ sleep 1
>
> Performance counter stats for 'system wide':
>
> 96,757,023,244,321 cxl_pmu_mem0.0/h2d_req_snpcur/
> 96,757,023,244,365 cxl_pmu_mem0.0/h2d_req_snpdata/
> 193,514,046,488,653 cxl_pmu_mem0.0/clock_ticks/
>
> 1.090539600 seconds time elapsed
>
> Reviewed-by: Dave Jiang <dave.jiang@intel.com>
> Reviewed-by: Kan Liang <kan.liang@linux.intel.com>
> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
> ---
> v7: THanks to Namhyung and Stephane for reviews
> - Set the cpu for an event only after validation checks.
> - Added -a to make perf default of all CPUs explicit in command line.
> - Fix wrong event names in example.
> ---
[SNIP]
> +static void cxl_pmu_event_start(struct perf_event *event, int flags)
> +{
> + struct cxl_pmu_info *info = pmu_to_cxl_pmu_info(event->pmu);
> + struct hw_perf_event *hwc = &event->hw;
> + void __iomem *base = info->base;
> + u64 cfg;
> +
> + /*
> + * All paths to here should either set these flags directly or
> + * call cxl_pmu_event_stop() which will ensure the correct state.
> + */
> + if (WARN_ON_ONCE(!(hwc->state & PERF_HES_STOPPED)))
> + return;
> +
> + WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE));
> + hwc->state = 0;
> +
> + /*
> + * Currently only hdm filter control is implemnted, this code will
Typo: implemented
> + * want generalizing when more filters are added.
> + */
> + if (info->filter_hdm) {
> + if (cxl_pmu_config1_hdm_filter_en(event))
> + cfg = cxl_pmu_config2_get_hdm_decoder(event);
> + else
> + cfg = GENMASK(15, 0); /* No filtering if 0xFFFF_FFFF */
> + writeq(cfg, base + CXL_PMU_FILTER_CFG_REG(hwc->idx, 0));
> + }
> +
> + cfg = readq(base + CXL_PMU_COUNTER_CFG_REG(hwc->idx));
> + cfg |= FIELD_PREP(CXL_PMU_COUNTER_CFG_INT_ON_OVRFLW, 1);
> + cfg |= FIELD_PREP(CXL_PMU_COUNTER_CFG_FREEZE_ON_OVRFLW, 1);
> + cfg |= FIELD_PREP(CXL_PMU_COUNTER_CFG_ENABLE, 1);
> + cfg |= FIELD_PREP(CXL_PMU_COUNTER_CFG_EDGE,
> + cxl_pmu_config1_get_edge(event) ? 1 : 0);
> + cfg |= FIELD_PREP(CXL_PMU_COUNTER_CFG_INVERT,
> + cxl_pmu_config1_get_invert(event) ? 1 : 0);
> +
> + /* Fixed purpose counters have next two fields RO */
> + if (test_bit(hwc->idx, info->conf_counter_bm)) {
> + cfg |= FIELD_PREP(CXL_PMU_COUNTER_CFG_EVENT_GRP_ID_IDX_MSK,
> + hwc->event_base);
> + cfg |= FIELD_PREP(CXL_PMU_COUNTER_CFG_EVENTS_MSK,
> + cxl_pmu_config_get_mask(event));
> + }
> + cfg &= ~CXL_PMU_COUNTER_CFG_THRESHOLD_MSK;
> + /*
> + * For events that generate only 1 count per clock the CXL 3.0 spec
> + * states the threshold shall be set to 1 but if set to 0 it will
> + * count the raw value anwyay?
Typo: anyway?
> + * There is no definition of what events will count multiple per cycle
> + * and hence to which non 1 values of threshold can apply.
> + * (CXL 3.0 8.2.7.2.1 Counter Configuration - threshold field definition)
> + */
> + cfg |= FIELD_PREP(CXL_PMU_COUNTER_CFG_THRESHOLD_MSK,
> + cxl_pmu_config1_get_threshold(event));
> + writeq(cfg, base + CXL_PMU_COUNTER_CFG_REG(hwc->idx));
> +
> + local64_set(&hwc->prev_count, 0);
> + writeq(0, base + CXL_PMU_COUNTER_REG(hwc->idx));
> +
> + perf_event_update_userpage(event);
> +}
> +
[SNIP]
> +
> +static int cxl_pmu_online_cpu(unsigned int cpu, struct hlist_node *node)
> +{
> + struct cxl_pmu_info *info = hlist_entry_safe(node, struct cxl_pmu_info, node);
> +
> + if (info->on_cpu != -1)
> + return 0;
> +
> + info->on_cpu = cpu;
> + /*
> + * CPU HP lock is held so we should be guaranteed that the CPU hasn't yet
> + * gone away again.
> + */
> + WARN_ON(irq_set_affinity(info->irq, cpumask_of(cpu)));
> +
> + return 0;
> +}
> +
> +static int cxl_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node)
> +{
> + struct cxl_pmu_info *info = hlist_entry_safe(node, struct cxl_pmu_info, node);
> + unsigned int target;
> +
> + if (info->on_cpu != cpu)
> + return 0;
> +
> + info->on_cpu = -1;
> + target = cpumask_any_but(cpu_online_mask, cpu);
> + if (target >= nr_cpu_ids) {
> + dev_err(info->pmu.dev, "Unable to find a suitable CPU\n");
> + return 0;
> + }
> +
> + perf_pmu_migrate_context(&info->pmu, cpu, target);
> + info->on_cpu = target;
> + /*
> + * CPU HP lock is held so we should be guaranteed that this CPU hasn't yet
> + * gone away.
> + */
> + WARN_ON(irq_set_affinity(info->irq, cpumask_of(target)));
> +
> + return 0;
> +}
IIUC a CXL PMU hardware (say cxl_pmu_mem0.0) is shared across
all CPUs and it would return the same value when read from any CPU,
right?
Thanks,
Namhyung
> +
> +static __init int cxl_pmu_init(void)
> +{
> + int rc;
> +
> + rc = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN,
> + "AP_PERF_CXL_PMU_ONLINE",
> + cxl_pmu_online_cpu, cxl_pmu_offline_cpu);
> + if (rc < 0)
> + return rc;
> + cxl_pmu_cpuhp_state_num = rc;
> +
> + rc = cxl_driver_register(&cxl_pmu_driver);
> + if (rc)
> + cpuhp_remove_multi_state(cxl_pmu_cpuhp_state_num);
> +
> + return rc;
> +}
> +
> +static __exit void cxl_pmu_exit(void)
> +{
> + cxl_driver_unregister(&cxl_pmu_driver);
> + cpuhp_remove_multi_state(cxl_pmu_cpuhp_state_num);
> +}
> +
> +MODULE_LICENSE("GPL");
> +MODULE_IMPORT_NS(CXL);
> +module_init(cxl_pmu_init);
> +module_exit(cxl_pmu_exit);
> +MODULE_ALIAS_CXL(CXL_DEVICE_PMU);
> --
> 2.39.2
>
Hi,
Tidied up the typos. Thanks,
> > +static int cxl_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node)
> > +{
> > + struct cxl_pmu_info *info = hlist_entry_safe(node, struct cxl_pmu_info, node);
> > + unsigned int target;
> > +
> > + if (info->on_cpu != cpu)
> > + return 0;
> > +
> > + info->on_cpu = -1;
> > + target = cpumask_any_but(cpu_online_mask, cpu);
> > + if (target >= nr_cpu_ids) {
> > + dev_err(info->pmu.dev, "Unable to find a suitable CPU\n");
> > + return 0;
> > + }
> > +
> > + perf_pmu_migrate_context(&info->pmu, cpu, target);
> > + info->on_cpu = target;
> > + /*
> > + * CPU HP lock is held so we should be guaranteed that this CPU hasn't yet
> > + * gone away.
> > + */
> > + WARN_ON(irq_set_affinity(info->irq, cpumask_of(target)));
> > +
> > + return 0;
> > +}
>
> IIUC a CXL PMU hardware (say cxl_pmu_mem0.0) is shared across
> all CPUs and it would return the same value when read from any CPU,
> right?
Correct, it will return the same value when used from any CPU.
I'm not sure what issue you are indicating.
My understanding is that, even for such cases, perf uses percpu
variables that mean we still have to ensure that the interrupt
handling occurs on the CPU we have migrated the context to.
There are a lot of similar driver in perf already from a quick
git grep cpumask_any_but\(cpu_online_mask,
It might be nice to enable perf to operate for these devices without
the percpu context though. I haven't looked into whether that
is worth doing.
Jonathan
>
> Thanks,
> Namhyung
>
>
> > +
> > +static __init int cxl_pmu_init(void)
> > +{
> > + int rc;
> > +
> > + rc = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN,
> > + "AP_PERF_CXL_PMU_ONLINE",
> > + cxl_pmu_online_cpu, cxl_pmu_offline_cpu);
> > + if (rc < 0)
> > + return rc;
> > + cxl_pmu_cpuhp_state_num = rc;
> > +
> > + rc = cxl_driver_register(&cxl_pmu_driver);
> > + if (rc)
> > + cpuhp_remove_multi_state(cxl_pmu_cpuhp_state_num);
> > +
> > + return rc;
> > +}
> > +
> > +static __exit void cxl_pmu_exit(void)
> > +{
> > + cxl_driver_unregister(&cxl_pmu_driver);
> > + cpuhp_remove_multi_state(cxl_pmu_cpuhp_state_num);
> > +}
> > +
> > +MODULE_LICENSE("GPL");
> > +MODULE_IMPORT_NS(CXL);
> > +module_init(cxl_pmu_init);
> > +module_exit(cxl_pmu_exit);
> > +MODULE_ALIAS_CXL(CXL_DEVICE_PMU);
> > --
> > 2.39.2
> >
>
On Tue, May 30, 2023 at 6:50 AM Jonathan Cameron
<Jonathan.Cameron@huawei.com> wrote:
>
>
> Hi,
>
> Tidied up the typos. Thanks,
>
> > > +static int cxl_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node)
> > > +{
> > > + struct cxl_pmu_info *info = hlist_entry_safe(node, struct cxl_pmu_info, node);
> > > + unsigned int target;
> > > +
> > > + if (info->on_cpu != cpu)
> > > + return 0;
> > > +
> > > + info->on_cpu = -1;
> > > + target = cpumask_any_but(cpu_online_mask, cpu);
> > > + if (target >= nr_cpu_ids) {
> > > + dev_err(info->pmu.dev, "Unable to find a suitable CPU\n");
> > > + return 0;
> > > + }
> > > +
> > > + perf_pmu_migrate_context(&info->pmu, cpu, target);
> > > + info->on_cpu = target;
> > > + /*
> > > + * CPU HP lock is held so we should be guaranteed that this CPU hasn't yet
> > > + * gone away.
> > > + */
> > > + WARN_ON(irq_set_affinity(info->irq, cpumask_of(target)));
> > > +
> > > + return 0;
> > > +}
> >
> > IIUC a CXL PMU hardware (say cxl_pmu_mem0.0) is shared across
> > all CPUs and it would return the same value when read from any CPU,
> > right?
>
> Correct, it will return the same value when used from any CPU.
> I'm not sure what issue you are indicating.
>
> My understanding is that, even for such cases, perf uses percpu
> variables that mean we still have to ensure that the interrupt
> handling occurs on the CPU we have migrated the context to.
>
> There are a lot of similar driver in perf already from a quick
> git grep cpumask_any_but\(cpu_online_mask,
>
> It might be nice to enable perf to operate for these devices without
> the percpu context though. I haven't looked into whether that
> is worth doing.
I was thinking if we could avoid IPIs for read from other
CPUs. It has PERF_EV_CAP_READ_ACTIVE_PKG for some uncore events
which can be shared among CPUs in the same package
to skip the IPI and to read it from the local CPU.
But I think this can be a separate step.
Thanks,
Namhyung
Jonathan Cameron wrote: > > Hi, > > Tidied up the typos. Thanks, Care to send that replacement patch? It looks like, in linux-next, no other tree has picked up: http://lore.kernel.org/r/20230526095824.16336-2-Jonathan.Cameron@huawei.com ...so I'll queue that as well.
© 2016 - 2024 Red Hat, Inc.