diff mbox series

[v3,4/6] kvm powerpc/book3s-apiv2: Introduce kvm-hv specific PMU

Message ID 20250123120749.90505-5-vaibhav@linux.ibm.com (mailing list archive)
State New
Headers show
Series kvm powerpc/book3s-hv: Expose Hostwide counters as perf-events | expand

Commit Message

Vaibhav Jain Jan. 23, 2025, 12:07 p.m. UTC
Introduce a new PMU named 'kvm-hv' to report Book3s kvm-hv specific
performance counters. This will expose KVM-HV specific performance
attributes to user-space via kernel's PMU infrastructure and would enable
users to monitor active kvm-hv based guests.

The patch creates necessary scaffolding to for the new PMU callbacks and
introduces two new exports kvmppc_{,un}register_pmu() that are called from
kvm-hv init and exit function to perform initialize and cleanup for the
'kvm-hv' PMU. The patch doesn't introduce any perf-events yet, which will
be introduced in later patches

Signed-off-by: Vaibhav Jain <vaibhav@linux.ibm.com>

---
Changelog

v2->v3:
* Fixed a build warning reported by kernel build robot.
Link:
https://lore.kernel.org/oe-kbuild-all/202501171030.3x0gqW8G-lkp@intel.com

v1->v2:
* Fixed an issue of kvm-hv not loading on baremetal kvm [Gautam]
---
 arch/powerpc/include/asm/kvm_book3s.h |  20 ++++
 arch/powerpc/kvm/Makefile             |   6 ++
 arch/powerpc/kvm/book3s_hv.c          |   9 ++
 arch/powerpc/kvm/book3s_hv_pmu.c      | 133 ++++++++++++++++++++++++++
 4 files changed, 168 insertions(+)
 create mode 100644 arch/powerpc/kvm/book3s_hv_pmu.c

Comments

Athira Rajeev Jan. 24, 2025, 6:50 p.m. UTC | #1
> On 23 Jan 2025, at 5:37 PM, Vaibhav Jain <vaibhav@linux.ibm.com> wrote:
> 
> Introduce a new PMU named 'kvm-hv' to report Book3s kvm-hv specific
> performance counters. This will expose KVM-HV specific performance
> attributes to user-space via kernel's PMU infrastructure and would enable
> users to monitor active kvm-hv based guests.
> 
> The patch creates necessary scaffolding to for the new PMU callbacks and
> introduces two new exports kvmppc_{,un}register_pmu() that are called from
> kvm-hv init and exit function to perform initialize and cleanup for the
> 'kvm-hv' PMU. The patch doesn't introduce any perf-events yet, which will
> be introduced in later patches
> 
> Signed-off-by: Vaibhav Jain <vaibhav@linux.ibm.com>
> 
> ---
> Changelog
> 
> v2->v3:
> * Fixed a build warning reported by kernel build robot.
> Link:
> https://lore.kernel.org/oe-kbuild-all/202501171030.3x0gqW8G-lkp@intel.com
> 
> v1->v2:
> * Fixed an issue of kvm-hv not loading on baremetal kvm [Gautam]
> ---
> arch/powerpc/include/asm/kvm_book3s.h |  20 ++++
> arch/powerpc/kvm/Makefile             |   6 ++
> arch/powerpc/kvm/book3s_hv.c          |   9 ++
> arch/powerpc/kvm/book3s_hv_pmu.c      | 133 ++++++++++++++++++++++++++
> 4 files changed, 168 insertions(+)
> create mode 100644 arch/powerpc/kvm/book3s_hv_pmu.c
> 
> diff --git a/arch/powerpc/include/asm/kvm_book3s.h b/arch/powerpc/include/asm/kvm_book3s.h
> index e1ff291ba891..7a7854c65ebb 100644
> --- a/arch/powerpc/include/asm/kvm_book3s.h
> +++ b/arch/powerpc/include/asm/kvm_book3s.h
> @@ -348,6 +348,26 @@ static inline bool kvmhv_is_nestedv1(void)
> 
> #endif
> 
> +/* kvm-ppc pmu registration */
> +#if IS_ENABLED(CONFIG_KVM_BOOK3S_64_HV)
> +#ifdef CONFIG_PERF_EVENTS
> +int kvmppc_register_pmu(void);
> +void kvmppc_unregister_pmu(void);
> +
> +#else
> +
> +static inline int kvmppc_register_pmu(void)
> +{
> + return 0;
> +}
> +
> +static inline void kvmppc_unregister_pmu(void)
> +{
> + /* do nothing */
> +}
> +#endif /* CONFIG_PERF_EVENTS */
> +#endif /* CONFIG_KVM_BOOK3S_64_HV */
> +
> int __kvmhv_nestedv2_reload_ptregs(struct kvm_vcpu *vcpu, struct pt_regs *regs);
> int __kvmhv_nestedv2_mark_dirty_ptregs(struct kvm_vcpu *vcpu, struct pt_regs *regs);
> int __kvmhv_nestedv2_mark_dirty(struct kvm_vcpu *vcpu, u16 iden);
> diff --git a/arch/powerpc/kvm/Makefile b/arch/powerpc/kvm/Makefile
> index 4bd9d1230869..7645307ff277 100644
> --- a/arch/powerpc/kvm/Makefile
> +++ b/arch/powerpc/kvm/Makefile
> @@ -92,6 +92,12 @@ kvm-book3s_64-builtin-objs-$(CONFIG_KVM_BOOK3S_64_HANDLER) += \
> $(kvm-book3s_64-builtin-tm-objs-y) \
> $(kvm-book3s_64-builtin-xics-objs-y)
> 
> +# enable kvm_hv perf events
> +ifdef CONFIG_PERF_EVENTS
> +kvm-book3s_64-builtin-objs-$(CONFIG_KVM_BOOK3S_64_HANDLER) += \
> + book3s_hv_pmu.o
> +endif
> +
> obj-$(CONFIG_GUEST_STATE_BUFFER_TEST) += test-guest-state-buffer.o
> endif
> 
> diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
> index 25429905ae90..6365b8126574 100644
> --- a/arch/powerpc/kvm/book3s_hv.c
> +++ b/arch/powerpc/kvm/book3s_hv.c
> @@ -6662,6 +6662,14 @@ static int kvmppc_book3s_init_hv(void)
> return r;
> }
> 
> + r = kvmppc_register_pmu();
> + if (r == -EOPNOTSUPP) {
> + pr_info("KVM-HV: PMU not supported %d\n", r);
> + } else if (r) {
> + pr_err("KVM-HV: Unable to register PMUs %d\n", r);
> + goto err;
> + }
> +
> kvm_ops_hv.owner = THIS_MODULE;
> kvmppc_hv_ops = &kvm_ops_hv;
> 
> @@ -6676,6 +6684,7 @@ static int kvmppc_book3s_init_hv(void)
> 
> static void kvmppc_book3s_exit_hv(void)
> {
> + kvmppc_unregister_pmu();
> kvmppc_uvmem_free();
> kvmppc_free_host_rm_ops();
> if (kvmppc_radix_possible())
> diff --git a/arch/powerpc/kvm/book3s_hv_pmu.c b/arch/powerpc/kvm/book3s_hv_pmu.c
> new file mode 100644
> index 000000000000..8c6ed30b7654
> --- /dev/null
> +++ b/arch/powerpc/kvm/book3s_hv_pmu.c
> @@ -0,0 +1,133 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Description: PMUs specific to running nested KVM-HV guests
> + * on Book3S processors (specifically POWER9 and later).
> + */
> +
> +#define pr_fmt(fmt)  "kvmppc-pmu: " fmt

Hi Vaibhav

All PMU specific code is under “arch/powerpc/perf in the kernel source. Here since we are introducing a kvm-hv specific PMU, can we please have it in arch/powerpc/perf ?

Thanks
Athira
> +
> +#include "asm-generic/local64.h"
> +#include <linux/kernel.h>
> +#include <linux/errno.h>
> +#include <linux/ratelimit.h>
> +#include <linux/kvm_host.h>
> +#include <linux/gfp_types.h>
> +#include <linux/pgtable.h>
> +#include <linux/perf_event.h>
> +#include <linux/spinlock_types.h>
> +#include <linux/spinlock.h>
> +
> +#include <asm/types.h>
> +#include <asm/kvm_ppc.h>
> +#include <asm/kvm_book3s.h>
> +#include <asm/mmu.h>
> +#include <asm/pgalloc.h>
> +#include <asm/pte-walk.h>
> +#include <asm/reg.h>
> +#include <asm/plpar_wrappers.h>
> +#include <asm/firmware.h>
> +
> +enum kvmppc_pmu_eventid {
> + KVMPPC_EVENT_MAX,
> +};
> +
> +static struct attribute *kvmppc_pmu_events_attr[] = {
> + NULL,
> +};
> +
> +static const struct attribute_group kvmppc_pmu_events_group = {
> + .name = "events",
> + .attrs = kvmppc_pmu_events_attr,
> +};
> +
> +PMU_FORMAT_ATTR(event, "config:0");
> +static struct attribute *kvmppc_pmu_format_attr[] = {
> + &format_attr_event.attr,
> + NULL,
> +};
> +
> +static struct attribute_group kvmppc_pmu_format_group = {
> + .name = "format",
> + .attrs = kvmppc_pmu_format_attr,
> +};
> +
> +static const struct attribute_group *kvmppc_pmu_attr_groups[] = {
> + &kvmppc_pmu_events_group,
> + &kvmppc_pmu_format_group,
> + NULL,
> +};
> +
> +static int kvmppc_pmu_event_init(struct perf_event *event)
> +{
> + unsigned int config = event->attr.config;
> +
> + pr_debug("%s: Event(%p) id=%llu cpu=%x on_cpu=%x config=%u",
> + __func__, event, event->id, event->cpu,
> + event->oncpu, config);
> +
> + if (event->attr.type != event->pmu->type)
> + return -ENOENT;
> +
> + if (config >= KVMPPC_EVENT_MAX)
> + return -EINVAL;
> +
> + local64_set(&event->hw.prev_count, 0);
> + local64_set(&event->count, 0);
> +
> + return 0;
> +}
> +
> +static void kvmppc_pmu_del(struct perf_event *event, int flags)
> +{
> +}
> +
> +static int kvmppc_pmu_add(struct perf_event *event, int flags)
> +{
> + return 0;
> +}
> +
> +static void kvmppc_pmu_read(struct perf_event *event)
> +{
> +}
> +
> +/* L1 wide counters PMU */
> +static struct pmu kvmppc_pmu = {
> + .task_ctx_nr = perf_sw_context,
> + .name = "kvm-hv",
> + .event_init = kvmppc_pmu_event_init,
> + .add = kvmppc_pmu_add,
> + .del = kvmppc_pmu_del,
> + .read = kvmppc_pmu_read,
> + .attr_groups = kvmppc_pmu_attr_groups,
> + .type = -1,
> +};
> +
> +int kvmppc_register_pmu(void)
> +{
> + int rc = -EOPNOTSUPP;
> +
> + /* only support events for nestedv2 right now */
> + if (kvmhv_is_nestedv2()) {
> + /* Setup done now register the PMU */
> + pr_info("Registering kvm-hv pmu");
> +
> + /* Register only if we arent already registered */
> + rc = (kvmppc_pmu.type == -1) ?
> +     perf_pmu_register(&kvmppc_pmu, kvmppc_pmu.name,
> +       -1) : 0;
> + }
> +
> + return rc;
> +}
> +EXPORT_SYMBOL_GPL(kvmppc_register_pmu);
> +
> +void kvmppc_unregister_pmu(void)
> +{
> + if (kvmhv_is_nestedv2()) {
> + if (kvmppc_pmu.type != -1)
> + perf_pmu_unregister(&kvmppc_pmu);
> +
> + pr_info("kvmhv_pmu unregistered.\n");
> + }
> +}
> +EXPORT_SYMBOL_GPL(kvmppc_unregister_pmu);
> -- 
> 2.48.1
> 
> 
>
Vaibhav Jain Jan. 27, 2025, 7:36 a.m. UTC | #2
Hi Athira,

Thanks for reviewing this patch series. My responses to your review
comment inline below:


Athira Rajeev <atrajeev@linux.vnet.ibm.com> writes:

>> On 23 Jan 2025, at 5:37 PM, Vaibhav Jain <vaibhav@linux.ibm.com> wrote:
>> 
>> Introduce a new PMU named 'kvm-hv' to report Book3s kvm-hv specific
<snip>
>
> Hi Vaibhav
>
> All PMU specific code is under “arch/powerpc/perf in the kernel source. Here since we are introducing a kvm-hv specific PMU, can we please have it in arch/powerpc/perf ?

Is it common convention to put PMU specific code in
arch/powerpc/perf across ppc achitecture variants ? If its there can you
please mention the reasons behind it.

Also the code for this PMU, will be part of kvm-hv kernel module as it
utilizes the functionality implemented there. Moving this PMU code to
arch/powerpc/perf will need this to be converted in yet another new
kernel module, adding a dependency to kvm-hv module and exporting a
bunch of functionality from kvm-hv. Which looks bit messy to me

<snip>
Madhavan Srinivasan Jan. 28, 2025, 11:57 a.m. UTC | #3
On 1/27/25 1:06 PM, Vaibhav Jain wrote:
> Hi Athira,
> 
> Thanks for reviewing this patch series. My responses to your review
> comment inline below:
> 
> 
> Athira Rajeev <atrajeev@linux.vnet.ibm.com> writes:
> 
>>> On 23 Jan 2025, at 5:37 PM, Vaibhav Jain <vaibhav@linux.ibm.com> wrote:
>>>
>>> Introduce a new PMU named 'kvm-hv' to report Book3s kvm-hv specific
> <snip>
>>
>> Hi Vaibhav
>>
>> All PMU specific code is under “arch/powerpc/perf in the kernel source. Here since we are introducing a kvm-hv specific PMU, can we please have it in arch/powerpc/perf ?
> 
> Is it common convention to put PMU specific code in
> arch/powerpc/perf across ppc achitecture variants ? If its there can you
> please mention the reasons behind it.
> 

My concern is about fragmentation. Would prefer to have
the pmu code under perf folder. Secondly, we did handle
module case for vpa-pmu.

Maddy


> Also the code for this PMU, will be part of kvm-hv kernel module as it
> utilizes the functionality implemented there. Moving this PMU code to
> arch/powerpc/perf will need this to be converted in yet another new
> kernel module, adding a dependency to kvm-hv module and exporting a
> bunch of functionality from kvm-hv. Which looks bit messy to me
> 
> <snip>
>
diff mbox series

Patch

diff --git a/arch/powerpc/include/asm/kvm_book3s.h b/arch/powerpc/include/asm/kvm_book3s.h
index e1ff291ba891..7a7854c65ebb 100644
--- a/arch/powerpc/include/asm/kvm_book3s.h
+++ b/arch/powerpc/include/asm/kvm_book3s.h
@@ -348,6 +348,26 @@  static inline bool kvmhv_is_nestedv1(void)
 
 #endif
 
+/* kvm-ppc pmu registration */
+#if IS_ENABLED(CONFIG_KVM_BOOK3S_64_HV)
+#ifdef CONFIG_PERF_EVENTS
+int kvmppc_register_pmu(void);
+void kvmppc_unregister_pmu(void);
+
+#else
+
+static inline int kvmppc_register_pmu(void)
+{
+	return 0;
+}
+
+static inline void kvmppc_unregister_pmu(void)
+{
+	/* do nothing */
+}
+#endif /* CONFIG_PERF_EVENTS */
+#endif /* CONFIG_KVM_BOOK3S_64_HV */
+
 int __kvmhv_nestedv2_reload_ptregs(struct kvm_vcpu *vcpu, struct pt_regs *regs);
 int __kvmhv_nestedv2_mark_dirty_ptregs(struct kvm_vcpu *vcpu, struct pt_regs *regs);
 int __kvmhv_nestedv2_mark_dirty(struct kvm_vcpu *vcpu, u16 iden);
diff --git a/arch/powerpc/kvm/Makefile b/arch/powerpc/kvm/Makefile
index 4bd9d1230869..7645307ff277 100644
--- a/arch/powerpc/kvm/Makefile
+++ b/arch/powerpc/kvm/Makefile
@@ -92,6 +92,12 @@  kvm-book3s_64-builtin-objs-$(CONFIG_KVM_BOOK3S_64_HANDLER) += \
 	$(kvm-book3s_64-builtin-tm-objs-y) \
 	$(kvm-book3s_64-builtin-xics-objs-y)
 
+# enable kvm_hv perf events
+ifdef CONFIG_PERF_EVENTS
+kvm-book3s_64-builtin-objs-$(CONFIG_KVM_BOOK3S_64_HANDLER) += \
+	book3s_hv_pmu.o
+endif
+
 obj-$(CONFIG_GUEST_STATE_BUFFER_TEST) += test-guest-state-buffer.o
 endif
 
diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index 25429905ae90..6365b8126574 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -6662,6 +6662,14 @@  static int kvmppc_book3s_init_hv(void)
 		return r;
 	}
 
+	r = kvmppc_register_pmu();
+	if (r == -EOPNOTSUPP) {
+		pr_info("KVM-HV: PMU not supported %d\n", r);
+	} else if (r) {
+		pr_err("KVM-HV: Unable to register PMUs %d\n", r);
+		goto err;
+	}
+
 	kvm_ops_hv.owner = THIS_MODULE;
 	kvmppc_hv_ops = &kvm_ops_hv;
 
@@ -6676,6 +6684,7 @@  static int kvmppc_book3s_init_hv(void)
 
 static void kvmppc_book3s_exit_hv(void)
 {
+	kvmppc_unregister_pmu();
 	kvmppc_uvmem_free();
 	kvmppc_free_host_rm_ops();
 	if (kvmppc_radix_possible())
diff --git a/arch/powerpc/kvm/book3s_hv_pmu.c b/arch/powerpc/kvm/book3s_hv_pmu.c
new file mode 100644
index 000000000000..8c6ed30b7654
--- /dev/null
+++ b/arch/powerpc/kvm/book3s_hv_pmu.c
@@ -0,0 +1,133 @@ 
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Description: PMUs specific to running nested KVM-HV guests
+ * on Book3S processors (specifically POWER9 and later).
+ */
+
+#define pr_fmt(fmt)  "kvmppc-pmu: " fmt
+
+#include "asm-generic/local64.h"
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/ratelimit.h>
+#include <linux/kvm_host.h>
+#include <linux/gfp_types.h>
+#include <linux/pgtable.h>
+#include <linux/perf_event.h>
+#include <linux/spinlock_types.h>
+#include <linux/spinlock.h>
+
+#include <asm/types.h>
+#include <asm/kvm_ppc.h>
+#include <asm/kvm_book3s.h>
+#include <asm/mmu.h>
+#include <asm/pgalloc.h>
+#include <asm/pte-walk.h>
+#include <asm/reg.h>
+#include <asm/plpar_wrappers.h>
+#include <asm/firmware.h>
+
+enum kvmppc_pmu_eventid {
+	KVMPPC_EVENT_MAX,
+};
+
+static struct attribute *kvmppc_pmu_events_attr[] = {
+	NULL,
+};
+
+static const struct attribute_group kvmppc_pmu_events_group = {
+	.name = "events",
+	.attrs = kvmppc_pmu_events_attr,
+};
+
+PMU_FORMAT_ATTR(event, "config:0");
+static struct attribute *kvmppc_pmu_format_attr[] = {
+	&format_attr_event.attr,
+	NULL,
+};
+
+static struct attribute_group kvmppc_pmu_format_group = {
+	.name = "format",
+	.attrs = kvmppc_pmu_format_attr,
+};
+
+static const struct attribute_group *kvmppc_pmu_attr_groups[] = {
+	&kvmppc_pmu_events_group,
+	&kvmppc_pmu_format_group,
+	NULL,
+};
+
+static int kvmppc_pmu_event_init(struct perf_event *event)
+{
+	unsigned int config = event->attr.config;
+
+	pr_debug("%s: Event(%p) id=%llu cpu=%x on_cpu=%x config=%u",
+		 __func__, event, event->id, event->cpu,
+		 event->oncpu, config);
+
+	if (event->attr.type != event->pmu->type)
+		return -ENOENT;
+
+	if (config >= KVMPPC_EVENT_MAX)
+		return -EINVAL;
+
+	local64_set(&event->hw.prev_count, 0);
+	local64_set(&event->count, 0);
+
+	return 0;
+}
+
+static void kvmppc_pmu_del(struct perf_event *event, int flags)
+{
+}
+
+static int kvmppc_pmu_add(struct perf_event *event, int flags)
+{
+	return 0;
+}
+
+static void kvmppc_pmu_read(struct perf_event *event)
+{
+}
+
+/* L1 wide counters PMU */
+static struct pmu kvmppc_pmu = {
+	.task_ctx_nr = perf_sw_context,
+	.name = "kvm-hv",
+	.event_init = kvmppc_pmu_event_init,
+	.add = kvmppc_pmu_add,
+	.del = kvmppc_pmu_del,
+	.read = kvmppc_pmu_read,
+	.attr_groups = kvmppc_pmu_attr_groups,
+	.type = -1,
+};
+
+int kvmppc_register_pmu(void)
+{
+	int rc = -EOPNOTSUPP;
+
+	/* only support events for nestedv2 right now */
+	if (kvmhv_is_nestedv2()) {
+		/* Setup done now register the PMU */
+		pr_info("Registering kvm-hv pmu");
+
+		/* Register only if we arent already registered */
+		rc = (kvmppc_pmu.type == -1) ?
+			     perf_pmu_register(&kvmppc_pmu, kvmppc_pmu.name,
+					       -1) : 0;
+	}
+
+	return rc;
+}
+EXPORT_SYMBOL_GPL(kvmppc_register_pmu);
+
+void kvmppc_unregister_pmu(void)
+{
+	if (kvmhv_is_nestedv2()) {
+		if (kvmppc_pmu.type != -1)
+			perf_pmu_unregister(&kvmppc_pmu);
+
+		pr_info("kvmhv_pmu unregistered.\n");
+	}
+}
+EXPORT_SYMBOL_GPL(kvmppc_unregister_pmu);