[PATCH 00/11] Tracefs support for pKVM

[PATCH 00/11] Tracefs support for pKVM

[Vincent Donnefort]

The growing set of features supported by the hypervisor in protected
mode necessitates debugging and profiling tools. Tracefs is the
ideal candidate for this task:

  * It is simple to use and to script.

  * It is supported by various tools, from the trace-cmd CLI to the
    Android web-based perfetto.

  * The ring-buffer, where are stored trace events consists of linked
    pages, making it an ideal structure for sharing between kernel and
    hypervisor.

This series introduces a method to create events and to generate them
from the hypervisor (hyp_enter/hyp_exit given as an example) as well as
a Tracefs user-space interface to read them.

A presentation was given on this matter during the tracing summit in
2022. [1]

1. ring-buffer
--------------

To setup the per-cpu ring-buffers, a new interface is created:

  ring_buffer_remote:	Describes what the kernel needs to know about the
			remote writer, that is, the set of pages forming the
			ring-buffer and a callback for the reader/head
			swapping (enables consuming read)

  ring_buffer_remote():	Creates a read-only ring-buffer from a
			ring_buffer_remote.

To keep the internals of `struct ring_buffer` in sync with the remote,
the meta-page is used. It was originally introduced to enable user-space
mapping of the ring-buffer [1]. In this case, the kernel is not the
producer anymore but the reader. The function to read that meta-page is:

  ring_buffer_poll_remote():
			Update `struct ring_buffer` based on the remote
			meta-page. Wake-up readers if necessary.

The kernel has to poll the meta-page to be notified of newly written
events.

2. Tracefs interface
--------------------

The interface is a hypervisor/ folder at the root of the tracefs mount
point.  This folder is like an instance and you'll find there a subset
of the regular Tracefs user-space interface:

  hypervisor/
     buffer_size_kb
     trace_clock
     trace_pipe
     trace_pipe_raw
     trace
     per_cpu/
             cpuX/
                 trace
                 trace_pipe
                 trace_pipe_raw
     events/
            hypervisor/
                hyp_enter/
                          enable
                          id

Behind the scenes, kvm/hyp_trace.c must rebuild the tracing hierarchy
without relying on kernel/trace/trace.c. This is due to fundamental
differences:

  * Hypervisor tracing doesn't support trace_array's system-specific
    features (snapshots, tracers, etc.).

  * Logged event formats differ (e.g., no PID in hypervisor
    events).

  * Buffer operations require specific hypervisor interactions.

3. Events
---------

In the hypervisor, "hyp events" can be generated with trace_<event_name>
in a similar fashion to what the kernel does. They're also created with
similar macros than the kernel (see kvm_hypevents.h)

HYP_EVENT("foboar",
	HE_PROTO(void),
	HE_STRUCT(),
	HE_ASSIGN(),
	HE_PRINTK(" ")
)

Despite the apparent similarities with TRACE_EVENT(), those macros
internally differs: they must be used in parallel between the hypervisor
(for the writing part) and the kernel (for the reading part) which makes
it difficult to share anything with their kernel counterpart.

Also, events directory isn't using eventfs.

4. Limitations:
---------------

Non-consuming reading of the buffer isn't supported (i.e. cat trace ->
-EPERM) due to current the lack of support in the ring-buffer meta-page.

[1] https://tracingsummit.org/ts/2022/hypervisortracing/
[2] https://lore.kernel.org/all/20240510140435.3550353-1-vdonnefort@google.com/

Changes since v2: https://lore.kernel.org/all/20250108114536.627715-1-vdonnefort@google.com/

  - Fix ring-buffer remote reset
  - Fix fast-forward in rb_page_desc()
  - Refactor nvhe/trace.c
  - struct hyp_buffer_page more compact
  - Add a struct_len to trace_page_desc
  - Extend reset testing
  - Rebase on 6.14-rc3

Changes since v1: https://lore.kernel.org/all/20240911093029.3279154-1-vdonnefort@google.com/

  - Add 128-bits mult fallback in the unlikely event of an overflow. (John)
  - Fix ELF section sort.
  - __always_inline trace_* event macros.
  - Fix events/<event>/enable permissions.
  - Rename ring-buffer "writer" to "remote".
  - Rename CONFIG_PROTECTED_NVHE_TESTING to PKVM_SELFTEST to align with
    Quentin's upcoming selftest
  - Rebase on 6.13-rc3.

Changes since RFC: https://lore.kernel.org/all/20240805173234.3542917-1-vdonnefort@google.com/

  - hypervisor trace clock:
     - mult/shift computed in hyp_trace.c. (John)
     - Update clock when it deviates from kernel boot clock. (John)
     - Add trace_clock file.
     - Separate patch for better readability.
  - Add a proper reset interface which does not need to teardown the
    tracing buffers. (Steven)
  - Return -EPERM on trace access. (Steven)
  - Add per-cpu trace file.
  - Automatically teardown and free the tracing buffer when it is empty,
    without readers and not currently tracing.
  - Show in buffer_size_kb if the buffer is loaded in the hypervisor or
    not.
  - Extend tests to cover reset and unload.
  - CC timekeeping folks on relevant patches (Marc)

Vincent Donnefort (11):
  ring-buffer: Introduce ring-buffer remote
  ring-buffer: Expose buffer_data_page material
  KVM: arm64: Support unaligned fixmap in the nVHE hyp
  KVM: arm64: Add clock support in the nVHE hyp
  KVM: arm64: Add tracing support for the pKVM hyp
  KVM: arm64: Add hyp tracing to tracefs
  KVM: arm64: Add clock for hyp tracefs
  KVM: arm64: Add raw interface for hyp tracefs
  KVM: arm64: Add trace interface for hyp tracefs
  KVM: arm64: Add support for hyp events
  KVM: arm64: Add kselftest for tracefs hyp tracefs

 arch/arm64/include/asm/kvm_asm.h              |   8 +
 arch/arm64/include/asm/kvm_define_hypevents.h |  61 ++
 arch/arm64/include/asm/kvm_hyp.h              |   1 -
 arch/arm64/include/asm/kvm_hypevents.h        |  41 +
 arch/arm64/include/asm/kvm_hypevents_defs.h   |  41 +
 arch/arm64/include/asm/kvm_hyptrace.h         |  37 +
 arch/arm64/kernel/image-vars.h                |   4 +
 arch/arm64/kernel/vmlinux.lds.S               |  18 +
 arch/arm64/kvm/Kconfig                        |  10 +
 arch/arm64/kvm/Makefile                       |   2 +
 arch/arm64/kvm/arm.c                          |   6 +
 arch/arm64/kvm/hyp/hyp-constants.c            |   4 +
 arch/arm64/kvm/hyp/include/nvhe/arm-smccc.h   |  13 +
 arch/arm64/kvm/hyp/include/nvhe/clock.h       |  16 +
 .../kvm/hyp/include/nvhe/define_events.h      |  21 +
 arch/arm64/kvm/hyp/include/nvhe/trace.h       |  60 ++
 arch/arm64/kvm/hyp/nvhe/Makefile              |   1 +
 arch/arm64/kvm/hyp/nvhe/clock.c               |  65 ++
 arch/arm64/kvm/hyp/nvhe/events.c              |  36 +
 arch/arm64/kvm/hyp/nvhe/ffa.c                 |   2 +-
 arch/arm64/kvm/hyp/nvhe/hyp-main.c            |  85 ++
 arch/arm64/kvm/hyp/nvhe/hyp.lds.S             |   6 +
 arch/arm64/kvm/hyp/nvhe/mm.c                  |   2 +-
 arch/arm64/kvm/hyp/nvhe/psci-relay.c          |  14 +-
 arch/arm64/kvm/hyp/nvhe/switch.c              |   5 +-
 arch/arm64/kvm/hyp/nvhe/trace.c               | 628 +++++++++++
 arch/arm64/kvm/hyp_events.c                   | 159 +++
 arch/arm64/kvm/hyp_trace.c                    | 985 ++++++++++++++++++
 arch/arm64/kvm/hyp_trace.h                    |  15 +
 include/linux/ring_buffer.h                   | 100 +-
 kernel/trace/ring_buffer.c                    | 252 ++++-
 tools/testing/selftests/hyp-trace/Makefile    |   6 +
 tools/testing/selftests/hyp-trace/config      |   4 +
 .../selftests/hyp-trace/hyp-trace-test        | 264 +++++
 34 files changed, 2921 insertions(+), 51 deletions(-)
 create mode 100644 arch/arm64/include/asm/kvm_define_hypevents.h
 create mode 100644 arch/arm64/include/asm/kvm_hypevents.h
 create mode 100644 arch/arm64/include/asm/kvm_hypevents_defs.h
 create mode 100644 arch/arm64/include/asm/kvm_hyptrace.h
 create mode 100644 arch/arm64/kvm/hyp/include/nvhe/arm-smccc.h
 create mode 100644 arch/arm64/kvm/hyp/include/nvhe/clock.h
 create mode 100644 arch/arm64/kvm/hyp/include/nvhe/define_events.h
 create mode 100644 arch/arm64/kvm/hyp/include/nvhe/trace.h
 create mode 100644 arch/arm64/kvm/hyp/nvhe/clock.c
 create mode 100644 arch/arm64/kvm/hyp/nvhe/events.c
 create mode 100644 arch/arm64/kvm/hyp/nvhe/trace.c
 create mode 100644 arch/arm64/kvm/hyp_events.c
 create mode 100644 arch/arm64/kvm/hyp_trace.c
 create mode 100644 arch/arm64/kvm/hyp_trace.h
 create mode 100644 tools/testing/selftests/hyp-trace/Makefile
 create mode 100644 tools/testing/selftests/hyp-trace/config
 create mode 100755 tools/testing/selftests/hyp-trace/hyp-trace-test


base-commit: 0ad2507d5d93f39619fc42372c347d6006b64319
-- 
2.48.1.601.g30ceb7b040-goog

[PATCH 01/11] ring-buffer: Introduce ring-buffer remote

[Vincent Donnefort]

A ring-buffer remote is an entity outside of the kernel (most likely a
firmware or a hypervisor) capable of writing events in a ring-buffer
following the same format as the tracefs ring-buffer.

To setup the ring-buffer on the kernel side, a description of the pages
(struct trace_page_desc) is necessary. A callback (get_reader_page) must
also be provided. It is called whenever it is done reading the previous
reader page.

It is expected from the remote to keep the meta-page updated.

Signed-off-by: Vincent Donnefort <vdonnefort@google.com>

diff --git a/include/linux/ring_buffer.h b/include/linux/ring_buffer.h
index 17fbb7855295..2a1330a65edb 100644
--- a/include/linux/ring_buffer.h
+++ b/include/linux/ring_buffer.h
@@ -248,4 +248,59 @@ int ring_buffer_map(struct trace_buffer *buffer, int cpu,
 		    struct vm_area_struct *vma);
 int ring_buffer_unmap(struct trace_buffer *buffer, int cpu);
 int ring_buffer_map_get_reader(struct trace_buffer *buffer, int cpu);
+
+#define meta_pages_lost(__meta) \
+	((__meta)->Reserved1)
+#define meta_pages_touched(__meta) \
+	((__meta)->Reserved2)
+
+struct rb_page_desc {
+	unsigned int	cpu;
+	unsigned int	nr_page_va; /* excludes the meta page */
+	unsigned long	meta_va;
+	unsigned long	page_va[];
+};
+
+struct trace_page_desc {
+	size_t		struct_len;
+	unsigned int	nr_cpus;
+	char		__data[]; /* list of rb_page_desc */
+};
+
+static inline
+struct rb_page_desc *__next_rb_page_desc(struct rb_page_desc *pdesc)
+{
+	size_t len = struct_size(pdesc, page_va, pdesc->nr_page_va);
+
+	return (struct rb_page_desc *)((void *)pdesc + len);
+}
+
+static inline
+struct rb_page_desc *__first_rb_page_desc(struct trace_page_desc *trace_pdesc)
+{
+	return (struct rb_page_desc *)(&trace_pdesc->__data[0]);
+}
+
+#define for_each_rb_page_desc(__pdesc, __cpu, __trace_pdesc)		\
+	for (__pdesc = __first_rb_page_desc(__trace_pdesc), __cpu = 0;	\
+	     __cpu < (__trace_pdesc)->nr_cpus;				\
+	     __cpu++, __pdesc = __next_rb_page_desc(__pdesc))
+
+struct ring_buffer_remote {
+	struct trace_page_desc	*pdesc;
+	int			(*get_reader_page)(int cpu);
+	int			(*reset)(int cpu);
+};
+
+int ring_buffer_poll_remote(struct trace_buffer *buffer, int cpu);
+
+struct trace_buffer *
+__ring_buffer_alloc_remote(struct ring_buffer_remote *remote,
+			   struct lock_class_key *key);
+
+#define ring_buffer_remote(remote)				\
+({								\
+	static struct lock_class_key __key;			\
+	__ring_buffer_alloc_remote(remote, &__key);		\
+})
 #endif /* _LINUX_RING_BUFFER_H */
diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index bb6089c2951e..c27516a384a8 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -519,6 +519,8 @@ struct ring_buffer_per_cpu {
 	struct trace_buffer_meta	*meta_page;
 	struct ring_buffer_meta		*ring_meta;
 
+	struct ring_buffer_remote	*remote;
+
 	/* ring buffer pages to update, > 0 to add, < 0 to remove */
 	long				nr_pages_to_update;
 	struct list_head		new_pages; /* new pages to add */
@@ -541,6 +543,8 @@ struct trace_buffer {
 
 	struct ring_buffer_per_cpu	**buffers;
 
+	struct ring_buffer_remote	*remote;
+
 	struct hlist_node		node;
 	u64				(*clock)(void);
 
@@ -2155,6 +2159,41 @@ static int __rb_allocate_pages(struct ring_buffer_per_cpu *cpu_buffer,
 	return -ENOMEM;
 }
 
+static struct rb_page_desc *rb_page_desc(struct trace_page_desc *trace_pdesc, int cpu)
+{
+	struct rb_page_desc *pdesc, *end;
+	size_t len;
+	int i;
+
+	if (!trace_pdesc)
+		return NULL;
+
+	if (cpu >= trace_pdesc->nr_cpus)
+		return NULL;
+
+	end = (struct rb_page_desc *)((void *)trace_pdesc + trace_pdesc->struct_len);
+	pdesc = __first_rb_page_desc(trace_pdesc);
+	len = struct_size(pdesc, page_va, pdesc->nr_page_va);
+	pdesc = (struct rb_page_desc *)((void *)pdesc + (len * cpu));
+
+	if (pdesc < end && pdesc->cpu == cpu)
+		return pdesc;
+
+	/* Missing CPUs, need to linear search */
+	for_each_rb_page_desc(pdesc, i, trace_pdesc) {
+		if (pdesc->cpu == cpu)
+			return pdesc;
+	}
+
+	return NULL;
+}
+
+static void *rb_page_desc_page(struct rb_page_desc *pdesc, int page_id)
+{
+	return page_id > pdesc->nr_page_va ? NULL : (void *)pdesc->page_va[page_id];
+}
+
+
 static int rb_allocate_pages(struct ring_buffer_per_cpu *cpu_buffer,
 			     unsigned long nr_pages)
 {
@@ -2215,6 +2254,31 @@ rb_allocate_cpu_buffer(struct trace_buffer *buffer, long nr_pages, int cpu)
 
 	cpu_buffer->reader_page = bpage;
 
+	if (buffer->remote) {
+		struct rb_page_desc *pdesc = rb_page_desc(buffer->remote->pdesc, cpu);
+
+		if (!pdesc)
+			goto fail_free_reader;
+
+		cpu_buffer->remote = buffer->remote;
+		cpu_buffer->meta_page = (struct trace_buffer_meta *)(void *)pdesc->meta_va;
+		cpu_buffer->subbuf_ids = pdesc->page_va;
+		cpu_buffer->nr_pages = pdesc->nr_page_va - 1;
+		atomic_inc(&cpu_buffer->record_disabled);
+		atomic_inc(&cpu_buffer->resize_disabled);
+
+		bpage->page = rb_page_desc_page(pdesc,
+						cpu_buffer->meta_page->reader.id);
+		if (!bpage->page)
+			goto fail_free_reader;
+		/*
+		 * The meta-page can only describe which of the ring-buffer page
+		 * is the reader. There is no need to init the rest of the
+		 * ring-buffer.
+		 */
+		return cpu_buffer;
+	}
+
 	if (buffer->range_addr_start) {
 		/*
 		 * Range mapped buffers have the same restrictions as memory
@@ -2292,6 +2356,10 @@ static void rb_free_cpu_buffer(struct ring_buffer_per_cpu *cpu_buffer)
 
 	irq_work_sync(&cpu_buffer->irq_work.work);
 
+	/* remote ring-buffer. We do not own the data pages */
+	if (cpu_buffer->remote)
+		cpu_buffer->reader_page->page = NULL;
+
 	free_buffer_page(cpu_buffer->reader_page);
 
 	if (head) {
@@ -2313,7 +2381,8 @@ static void rb_free_cpu_buffer(struct ring_buffer_per_cpu *cpu_buffer)
 static struct trace_buffer *alloc_buffer(unsigned long size, unsigned flags,
 					 int order, unsigned long start,
 					 unsigned long end,
-					 struct lock_class_key *key)
+					 struct lock_class_key *key,
+					 struct ring_buffer_remote *remote)
 {
 	struct trace_buffer *buffer;
 	long nr_pages;
@@ -2341,6 +2410,11 @@ static struct trace_buffer *alloc_buffer(unsigned long size, unsigned flags,
 	buffer->flags = flags;
 	buffer->clock = trace_clock_local;
 	buffer->reader_lock_key = key;
+	if (remote) {
+		buffer->remote = remote;
+		/* The writer is remote. This ring-buffer is read-only */
+		atomic_inc(&buffer->record_disabled);
+	}
 
 	init_irq_work(&buffer->irq_work.work, rb_wake_up_waiters);
 	init_waitqueue_head(&buffer->irq_work.waiters);
@@ -2447,7 +2521,7 @@ struct trace_buffer *__ring_buffer_alloc(unsigned long size, unsigned flags,
 					struct lock_class_key *key)
 {
 	/* Default buffer page size - one system page */
-	return alloc_buffer(size, flags, 0, 0, 0,key);
+	return alloc_buffer(size, flags, 0, 0, 0, key, NULL);
 
 }
 EXPORT_SYMBOL_GPL(__ring_buffer_alloc);
@@ -2471,7 +2545,18 @@ struct trace_buffer *__ring_buffer_alloc_range(unsigned long size, unsigned flag
 					       unsigned long range_size,
 					       struct lock_class_key *key)
 {
-	return alloc_buffer(size, flags, order, start, start + range_size, key);
+	return alloc_buffer(size, flags, order, start, start + range_size, key, NULL);
+}
+
+/**
+ * __ring_buffer_alloc_remote - allocate a new ring_buffer from a remote
+ * @remote: Contains a description of the ring-buffer pages and remote callbacks.
+ * @key: ring buffer reader_lock_key.
+ */
+struct trace_buffer *__ring_buffer_alloc_remote(struct ring_buffer_remote *remote,
+						struct lock_class_key *key)
+{
+	return alloc_buffer(0, 0, 0, 0, 0, key, remote);
 }
 
 /**
@@ -5225,8 +5310,56 @@ rb_update_iter_read_stamp(struct ring_buffer_iter *iter,
 	}
 }
 
+static bool rb_read_remote_meta_page(struct ring_buffer_per_cpu *cpu_buffer)
+{
+	local_set(&cpu_buffer->entries, READ_ONCE(cpu_buffer->meta_page->entries));
+	local_set(&cpu_buffer->overrun, READ_ONCE(cpu_buffer->meta_page->overrun));
+	local_set(&cpu_buffer->pages_touched, READ_ONCE(meta_pages_touched(cpu_buffer->meta_page)));
+	local_set(&cpu_buffer->pages_lost, READ_ONCE(meta_pages_lost(cpu_buffer->meta_page)));
+	/*
+	 * No need to get the "read" field, it can be tracked here as any
+	 * reader will have to go through a rign_buffer_per_cpu.
+	 */
+
+	return rb_num_of_entries(cpu_buffer);
+}
+
 static struct buffer_page *
-rb_get_reader_page(struct ring_buffer_per_cpu *cpu_buffer)
+__rb_get_reader_page_from_remote(struct ring_buffer_per_cpu *cpu_buffer)
+{
+	u32 prev_reader;
+
+	if (!rb_read_remote_meta_page(cpu_buffer))
+		return NULL;
+
+	/* More to read on the reader page */
+	if (cpu_buffer->reader_page->read < rb_page_size(cpu_buffer->reader_page)) {
+		if (!cpu_buffer->reader_page->read)
+			cpu_buffer->read_stamp = cpu_buffer->reader_page->page->time_stamp;
+		return cpu_buffer->reader_page;
+	}
+
+	prev_reader = cpu_buffer->meta_page->reader.id;
+
+	WARN_ON(cpu_buffer->remote->get_reader_page(cpu_buffer->cpu));
+	/* nr_pages doesn't include the reader page */
+	if (WARN_ON(cpu_buffer->meta_page->reader.id > cpu_buffer->nr_pages))
+		return NULL;
+
+	cpu_buffer->reader_page->page =
+		(void *)cpu_buffer->subbuf_ids[cpu_buffer->meta_page->reader.id];
+	cpu_buffer->reader_page->id = cpu_buffer->meta_page->reader.id;
+	cpu_buffer->reader_page->read = 0;
+	cpu_buffer->read_stamp = cpu_buffer->reader_page->page->time_stamp;
+	cpu_buffer->lost_events = cpu_buffer->meta_page->reader.lost_events;
+
+	WARN_ON(prev_reader == cpu_buffer->meta_page->reader.id);
+
+	return rb_page_size(cpu_buffer->reader_page) ? cpu_buffer->reader_page : NULL;
+}
+
+static struct buffer_page *
+__rb_get_reader_page(struct ring_buffer_per_cpu *cpu_buffer)
 {
 	struct buffer_page *reader = NULL;
 	unsigned long bsize = READ_ONCE(cpu_buffer->buffer->subbuf_size);
@@ -5397,6 +5530,13 @@ rb_get_reader_page(struct ring_buffer_per_cpu *cpu_buffer)
 	return reader;
 }
 
+static struct buffer_page *
+rb_get_reader_page(struct ring_buffer_per_cpu *cpu_buffer)
+{
+	return cpu_buffer->remote ? __rb_get_reader_page_from_remote(cpu_buffer) :
+				    __rb_get_reader_page(cpu_buffer);
+}
+
 static void rb_advance_reader(struct ring_buffer_per_cpu *cpu_buffer)
 {
 	struct ring_buffer_event *event;
@@ -5801,7 +5941,7 @@ ring_buffer_read_prepare(struct trace_buffer *buffer, int cpu, gfp_t flags)
 	struct ring_buffer_per_cpu *cpu_buffer;
 	struct ring_buffer_iter *iter;
 
-	if (!cpumask_test_cpu(cpu, buffer->cpumask))
+	if (!cpumask_test_cpu(cpu, buffer->cpumask) || buffer->remote)
 		return NULL;
 
 	iter = kzalloc(sizeof(*iter), flags);
@@ -5971,6 +6111,23 @@ rb_reset_cpu(struct ring_buffer_per_cpu *cpu_buffer)
 {
 	struct buffer_page *page;
 
+	if (cpu_buffer->remote) {
+		if (!cpu_buffer->remote->reset)
+			return;
+
+		cpu_buffer->remote->reset(cpu_buffer->cpu);
+		rb_read_remote_meta_page(cpu_buffer);
+
+		/* Read related values, not covered by the meta-page */
+		local_set(&cpu_buffer->pages_read, 0);
+		cpu_buffer->read = 0;
+		cpu_buffer->read_bytes = 0;
+		cpu_buffer->last_overrun = 0;
+		cpu_buffer->reader_page->read = 0;
+
+		return;
+	}
+
 	rb_head_page_deactivate(cpu_buffer);
 
 	cpu_buffer->head_page
@@ -6218,6 +6375,49 @@ bool ring_buffer_empty_cpu(struct trace_buffer *buffer, int cpu)
 }
 EXPORT_SYMBOL_GPL(ring_buffer_empty_cpu);
 
+int ring_buffer_poll_remote(struct trace_buffer *buffer, int cpu)
+{
+	struct ring_buffer_per_cpu *cpu_buffer;
+	unsigned long flags;
+
+	if (cpu != RING_BUFFER_ALL_CPUS) {
+		if (!cpumask_test_cpu(cpu, buffer->cpumask))
+			return -EINVAL;
+
+		cpu_buffer = buffer->buffers[cpu];
+
+		raw_spin_lock_irqsave(&cpu_buffer->reader_lock, flags);
+		if (rb_read_remote_meta_page(cpu_buffer))
+			rb_wakeups(buffer, cpu_buffer);
+		raw_spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags);
+
+		return 0;
+	}
+
+	/*
+	 * Make sure all the ring buffers are up to date before we start reading
+	 * them.
+	 */
+	for_each_buffer_cpu(buffer, cpu) {
+		cpu_buffer = buffer->buffers[cpu];
+
+		raw_spin_lock_irqsave(&cpu_buffer->reader_lock, flags);
+		rb_read_remote_meta_page(buffer->buffers[cpu]);
+		raw_spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags);
+	}
+
+	for_each_buffer_cpu(buffer, cpu) {
+		cpu_buffer = buffer->buffers[cpu];
+
+		raw_spin_lock_irqsave(&cpu_buffer->reader_lock, flags);
+		if (rb_num_of_entries(cpu_buffer))
+			rb_wakeups(buffer, buffer->buffers[cpu]);
+		raw_spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags);
+	}
+
+	return 0;
+}
+
 #ifdef CONFIG_RING_BUFFER_ALLOW_SWAP
 /**
  * ring_buffer_swap_cpu - swap a CPU buffer between two ring buffers
@@ -6469,6 +6669,7 @@ int ring_buffer_read_page(struct trace_buffer *buffer,
 	unsigned int commit;
 	unsigned int read;
 	u64 save_timestamp;
+	bool force_memcpy;
 	int ret = -1;
 
 	if (!cpumask_test_cpu(cpu, buffer->cpumask))
@@ -6506,6 +6707,8 @@ int ring_buffer_read_page(struct trace_buffer *buffer,
 	/* Check if any events were dropped */
 	missed_events = cpu_buffer->lost_events;
 
+	force_memcpy = cpu_buffer->mapped || cpu_buffer->remote;
+
 	/*
 	 * If this page has been partially read or
 	 * if len is not big enough to read the rest of the page or
@@ -6515,7 +6718,7 @@ int ring_buffer_read_page(struct trace_buffer *buffer,
 	 */
 	if (read || (len < (commit - read)) ||
 	    cpu_buffer->reader_page == cpu_buffer->commit_page ||
-	    cpu_buffer->mapped) {
+	    force_memcpy) {
 		struct buffer_data_page *rpage = cpu_buffer->reader_page->page;
 		unsigned int rpos = read;
 		unsigned int pos = 0;
@@ -7097,7 +7300,7 @@ int ring_buffer_map(struct trace_buffer *buffer, int cpu,
 	unsigned long flags, *subbuf_ids;
 	int err = 0;
 
-	if (!cpumask_test_cpu(cpu, buffer->cpumask))
+	if (!cpumask_test_cpu(cpu, buffer->cpumask) || buffer->remote)
 		return -EINVAL;
 
 	cpu_buffer = buffer->buffers[cpu];
-- 
2.48.1.601.g30ceb7b040-goog

[PATCH 02/11] ring-buffer: Expose buffer_data_page material

[Vincent Donnefort]

In preparation for allowing the write of ring-buffer compliant pages
outside of ring_buffer.c, move to the header, struct buffer_data_page and
timestamp encoding functions into the publicly available ring_buffer.h.

Signed-off-by: Vincent Donnefort <vdonnefort@google.com>

diff --git a/include/linux/ring_buffer.h b/include/linux/ring_buffer.h
index 2a1330a65edb..75cd0cb46768 100644
--- a/include/linux/ring_buffer.h
+++ b/include/linux/ring_buffer.h
@@ -3,8 +3,10 @@
 #define _LINUX_RING_BUFFER_H
 
 #include <linux/mm.h>
-#include <linux/seq_file.h>
 #include <linux/poll.h>
+#include <linux/seq_file.h>
+
+#include <asm/local.h>
 
 #include <uapi/linux/trace_mmap.h>
 
@@ -20,6 +22,8 @@ struct ring_buffer_event {
 	u32		array[];
 };
 
+#define RB_EVNT_HDR_SIZE (offsetof(struct ring_buffer_event, array))
+
 /**
  * enum ring_buffer_type - internal ring buffer types
  *
@@ -61,11 +65,50 @@ enum ring_buffer_type {
 	RINGBUF_TYPE_TIME_STAMP,
 };
 
+#define TS_SHIFT        27
+#define TS_MASK         ((1ULL << TS_SHIFT) - 1)
+#define TS_DELTA_TEST   (~TS_MASK)
+
+/*
+ * We need to fit the time_stamp delta into 27 bits.
+ */
+static inline bool test_time_stamp(u64 delta)
+{
+	return !!(delta & TS_DELTA_TEST);
+}
+
 unsigned ring_buffer_event_length(struct ring_buffer_event *event);
 void *ring_buffer_event_data(struct ring_buffer_event *event);
 u64 ring_buffer_event_time_stamp(struct trace_buffer *buffer,
 				 struct ring_buffer_event *event);
 
+#define BUF_PAGE_HDR_SIZE offsetof(struct buffer_data_page, data)
+
+/* Max payload is BUF_PAGE_SIZE - header (8bytes) */
+#define BUF_MAX_DATA_SIZE (BUF_PAGE_SIZE - (sizeof(u32) * 2))
+
+#define BUF_PAGE_SIZE (PAGE_SIZE - BUF_PAGE_HDR_SIZE)
+
+#define RB_ALIGNMENT		4U
+#define RB_MAX_SMALL_DATA	(RB_ALIGNMENT * RINGBUF_TYPE_DATA_TYPE_LEN_MAX)
+#define RB_EVNT_MIN_SIZE	8U	/* two 32bit words */
+
+#ifndef CONFIG_HAVE_64BIT_ALIGNED_ACCESS
+# define RB_FORCE_8BYTE_ALIGNMENT	0
+# define RB_ARCH_ALIGNMENT		RB_ALIGNMENT
+#else
+# define RB_FORCE_8BYTE_ALIGNMENT	1
+# define RB_ARCH_ALIGNMENT		8U
+#endif
+
+#define RB_ALIGN_DATA		__aligned(RB_ARCH_ALIGNMENT)
+
+struct buffer_data_page {
+	u64		 time_stamp;	/* page time stamp */
+	local_t		 commit;	/* write committed index */
+	unsigned char	 data[] RB_ALIGN_DATA;	/* data of buffer page */
+};
+
 /*
  * ring_buffer_discard_commit will remove an event that has not
  *   been committed yet. If this is used, then ring_buffer_unlock_commit
diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index c27516a384a8..e70f39e0adb1 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -152,23 +152,6 @@ int ring_buffer_print_entry_header(struct trace_seq *s)
 /* Used for individual buffers (after the counter) */
 #define RB_BUFFER_OFF		(1 << 20)
 
-#define BUF_PAGE_HDR_SIZE offsetof(struct buffer_data_page, data)
-
-#define RB_EVNT_HDR_SIZE (offsetof(struct ring_buffer_event, array))
-#define RB_ALIGNMENT		4U
-#define RB_MAX_SMALL_DATA	(RB_ALIGNMENT * RINGBUF_TYPE_DATA_TYPE_LEN_MAX)
-#define RB_EVNT_MIN_SIZE	8U	/* two 32bit words */
-
-#ifndef CONFIG_HAVE_64BIT_ALIGNED_ACCESS
-# define RB_FORCE_8BYTE_ALIGNMENT	0
-# define RB_ARCH_ALIGNMENT		RB_ALIGNMENT
-#else
-# define RB_FORCE_8BYTE_ALIGNMENT	1
-# define RB_ARCH_ALIGNMENT		8U
-#endif
-
-#define RB_ALIGN_DATA		__aligned(RB_ARCH_ALIGNMENT)
-
 /* define RINGBUF_TYPE_DATA for 'case RINGBUF_TYPE_DATA:' */
 #define RINGBUF_TYPE_DATA 0 ... RINGBUF_TYPE_DATA_TYPE_LEN_MAX
 
@@ -311,10 +294,6 @@ EXPORT_SYMBOL_GPL(ring_buffer_event_data);
 #define for_each_online_buffer_cpu(buffer, cpu)		\
 	for_each_cpu_and(cpu, buffer->cpumask, cpu_online_mask)
 
-#define TS_SHIFT	27
-#define TS_MASK		((1ULL << TS_SHIFT) - 1)
-#define TS_DELTA_TEST	(~TS_MASK)
-
 static u64 rb_event_time_stamp(struct ring_buffer_event *event)
 {
 	u64 ts;
@@ -333,12 +312,6 @@ static u64 rb_event_time_stamp(struct ring_buffer_event *event)
 
 #define RB_MISSED_MASK		(3 << 30)
 
-struct buffer_data_page {
-	u64		 time_stamp;	/* page time stamp */
-	local_t		 commit;	/* write committed index */
-	unsigned char	 data[] RB_ALIGN_DATA;	/* data of buffer page */
-};
-
 struct buffer_data_read_page {
 	unsigned		order;	/* order of the page */
 	struct buffer_data_page	*data;	/* actual data, stored in this page */
@@ -397,14 +370,6 @@ static void free_buffer_page(struct buffer_page *bpage)
 	kfree(bpage);
 }
 
-/*
- * We need to fit the time_stamp delta into 27 bits.
- */
-static inline bool test_time_stamp(u64 delta)
-{
-	return !!(delta & TS_DELTA_TEST);
-}
-
 struct rb_irq_work {
 	struct irq_work			work;
 	wait_queue_head_t		waiters;
-- 
2.48.1.601.g30ceb7b040-goog

[PATCH 03/11] KVM: arm64: Support unaligned fixmap in the nVHE hyp

[Vincent Donnefort]

Return the fixmap VA with the page offset, instead of the page base
address.

Signed-off-by: Vincent Donnefort <vdonnefort@google.com>

diff --git a/arch/arm64/kvm/hyp/nvhe/mm.c b/arch/arm64/kvm/hyp/nvhe/mm.c
index f41c7440b34b..720cc3b36596 100644
--- a/arch/arm64/kvm/hyp/nvhe/mm.c
+++ b/arch/arm64/kvm/hyp/nvhe/mm.c
@@ -240,7 +240,7 @@ void *hyp_fixmap_map(phys_addr_t phys)
 	WRITE_ONCE(*ptep, pte);
 	dsb(ishst);
 
-	return (void *)slot->addr;
+	return (void *)slot->addr + offset_in_page(phys);
 }
 
 static void fixmap_clear_slot(struct hyp_fixmap_slot *slot)
-- 
2.48.1.601.g30ceb7b040-goog

[PATCH 04/11] KVM: arm64: Add clock support in the nVHE hyp

[Vincent Donnefort]

By default, the arm64 host kernel is using the arch timer as a source
for sched_clock. Conveniently, EL2 has access to that same counter,
allowing to generate clock values that are synchronized.

The clock needs nonetheless to be setup with the same slope values as
the kernel. Introducing at the same time trace_clock() which is expected
to be later configured by the hypervisor tracing.

Signed-off-by: Vincent Donnefort <vdonnefort@google.com>

diff --git a/arch/arm64/include/asm/kvm_hyp.h b/arch/arm64/include/asm/kvm_hyp.h
index c838309e4ec4..355bae0056f0 100644
--- a/arch/arm64/include/asm/kvm_hyp.h
+++ b/arch/arm64/include/asm/kvm_hyp.h
@@ -144,5 +144,4 @@ extern u64 kvm_nvhe_sym(id_aa64smfr0_el1_sys_val);
 extern unsigned long kvm_nvhe_sym(__icache_flags);
 extern unsigned int kvm_nvhe_sym(kvm_arm_vmid_bits);
 extern unsigned int kvm_nvhe_sym(kvm_host_sve_max_vl);
-
 #endif /* __ARM64_KVM_HYP_H__ */
diff --git a/arch/arm64/kvm/hyp/include/nvhe/clock.h b/arch/arm64/kvm/hyp/include/nvhe/clock.h
new file mode 100644
index 000000000000..2bd05b3b89f9
--- /dev/null
+++ b/arch/arm64/kvm/hyp/include/nvhe/clock.h
@@ -0,0 +1,16 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __ARM64_KVM_HYP_NVHE_CLOCK_H
+#define __ARM64_KVM_HYP_NVHE_CLOCK_H
+#include <linux/types.h>
+
+#include <asm/kvm_hyp.h>
+
+#ifdef CONFIG_TRACING
+void trace_clock_update(u32 mult, u32 shift, u64 epoch_ns, u64 epoch_cyc);
+u64 trace_clock(void);
+#else
+static inline void
+trace_clock_update(u32 mult, u32 shift, u64 epoch_ns, u64 epoch_cyc) { }
+static inline u64 trace_clock(void) { return 0; }
+#endif
+#endif
diff --git a/arch/arm64/kvm/hyp/nvhe/Makefile b/arch/arm64/kvm/hyp/nvhe/Makefile
index b43426a493df..323e992089bd 100644
--- a/arch/arm64/kvm/hyp/nvhe/Makefile
+++ b/arch/arm64/kvm/hyp/nvhe/Makefile
@@ -28,6 +28,7 @@ hyp-obj-y := timer-sr.o sysreg-sr.o debug-sr.o switch.o tlb.o hyp-init.o host.o
 hyp-obj-y += ../vgic-v3-sr.o ../aarch32.o ../vgic-v2-cpuif-proxy.o ../entry.o \
 	 ../fpsimd.o ../hyp-entry.o ../exception.o ../pgtable.o
 hyp-obj-$(CONFIG_LIST_HARDENED) += list_debug.o
+hyp-obj-$(CONFIG_TRACING) += clock.o
 hyp-obj-y += $(lib-objs)
 
 ##
diff --git a/arch/arm64/kvm/hyp/nvhe/clock.c b/arch/arm64/kvm/hyp/nvhe/clock.c
new file mode 100644
index 000000000000..879c6b09d9ca
--- /dev/null
+++ b/arch/arm64/kvm/hyp/nvhe/clock.c
@@ -0,0 +1,65 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2025 Google LLC
+ * Author: Vincent Donnefort <vdonnefort@google.com>
+ */
+
+#include <nvhe/clock.h>
+
+#include <asm/arch_timer.h>
+#include <asm/div64.h>
+
+static struct clock_data {
+	struct {
+		u32 mult;
+		u32 shift;
+		u64 epoch_ns;
+		u64 epoch_cyc;
+		u64 cyc_overflow64;
+	} data[2];
+	u64 cur;
+} trace_clock_data;
+
+static u64 __clock_mult_uint128(u64 cyc, u32 mult, u32 shift)
+{
+	__uint128_t ns = (__uint128_t)cyc * mult;
+
+	ns >>= shift;
+
+	return (u64)ns;
+}
+
+/* Does not guarantee no reader on the modified bank. */
+void trace_clock_update(u32 mult, u32 shift, u64 epoch_ns, u64 epoch_cyc)
+{
+	struct clock_data *clock = &trace_clock_data;
+	u64 bank = clock->cur ^ 1;
+
+	clock->data[bank].mult			= mult;
+	clock->data[bank].shift			= shift;
+	clock->data[bank].epoch_ns		= epoch_ns;
+	clock->data[bank].epoch_cyc		= epoch_cyc;
+	clock->data[bank].cyc_overflow64	= ULONG_MAX / mult;
+
+	smp_store_release(&clock->cur, bank);
+}
+
+/* Using host provided data. Do not use for anything else than debugging. */
+u64 trace_clock(void)
+{
+	struct clock_data *clock = &trace_clock_data;
+	u64 bank = smp_load_acquire(&clock->cur);
+	u64 cyc, ns;
+
+	cyc = __arch_counter_get_cntpct() - clock->data[bank].epoch_cyc;
+
+	if (likely(cyc < clock->data[bank].cyc_overflow64)) {
+		ns = cyc * clock->data[bank].mult;
+		ns >>= clock->data[bank].shift;
+	} else {
+		ns = __clock_mult_uint128(cyc, clock->data[bank].mult,
+					   clock->data[bank].shift);
+	}
+
+	return (u64)ns + clock->data[bank].epoch_ns;
+}
-- 
2.48.1.601.g30ceb7b040-goog

[PATCH 05/11] KVM: arm64: Add tracing support for the pKVM hyp

[Vincent Donnefort]

When running with protected mode, the host has very little knowledge
about what is happening in the hypervisor. Of course this is an
essential feature for security but nonetheless, that piece of code
growing with more responsibilities, we need now a way to debug and
profile it. Tracefs by its reliatility, versatility and support for
user-space is the perfect tool.

There's no way the hypervisor could log events directly into the host
tracefs ring-buffers. So instead let's use our own, where the hypervisor
is the writer and the host the reader.

Signed-off-by: Vincent Donnefort <vdonnefort@google.com>

diff --git a/arch/arm64/include/asm/kvm_asm.h b/arch/arm64/include/asm/kvm_asm.h
index bec227f9500a..b5893e0afe8e 100644
--- a/arch/arm64/include/asm/kvm_asm.h
+++ b/arch/arm64/include/asm/kvm_asm.h
@@ -87,6 +87,10 @@ enum __kvm_host_smccc_func {
 	__KVM_HOST_SMCCC_FUNC___pkvm_vcpu_load,
 	__KVM_HOST_SMCCC_FUNC___pkvm_vcpu_put,
 	__KVM_HOST_SMCCC_FUNC___pkvm_tlb_flush_vmid,
+	__KVM_HOST_SMCCC_FUNC___pkvm_load_tracing,
+	__KVM_HOST_SMCCC_FUNC___pkvm_teardown_tracing,
+	__KVM_HOST_SMCCC_FUNC___pkvm_enable_tracing,
+	__KVM_HOST_SMCCC_FUNC___pkvm_swap_reader_tracing,
 };
 
 #define DECLARE_KVM_VHE_SYM(sym)	extern char sym[]
diff --git a/arch/arm64/include/asm/kvm_hyptrace.h b/arch/arm64/include/asm/kvm_hyptrace.h
new file mode 100644
index 000000000000..7da6a248c7fa
--- /dev/null
+++ b/arch/arm64/include/asm/kvm_hyptrace.h
@@ -0,0 +1,21 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+#ifndef __ARM64_KVM_HYPTRACE_H_
+#define __ARM64_KVM_HYPTRACE_H_
+#include <asm/kvm_hyp.h>
+
+#include <linux/ring_buffer.h>
+
+/*
+ * Host donations to the hypervisor to store the struct hyp_buffer_page.
+ */
+struct hyp_buffer_pages_backing {
+	unsigned long start;
+	size_t size;
+};
+
+struct hyp_trace_desc {
+	struct hyp_buffer_pages_backing	backing;
+	struct trace_page_desc		page_desc;
+
+};
+#endif
diff --git a/arch/arm64/kvm/hyp/include/nvhe/trace.h b/arch/arm64/kvm/hyp/include/nvhe/trace.h
new file mode 100644
index 000000000000..bf74a6ee322d
--- /dev/null
+++ b/arch/arm64/kvm/hyp/include/nvhe/trace.h
@@ -0,0 +1,32 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+#ifndef __ARM64_KVM_HYP_NVHE_TRACE_H
+#define __ARM64_KVM_HYP_NVHE_TRACE_H
+#include <asm/kvm_hyptrace.h>
+
+/* Internal struct exported for hyp-constants.c */
+struct hyp_buffer_page {
+	struct list_head	list;
+	struct buffer_data_page	*page;
+	u64			entries;
+	u32			write;
+	u32			id;
+};
+
+#ifdef CONFIG_TRACING
+void *tracing_reserve_entry(unsigned long length);
+void tracing_commit_entry(void);
+
+int __pkvm_load_tracing(unsigned long desc_va, size_t desc_size);
+void __pkvm_teardown_tracing(void);
+int __pkvm_enable_tracing(bool enable);
+int __pkvm_swap_reader_tracing(unsigned int cpu);
+#else
+static inline void *tracing_reserve_entry(unsigned long length) { return NULL; }
+static inline void tracing_commit_entry(void) { }
+
+static inline int __pkvm_load_tracing(unsigned long desc_va, size_t desc_size) { return -ENODEV; }
+static inline void __pkvm_teardown_tracing(void) { }
+static inline int __pkvm_enable_tracing(bool enable) { return -ENODEV; }
+static inline int __pkvm_swap_reader_tracing(unsigned int cpu) { return -ENODEV; }
+#endif
+#endif
diff --git a/arch/arm64/kvm/hyp/nvhe/Makefile b/arch/arm64/kvm/hyp/nvhe/Makefile
index 323e992089bd..40f243c44cf5 100644
--- a/arch/arm64/kvm/hyp/nvhe/Makefile
+++ b/arch/arm64/kvm/hyp/nvhe/Makefile
@@ -28,7 +28,7 @@ hyp-obj-y := timer-sr.o sysreg-sr.o debug-sr.o switch.o tlb.o hyp-init.o host.o
 hyp-obj-y += ../vgic-v3-sr.o ../aarch32.o ../vgic-v2-cpuif-proxy.o ../entry.o \
 	 ../fpsimd.o ../hyp-entry.o ../exception.o ../pgtable.o
 hyp-obj-$(CONFIG_LIST_HARDENED) += list_debug.o
-hyp-obj-$(CONFIG_TRACING) += clock.o
+hyp-obj-$(CONFIG_TRACING) += clock.o trace.o
 hyp-obj-y += $(lib-objs)
 
 ##
diff --git a/arch/arm64/kvm/hyp/nvhe/hyp-main.c b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
index 2c37680d954c..ced0a161d56e 100644
--- a/arch/arm64/kvm/hyp/nvhe/hyp-main.c
+++ b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
@@ -18,6 +18,7 @@
 #include <nvhe/mem_protect.h>
 #include <nvhe/mm.h>
 #include <nvhe/pkvm.h>
+#include <nvhe/trace.h>
 #include <nvhe/trap_handler.h>
 
 DEFINE_PER_CPU(struct kvm_nvhe_init_params, kvm_init_params);
@@ -570,6 +571,35 @@ static void handle___pkvm_teardown_vm(struct kvm_cpu_context *host_ctxt)
 	cpu_reg(host_ctxt, 1) = __pkvm_teardown_vm(handle);
 }
 
+static void handle___pkvm_load_tracing(struct kvm_cpu_context *host_ctxt)
+{
+	 DECLARE_REG(unsigned long, desc_hva, host_ctxt, 1);
+	 DECLARE_REG(size_t, desc_size, host_ctxt, 2);
+
+	 cpu_reg(host_ctxt, 1) = __pkvm_load_tracing(desc_hva, desc_size);
+}
+
+static void handle___pkvm_teardown_tracing(struct kvm_cpu_context *host_ctxt)
+{
+	__pkvm_teardown_tracing();
+
+	cpu_reg(host_ctxt, 1) = 0;
+}
+
+static void handle___pkvm_enable_tracing(struct kvm_cpu_context *host_ctxt)
+{
+	DECLARE_REG(bool, enable, host_ctxt, 1);
+
+	cpu_reg(host_ctxt, 1) = __pkvm_enable_tracing(enable);
+}
+
+static void handle___pkvm_swap_reader_tracing(struct kvm_cpu_context *host_ctxt)
+{
+	DECLARE_REG(unsigned int, cpu, host_ctxt, 1);
+
+	cpu_reg(host_ctxt, 1) = __pkvm_swap_reader_tracing(cpu);
+}
+
 typedef void (*hcall_t)(struct kvm_cpu_context *);
 
 #define HANDLE_FUNC(x)	[__KVM_HOST_SMCCC_FUNC_##x] = (hcall_t)handle_##x
@@ -609,6 +639,10 @@ static const hcall_t host_hcall[] = {
 	HANDLE_FUNC(__pkvm_vcpu_load),
 	HANDLE_FUNC(__pkvm_vcpu_put),
 	HANDLE_FUNC(__pkvm_tlb_flush_vmid),
+	HANDLE_FUNC(__pkvm_load_tracing),
+	HANDLE_FUNC(__pkvm_teardown_tracing),
+	HANDLE_FUNC(__pkvm_enable_tracing),
+	HANDLE_FUNC(__pkvm_swap_reader_tracing),
 };
 
 static void handle_host_hcall(struct kvm_cpu_context *host_ctxt)
diff --git a/arch/arm64/kvm/hyp/nvhe/trace.c b/arch/arm64/kvm/hyp/nvhe/trace.c
new file mode 100644
index 000000000000..4611cef64566
--- /dev/null
+++ b/arch/arm64/kvm/hyp/nvhe/trace.c
@@ -0,0 +1,558 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2025 Google LLC
+ * Author: Vincent Donnefort <vdonnefort@google.com>
+ */
+
+#include <nvhe/clock.h>
+#include <nvhe/mem_protect.h>
+#include <nvhe/mm.h>
+#include <nvhe/trace.h>
+
+#include <asm/percpu.h>
+#include <asm/kvm_mmu.h>
+#include <asm/local.h>
+
+struct hyp_rb_per_cpu {
+	struct trace_buffer_meta	*meta;
+	struct hyp_buffer_page		*tail_page;
+	struct hyp_buffer_page		*reader_page;
+	struct hyp_buffer_page		*head_page;
+	struct hyp_buffer_page		*bpages;
+	u32				nr_pages;
+	u32				status;
+	u64				last_overrun;
+	u64				write_stamp;
+};
+
+#define HYP_RB_UNAVAILABLE	0
+#define HYP_RB_READY		1
+#define HYP_RB_WRITING		2
+
+static struct hyp_buffer_pages_backing hyp_buffer_pages_backing;
+static DEFINE_PER_CPU(struct hyp_rb_per_cpu, trace_rb);
+static DEFINE_HYP_SPINLOCK(trace_rb_lock);
+
+#define HYP_BPAGE_LINK_HEAD	1UL
+#define HYP_BPAGE_LINK_MASK	~HYP_BPAGE_LINK_HEAD
+
+static bool hyp_bpage_try_shunt_link(struct hyp_buffer_page *bpage, struct hyp_buffer_page *dst,
+				     unsigned long old_flags, unsigned long flags)
+{
+	unsigned long *ptr = (unsigned long *)(&bpage->list.next);
+	unsigned long old = (*ptr & HYP_BPAGE_LINK_MASK) | old_flags;
+	unsigned long new = (unsigned long)(&dst->list) | flags;
+
+	return cmpxchg(ptr, old, new) == old;
+}
+
+static void hyp_bpage_set_link_flag(struct hyp_buffer_page *bpage, unsigned long flag)
+{
+	bpage->list.next = (struct list_head *)
+		(((unsigned long)bpage->list.next & HYP_BPAGE_LINK_MASK) | flag);
+}
+
+static struct hyp_buffer_page *hyp_bpage_from_link(struct list_head *list)
+{
+	unsigned long ptr = (unsigned long)list & HYP_BPAGE_LINK_MASK;
+
+	return container_of((struct list_head *)ptr, struct hyp_buffer_page, list);
+}
+
+static struct hyp_buffer_page *hyp_bpage_next_page(struct hyp_buffer_page *bpage)
+{
+	return hyp_bpage_from_link(bpage->list.next);
+}
+
+static bool hyp_bpage_is_head(struct hyp_buffer_page *bpage)
+{
+	return (unsigned long)bpage->list.prev->next & HYP_BPAGE_LINK_HEAD;
+}
+
+static void hyp_bpage_reset(struct hyp_buffer_page *bpage)
+{
+	bpage->write = 0;
+	bpage->entries = 0;
+
+	local_set(&bpage->page->commit, 0);
+}
+
+static int hyp_bpage_init(struct hyp_buffer_page *bpage, unsigned long hva)
+{
+	void *hyp_va = (void *)kern_hyp_va(hva);
+	int ret;
+
+	ret = hyp_pin_shared_mem(hyp_va, hyp_va + PAGE_SIZE);
+	if (ret)
+		return ret;
+
+	INIT_LIST_HEAD(&bpage->list);
+	bpage->page = (struct buffer_data_page *)hyp_va;
+
+	hyp_bpage_reset(bpage);
+
+	return 0;
+}
+
+#define hyp_rb_meta_inc(__meta, __inc)		\
+	WRITE_ONCE((__meta), (__meta + __inc))
+
+static bool hyp_rb_loaded(struct hyp_rb_per_cpu *cpu_buffer)
+{
+	return !!cpu_buffer->bpages;
+}
+
+static int hyp_rb_swap_reader(struct hyp_rb_per_cpu *cpu_buffer)
+{
+	struct hyp_buffer_page *last, *head, *reader;
+	unsigned long overrun;
+
+	if (!hyp_rb_loaded(cpu_buffer))
+		return -ENODEV;
+
+	head = cpu_buffer->head_page;
+	reader = cpu_buffer->reader_page;
+
+	do {
+		/* Run after the writer to find the head */
+		while (!hyp_bpage_is_head(head))
+			cpu_buffer->head_page = head = hyp_bpage_next_page(head);
+
+		/* Connect the reader page around the header page */
+		reader->list.next = head->list.next;
+		reader->list.prev = head->list.prev;
+
+		/* The last page before the head */
+		last = hyp_bpage_from_link(reader->list.next);
+
+		/* The reader page points to the new header page */
+		hyp_bpage_set_link_flag(reader, HYP_BPAGE_LINK_HEAD);
+
+		overrun = smp_load_acquire(&cpu_buffer->meta->overrun);
+	} while (!hyp_bpage_try_shunt_link(last, reader, HYP_BPAGE_LINK_HEAD, 0));
+
+	cpu_buffer->head_page = hyp_bpage_from_link(reader->list.next);
+	cpu_buffer->head_page->list.prev = &reader->list;
+	cpu_buffer->reader_page = head;
+	cpu_buffer->meta->reader.lost_events = overrun - cpu_buffer->last_overrun;
+	cpu_buffer->meta->reader.id = cpu_buffer->reader_page->id;
+	cpu_buffer->last_overrun = overrun;
+
+	return 0;
+}
+
+static struct hyp_buffer_page *hyp_rb_move_tail(struct hyp_rb_per_cpu *cpu_buffer)
+{
+	struct hyp_buffer_page *tail, *new_tail;
+
+	tail = cpu_buffer->tail_page;
+	new_tail = hyp_bpage_next_page(tail);
+
+	if (hyp_bpage_try_shunt_link(tail, new_tail, HYP_BPAGE_LINK_HEAD, 0)) {
+		/*
+		 * Oh no! we've caught the head. There is none anymore and swap_reader will spin
+		 * until we set the new one. Overrun must be written first, to make sure we report
+		 * the correct number of lost events.
+		 */
+		hyp_rb_meta_inc(cpu_buffer->meta->overrun, new_tail->entries);
+		hyp_rb_meta_inc(meta_pages_lost(cpu_buffer->meta), 1);
+
+		smp_store_release(&new_tail->list.next,
+				  (unsigned long)new_tail->list.next | HYP_BPAGE_LINK_HEAD);
+	}
+
+	hyp_bpage_reset(new_tail);
+	cpu_buffer->tail_page = new_tail;
+
+	hyp_rb_meta_inc(meta_pages_touched(cpu_buffer->meta), 1);
+
+	return new_tail;
+}
+
+static unsigned long rb_event_size(unsigned long length)
+{
+	struct ring_buffer_event *event;
+
+	return length + RB_EVNT_HDR_SIZE + sizeof(event->array[0]);
+}
+
+static struct ring_buffer_event *
+rb_add_ts_extend(struct ring_buffer_event *event, u64 delta)
+{
+	event->type_len = RINGBUF_TYPE_TIME_EXTEND;
+	event->time_delta = delta & TS_MASK;
+	event->array[0] = delta >> TS_SHIFT;
+
+	return (struct ring_buffer_event *)((unsigned long)event + 8);
+}
+
+static struct ring_buffer_event *
+hyp_rb_reserve_next(struct hyp_rb_per_cpu *cpu_buffer, unsigned long length)
+{
+	unsigned long ts_ext_size = 0, event_size = rb_event_size(length);
+	struct hyp_buffer_page *tail = cpu_buffer->tail_page;
+	struct ring_buffer_event *event;
+	u32 write, prev_write;
+	u64 ts, time_delta;
+
+	ts = trace_clock();
+
+	time_delta = ts - cpu_buffer->write_stamp;
+
+	if (test_time_stamp(time_delta))
+		ts_ext_size = 8;
+
+	prev_write = tail->write;
+	write = prev_write + event_size + ts_ext_size;
+
+	if (unlikely(write > BUF_PAGE_SIZE))
+		tail = hyp_rb_move_tail(cpu_buffer);
+
+	if (!tail->entries) {
+		tail->page->time_stamp = ts;
+		time_delta = 0;
+		ts_ext_size = 0;
+		write = event_size;
+		prev_write = 0;
+	}
+
+	tail->write = write;
+	tail->entries++;
+
+	cpu_buffer->write_stamp = ts;
+
+	event = (struct ring_buffer_event *)(tail->page->data + prev_write);
+	if (ts_ext_size) {
+		event = rb_add_ts_extend(event, time_delta);
+		time_delta = 0;
+	}
+
+	event->type_len = 0;
+	event->time_delta = time_delta;
+	event->array[0] = event_size - RB_EVNT_HDR_SIZE;
+
+	return event;
+}
+
+void *tracing_reserve_entry(unsigned long length)
+{
+	struct hyp_rb_per_cpu *cpu_buffer = this_cpu_ptr(&trace_rb);
+	struct ring_buffer_event *rb_event;
+
+	if (cmpxchg(&cpu_buffer->status, HYP_RB_READY, HYP_RB_WRITING) != HYP_RB_READY)
+		return NULL;
+
+	rb_event = hyp_rb_reserve_next(cpu_buffer, length);
+
+	return &rb_event->array[1];
+}
+
+void tracing_commit_entry(void)
+{
+	struct hyp_rb_per_cpu *cpu_buffer = this_cpu_ptr(&trace_rb);
+
+	local_set(&cpu_buffer->tail_page->page->commit,
+		  cpu_buffer->tail_page->write);
+	hyp_rb_meta_inc(cpu_buffer->meta->entries, 1);
+
+	/*
+	 * Paired with hyp_rb_disable_writing() to ensure data is
+	 * written to the ring-buffer before teardown.
+	 */
+	smp_store_release(&cpu_buffer->status, HYP_RB_READY);
+}
+
+static void hyp_rb_disable_writing(struct hyp_rb_per_cpu *cpu_buffer)
+{
+	u32 prev_status;
+
+	/* Wait for the buffer to be released */
+	do {
+		prev_status = cmpxchg_acquire(&cpu_buffer->status,
+					      HYP_RB_READY,
+					      HYP_RB_UNAVAILABLE);
+	} while (prev_status == HYP_RB_WRITING);
+}
+
+static int hyp_rb_enable_writing(struct hyp_rb_per_cpu *cpu_buffer)
+{
+	if (!hyp_rb_loaded(cpu_buffer))
+		return -ENODEV;
+
+	cmpxchg(&cpu_buffer->status, HYP_RB_UNAVAILABLE, HYP_RB_READY);
+
+	return 0;
+}
+
+static void hyp_rb_teardown(struct hyp_rb_per_cpu *cpu_buffer)
+{
+	int i;
+
+	if (!hyp_rb_loaded(cpu_buffer))
+		return;
+
+	hyp_rb_disable_writing(cpu_buffer);
+
+	hyp_unpin_shared_mem((void *)cpu_buffer->meta,
+			     (void *)(cpu_buffer->meta) + PAGE_SIZE);
+
+	for (i = 0; i < cpu_buffer->nr_pages; i++) {
+		struct hyp_buffer_page *bpage = &cpu_buffer->bpages[i];
+
+		if (!bpage->page)
+			continue;
+
+		hyp_unpin_shared_mem((void *)bpage->page,
+				     (void *)bpage->page + PAGE_SIZE);
+	}
+
+	cpu_buffer->bpages = 0;
+}
+
+static bool hyp_rb_fits_backing(u32 nr_pages, struct hyp_buffer_page *start)
+{
+	unsigned long max = hyp_buffer_pages_backing.start +
+			    hyp_buffer_pages_backing.size;
+	struct hyp_buffer_page *end = start + nr_pages;
+
+	return (unsigned long)end <= max;
+}
+
+static int hyp_rb_init(struct rb_page_desc *pdesc, struct hyp_buffer_page *start,
+		       struct hyp_rb_per_cpu *cpu_buffer)
+{
+	struct hyp_buffer_page *bpage = start;
+	int i, ret;
+
+	/* At least 1 reader page and one head */
+	if (pdesc->nr_page_va < 2)
+		return -EINVAL;
+
+	/* nr_page_va + 1 must fit nr_pages */
+	if (pdesc->nr_page_va >= U32_MAX)
+		return -EINVAL;
+
+	if (!hyp_rb_fits_backing(pdesc->nr_page_va, start))
+		return -EINVAL;
+
+	if (hyp_rb_loaded(cpu_buffer))
+		return -EBUSY;
+
+	cpu_buffer->bpages = start;
+
+	cpu_buffer->meta = (struct trace_buffer_meta *)kern_hyp_va(pdesc->meta_va);
+	ret = hyp_pin_shared_mem((void *)cpu_buffer->meta,
+				 ((void *)cpu_buffer->meta) + PAGE_SIZE);
+	if (ret)
+		return ret;
+
+	memset(cpu_buffer->meta, 0, sizeof(*cpu_buffer->meta));
+	cpu_buffer->meta->meta_page_size = PAGE_SIZE;
+	cpu_buffer->meta->nr_subbufs = cpu_buffer->nr_pages;
+
+	/* The reader page is not part of the ring initially */
+	ret = hyp_bpage_init(bpage, pdesc->page_va[0]);
+	if (ret)
+		goto err;
+
+	cpu_buffer->nr_pages = 1;
+
+	cpu_buffer->reader_page = bpage;
+	cpu_buffer->tail_page = bpage + 1;
+	cpu_buffer->head_page = bpage + 1;
+
+	for (i = 1; i < pdesc->nr_page_va; i++) {
+		ret = hyp_bpage_init(++bpage, pdesc->page_va[i]);
+		if (ret)
+			goto err;
+
+		bpage->list.next = &(bpage + 1)->list;
+		bpage->list.prev = &(bpage - 1)->list;
+		bpage->id = i;
+
+		cpu_buffer->nr_pages = i + 1;
+	}
+
+	/* Close the ring */
+	bpage->list.next = &cpu_buffer->tail_page->list;
+	cpu_buffer->tail_page->list.prev = &bpage->list;
+
+	/* The last init'ed page points to the head page */
+	hyp_bpage_set_link_flag(bpage, HYP_BPAGE_LINK_HEAD);
+
+	cpu_buffer->last_overrun = 0;
+
+	return 0;
+
+err:
+	hyp_rb_teardown(cpu_buffer);
+
+	return ret;
+}
+
+static int hyp_setup_bpage_backing(struct hyp_trace_desc *desc)
+{
+	unsigned long start = kern_hyp_va(desc->backing.start);
+	size_t size = desc->backing.size;
+	int ret;
+
+	if (hyp_buffer_pages_backing.size)
+		return -EBUSY;
+
+	if (!PAGE_ALIGNED(start) || !PAGE_ALIGNED(size))
+		return -EINVAL;
+
+	ret = __pkvm_host_donate_hyp(hyp_virt_to_pfn((void *)start), size >> PAGE_SHIFT);
+	if (ret)
+		return ret;
+
+	memset((void *)start, 0, size);
+
+	hyp_buffer_pages_backing.start = start;
+	hyp_buffer_pages_backing.size = size;
+
+	return 0;
+}
+
+static void hyp_teardown_bpage_backing(void)
+{
+	unsigned long start = hyp_buffer_pages_backing.start;
+	size_t size = hyp_buffer_pages_backing.size;
+
+	if (!size)
+		return;
+
+	memset((void *)start, 0, size);
+
+	WARN_ON(__pkvm_hyp_donate_host(hyp_virt_to_pfn(start), size >> PAGE_SHIFT));
+
+	hyp_buffer_pages_backing.start = 0;
+	hyp_buffer_pages_backing.size = 0;
+}
+
+int __pkvm_swap_reader_tracing(unsigned int cpu)
+{
+	int ret = 0;
+
+	if (cpu >= hyp_nr_cpus)
+		return -EINVAL;
+
+	hyp_spin_lock(&trace_rb_lock);
+	ret = hyp_rb_swap_reader(per_cpu_ptr(&trace_rb, cpu));
+	hyp_spin_unlock(&trace_rb_lock);
+
+	return ret;
+}
+
+static void __pkvm_teardown_tracing_locked(void)
+{
+	int cpu;
+
+	hyp_assert_lock_held(&trace_rb_lock);
+
+	for (cpu = 0; cpu < hyp_nr_cpus; cpu++) {
+		struct hyp_rb_per_cpu *cpu_buffer = per_cpu_ptr(&trace_rb, cpu);
+
+		hyp_rb_teardown(cpu_buffer);
+	}
+
+	hyp_teardown_bpage_backing();
+}
+
+void __pkvm_teardown_tracing(void)
+{
+	hyp_spin_lock(&trace_rb_lock);
+	__pkvm_teardown_tracing_locked();
+	hyp_spin_unlock(&trace_rb_lock);
+}
+
+static bool rb_page_desc_fits_desc(struct rb_page_desc *pdesc,
+				   unsigned long desc_end)
+{
+	unsigned long *end;
+
+	/* Check we can at least read nr_pages */
+	if ((unsigned long)&pdesc->nr_page_va >= desc_end)
+		return false;
+
+	end = &pdesc->page_va[pdesc->nr_page_va];
+
+	return (unsigned long)end <= desc_end;
+}
+
+int __pkvm_load_tracing(unsigned long desc_hva, size_t desc_size)
+{
+	struct hyp_trace_desc *desc = (struct hyp_trace_desc *)kern_hyp_va(desc_hva);
+	struct trace_page_desc *trace_pdesc = &desc->page_desc;
+	struct hyp_buffer_page *bpage_backing_start;
+	struct rb_page_desc *pdesc;
+	int ret, cpu;
+
+	if (!desc_size || !PAGE_ALIGNED(desc_hva) || !PAGE_ALIGNED(desc_size))
+		return -EINVAL;
+
+	ret = __pkvm_host_donate_hyp(hyp_virt_to_pfn((void *)desc),
+				     desc_size >> PAGE_SHIFT);
+	if (ret)
+		return ret;
+
+	hyp_spin_lock(&trace_rb_lock);
+
+	ret = hyp_setup_bpage_backing(desc);
+	if (ret)
+		goto err;
+
+	bpage_backing_start = (struct hyp_buffer_page *)hyp_buffer_pages_backing.start;
+
+	for_each_rb_page_desc(pdesc, cpu, trace_pdesc) {
+		struct hyp_rb_per_cpu *cpu_buffer;
+		int cpu;
+
+		ret = -EINVAL;
+		if (!rb_page_desc_fits_desc(pdesc, desc_hva + desc_size))
+			break;
+
+		cpu = pdesc->cpu;
+		if (cpu >= hyp_nr_cpus)
+			break;
+
+		cpu_buffer = per_cpu_ptr(&trace_rb, cpu);
+
+		ret = hyp_rb_init(pdesc, bpage_backing_start, cpu_buffer);
+		if (ret)
+			break;
+
+		bpage_backing_start += pdesc->nr_page_va;
+	}
+
+err:
+	if (ret)
+		__pkvm_teardown_tracing_locked();
+
+	hyp_spin_unlock(&trace_rb_lock);
+
+	WARN_ON(__pkvm_hyp_donate_host(hyp_virt_to_pfn((void *)desc),
+				       desc_size >> PAGE_SHIFT));
+	return ret;
+}
+
+int __pkvm_enable_tracing(bool enable)
+{
+	int cpu, ret = enable ? -EINVAL : 0;
+
+	hyp_spin_lock(&trace_rb_lock);
+	for (cpu = 0; cpu < hyp_nr_cpus; cpu++) {
+		struct hyp_rb_per_cpu *cpu_buffer = per_cpu_ptr(&trace_rb, cpu);
+
+		if (enable) {
+			if (!hyp_rb_enable_writing(cpu_buffer))
+				ret = 0;
+		} else {
+			hyp_rb_disable_writing(cpu_buffer);
+		}
+
+	}
+	hyp_spin_unlock(&trace_rb_lock);
+
+	return ret;
+}
-- 
2.48.1.601.g30ceb7b040-goog

[PATCH 06/11] KVM: arm64: Add hyp tracing to tracefs

[Vincent Donnefort]

When running with KVM protected mode, the hypervisor is able to generate
events into tracefs compatible ring-buffers. Plug those ring-buffers to
tracefs. The interface is found in hyp/ and contains the same hierarchy
as any host instances easing the support by existing user-space tools.

This currently doesn't provide any event support which will come later.

Signed-off-by: Vincent Donnefort <vdonnefort@google.com>

diff --git a/arch/arm64/kvm/Makefile b/arch/arm64/kvm/Makefile
index 3cf7adb2b503..865971bb8905 100644
--- a/arch/arm64/kvm/Makefile
+++ b/arch/arm64/kvm/Makefile
@@ -29,6 +29,8 @@ kvm-$(CONFIG_HW_PERF_EVENTS)  += pmu-emul.o pmu.o
 kvm-$(CONFIG_ARM64_PTR_AUTH)  += pauth.o
 kvm-$(CONFIG_PTDUMP_STAGE2_DEBUGFS) += ptdump.o
 
+kvm-$(CONFIG_TRACING) += hyp_trace.o
+
 always-y := hyp_constants.h hyp-constants.s
 
 define rule_gen_hyp_constants
diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c
index b8e55a441282..f3951d36b9c1 100644
--- a/arch/arm64/kvm/arm.c
+++ b/arch/arm64/kvm/arm.c
@@ -25,6 +25,7 @@
 
 #define CREATE_TRACE_POINTS
 #include "trace_arm.h"
+#include "hyp_trace.h"
 
 #include <linux/uaccess.h>
 #include <asm/ptrace.h>
@@ -2319,6 +2320,9 @@ static int __init init_subsystems(void)
 
 	kvm_register_perf_callbacks(NULL);
 
+	err = hyp_trace_init_tracefs();
+	if (err)
+		kvm_err("Failed to initialize Hyp tracing\n");
 out:
 	if (err)
 		hyp_cpu_pm_exit();
diff --git a/arch/arm64/kvm/hyp/hyp-constants.c b/arch/arm64/kvm/hyp/hyp-constants.c
index b257a3b4bfc5..5c4a797a701f 100644
--- a/arch/arm64/kvm/hyp/hyp-constants.c
+++ b/arch/arm64/kvm/hyp/hyp-constants.c
@@ -3,11 +3,15 @@
 #include <linux/kbuild.h>
 #include <nvhe/memory.h>
 #include <nvhe/pkvm.h>
+#include <nvhe/trace.h>
 
 int main(void)
 {
 	DEFINE(STRUCT_HYP_PAGE_SIZE,	sizeof(struct hyp_page));
 	DEFINE(PKVM_HYP_VM_SIZE,	sizeof(struct pkvm_hyp_vm));
 	DEFINE(PKVM_HYP_VCPU_SIZE,	sizeof(struct pkvm_hyp_vcpu));
+#ifdef CONFIG_TRACING
+	DEFINE(STRUCT_HYP_BUFFER_PAGE_SIZE,	sizeof(struct hyp_buffer_page));
+#endif
 	return 0;
 }
diff --git a/arch/arm64/kvm/hyp/nvhe/trace.c b/arch/arm64/kvm/hyp/nvhe/trace.c
index 4611cef64566..2f1e5005c5d4 100644
--- a/arch/arm64/kvm/hyp/nvhe/trace.c
+++ b/arch/arm64/kvm/hyp/nvhe/trace.c
@@ -123,7 +123,7 @@ static int hyp_rb_swap_reader(struct hyp_rb_per_cpu *cpu_buffer)
 		reader->list.prev = head->list.prev;
 
 		/* The last page before the head */
-		last = hyp_bpage_from_link(reader->list.next);
+		last = hyp_bpage_from_link(head->list.prev);
 
 		/* The reader page points to the new header page */
 		hyp_bpage_set_link_flag(reader, HYP_BPAGE_LINK_HEAD);
diff --git a/arch/arm64/kvm/hyp_trace.c b/arch/arm64/kvm/hyp_trace.c
new file mode 100644
index 000000000000..c08ae8c33052
--- /dev/null
+++ b/arch/arm64/kvm/hyp_trace.c
@@ -0,0 +1,666 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2025 Google LLC
+ * Author: Vincent Donnefort <vdonnefort@google.com>
+ */
+
+#include <linux/arm-smccc.h>
+#include <linux/percpu-defs.h>
+#include <linux/tracefs.h>
+
+#include <asm/kvm_host.h>
+#include <asm/kvm_hyptrace.h>
+
+#include "hyp_constants.h"
+#include "hyp_trace.h"
+
+#define RB_POLL_MS 100
+
+#define TRACEFS_DIR "hypervisor"
+#define TRACEFS_MODE_WRITE 0640
+#define TRACEFS_MODE_READ 0440
+
+static struct hyp_trace_buffer {
+	struct hyp_trace_desc		*desc;
+	struct ring_buffer_remote	remote;
+	struct trace_buffer		*trace_buffer;
+	size_t				desc_size;
+	bool				tracing_on;
+	int				nr_readers;
+	struct mutex			lock;
+} hyp_trace_buffer = {
+	.lock		= __MUTEX_INITIALIZER(hyp_trace_buffer.lock),
+};
+
+static size_t hyp_trace_buffer_size = 7 << 10;
+
+/* Number of pages the ring-buffer requires to accommodate for size */
+#define NR_PAGES(size) \
+	((PAGE_ALIGN(size) >> PAGE_SHIFT) + 1)
+
+static inline bool hyp_trace_buffer_loaded(struct hyp_trace_buffer *hyp_buffer)
+{
+	return !!hyp_buffer->trace_buffer;
+}
+
+static inline bool hyp_trace_buffer_used(struct hyp_trace_buffer *hyp_buffer)
+{
+	return hyp_buffer->nr_readers || hyp_buffer->tracing_on ||
+		!ring_buffer_empty(hyp_buffer->trace_buffer);
+}
+
+static int
+bpage_backing_alloc(struct hyp_buffer_pages_backing *bpage_backing, size_t size)
+{
+	size_t backing_size;
+	void *start;
+
+	backing_size = PAGE_ALIGN(STRUCT_HYP_BUFFER_PAGE_SIZE * NR_PAGES(size) *
+				  num_possible_cpus());
+
+	start = alloc_pages_exact(backing_size, GFP_KERNEL_ACCOUNT);
+	if (!start)
+		return -ENOMEM;
+
+	bpage_backing->start = (unsigned long)start;
+	bpage_backing->size = backing_size;
+
+	return 0;
+}
+
+static void
+bpage_backing_free(struct hyp_buffer_pages_backing *bpage_backing)
+{
+	free_pages_exact((void *)bpage_backing->start, bpage_backing->size);
+}
+
+static int __get_reader_page(int cpu)
+{
+	return kvm_call_hyp_nvhe(__pkvm_swap_reader_tracing, cpu);
+}
+
+static void hyp_trace_free_pages(struct hyp_trace_desc *desc)
+{
+	struct rb_page_desc *rb_desc;
+	int cpu, id;
+
+	for_each_rb_page_desc(rb_desc, cpu, &desc->page_desc) {
+		free_page(rb_desc->meta_va);
+		for (id = 0; id < rb_desc->nr_page_va; id++)
+			free_page(rb_desc->page_va[id]);
+	}
+}
+
+static int hyp_trace_alloc_pages(struct hyp_trace_desc *desc, size_t size)
+{
+	int err = 0, cpu, id, nr_pages = NR_PAGES(size);
+	struct trace_page_desc *trace_desc;
+	struct rb_page_desc *rb_desc;
+
+	trace_desc = &desc->page_desc;
+	trace_desc->nr_cpus = 0;
+	trace_desc->struct_len = offsetof(struct trace_page_desc, __data);
+
+	rb_desc = (struct rb_page_desc *)&trace_desc->__data[0];
+
+	for_each_possible_cpu(cpu) {
+		rb_desc->cpu = cpu;
+		rb_desc->nr_page_va = 0;
+		rb_desc->meta_va = (unsigned long)page_to_virt(alloc_page(GFP_KERNEL));
+		if (!rb_desc->meta_va) {
+			err = -ENOMEM;
+			break;
+		}
+		for (id = 0; id < nr_pages; id++) {
+			rb_desc->page_va[id] = (unsigned long)page_to_virt(alloc_page(GFP_KERNEL));
+			if (!rb_desc->page_va[id]) {
+				err = -ENOMEM;
+				break;
+			}
+			rb_desc->nr_page_va++;
+		}
+		trace_desc->nr_cpus++;
+		trace_desc->struct_len += offsetof(struct rb_page_desc, page_va);
+		trace_desc->struct_len += sizeof(rb_desc->page_va[0]) * rb_desc->nr_page_va;
+		rb_desc = __next_rb_page_desc(rb_desc);
+	}
+
+	if (err) {
+		hyp_trace_free_pages(desc);
+		return err;
+	}
+
+	return 0;
+}
+
+static int __load_page(unsigned long va)
+{
+	return kvm_call_hyp_nvhe(__pkvm_host_share_hyp, virt_to_pfn((void *)va), 1);
+}
+
+static void __teardown_page(unsigned long va)
+{
+	WARN_ON(kvm_call_hyp_nvhe(__pkvm_host_unshare_hyp, virt_to_pfn((void *)va), 1));
+}
+
+static void hyp_trace_teardown_pages(struct hyp_trace_desc *desc,
+				     int last_cpu)
+{
+	struct rb_page_desc *rb_desc;
+	int cpu, id;
+
+	for_each_rb_page_desc(rb_desc, cpu, &desc->page_desc) {
+		if (cpu > last_cpu)
+			break;
+		__teardown_page(rb_desc->meta_va);
+		for (id = 0; id < rb_desc->nr_page_va; id++)
+			__teardown_page(rb_desc->page_va[id]);
+	}
+}
+
+static int hyp_trace_load_pages(struct hyp_trace_desc *desc)
+{
+	int last_loaded_cpu = 0, cpu, id, err = -EINVAL;
+	struct rb_page_desc *rb_desc;
+
+	for_each_rb_page_desc(rb_desc, cpu, &desc->page_desc) {
+		err = __load_page(rb_desc->meta_va);
+		if (err)
+			break;
+
+		for (id = 0; id < rb_desc->nr_page_va; id++) {
+			err = __load_page(rb_desc->page_va[id]);
+			if (err)
+				break;
+		}
+
+		if (!err)
+			continue;
+
+		for (id--; id >= 0; id--)
+			__teardown_page(rb_desc->page_va[id]);
+
+		last_loaded_cpu = cpu - 1;
+
+		break;
+	}
+
+	if (!err)
+		return 0;
+
+	hyp_trace_teardown_pages(desc, last_loaded_cpu);
+
+	return err;
+}
+
+static int hyp_trace_buffer_load(struct hyp_trace_buffer *hyp_buffer, size_t size)
+{
+	int ret, nr_pages = NR_PAGES(size);
+	struct rb_page_desc *rbdesc;
+	struct hyp_trace_desc *desc;
+	size_t desc_size;
+
+	if (hyp_trace_buffer_loaded(hyp_buffer))
+		return 0;
+
+	desc_size = size_add(offsetof(struct hyp_trace_desc, page_desc),
+			     offsetof(struct trace_page_desc, __data));
+	desc_size = size_add(desc_size,
+			     size_mul(num_possible_cpus(),
+				      struct_size(rbdesc, page_va, nr_pages)));
+	if (desc_size == SIZE_MAX)
+		return -E2BIG;
+
+	/*
+	 * The hypervisor will unmap the descriptor from the host to protect the
+	 * reading. Page granularity for the allocation ensures no other
+	 * useful data will be unmapped.
+	 */
+	desc_size = PAGE_ALIGN(desc_size);
+
+	desc = (struct hyp_trace_desc *)alloc_pages_exact(desc_size, GFP_KERNEL);
+	if (!desc)
+		return -ENOMEM;
+
+	ret = hyp_trace_alloc_pages(desc, size);
+	if (ret)
+		goto err_free_desc;
+
+	ret = bpage_backing_alloc(&desc->backing, size);
+	if (ret)
+		goto err_free_pages;
+
+	ret = hyp_trace_load_pages(desc);
+	if (ret)
+		goto err_free_backing;
+
+	ret = kvm_call_hyp_nvhe(__pkvm_load_tracing, (unsigned long)desc, desc_size);
+	if (ret)
+		goto err_teardown_pages;
+
+	hyp_buffer->remote.pdesc = &desc->page_desc;
+	hyp_buffer->remote.get_reader_page = __get_reader_page;
+	hyp_buffer->trace_buffer = ring_buffer_remote(&hyp_buffer->remote);
+	if (!hyp_buffer->trace_buffer) {
+		ret = -ENOMEM;
+		goto err_teardown_tracing;
+	}
+
+	hyp_buffer->desc = desc;
+	hyp_buffer->desc_size = desc_size;
+
+	return 0;
+
+err_teardown_tracing:
+	kvm_call_hyp_nvhe(__pkvm_teardown_tracing);
+err_teardown_pages:
+	hyp_trace_teardown_pages(desc, INT_MAX);
+err_free_backing:
+	bpage_backing_free(&desc->backing);
+err_free_pages:
+	hyp_trace_free_pages(desc);
+err_free_desc:
+	free_pages_exact(desc, desc_size);
+
+	return ret;
+}
+
+static void hyp_trace_buffer_teardown(struct hyp_trace_buffer *hyp_buffer)
+{
+	struct hyp_trace_desc *desc = hyp_buffer->desc;
+	size_t desc_size = hyp_buffer->desc_size;
+
+	if (!hyp_trace_buffer_loaded(hyp_buffer))
+		return;
+
+	if (hyp_trace_buffer_used(hyp_buffer))
+		return;
+
+	if (kvm_call_hyp_nvhe(__pkvm_teardown_tracing))
+		return;
+
+	ring_buffer_free(hyp_buffer->trace_buffer);
+	hyp_trace_teardown_pages(desc, INT_MAX);
+	bpage_backing_free(&desc->backing);
+	hyp_trace_free_pages(desc);
+	free_pages_exact(desc, desc_size);
+	hyp_buffer->trace_buffer = NULL;
+}
+
+static int hyp_trace_start(void)
+{
+	struct hyp_trace_buffer *hyp_buffer = &hyp_trace_buffer;
+	int ret = 0;
+
+	mutex_lock(&hyp_buffer->lock);
+
+	if (hyp_buffer->tracing_on)
+		goto out;
+
+	ret = hyp_trace_buffer_load(hyp_buffer, hyp_trace_buffer_size);
+	if (ret)
+		goto out;
+
+	ret = kvm_call_hyp_nvhe(__pkvm_enable_tracing, true);
+	if (ret) {
+		hyp_trace_buffer_teardown(hyp_buffer);
+		goto out;
+	}
+
+	hyp_buffer->tracing_on = true;
+
+out:
+	mutex_unlock(&hyp_buffer->lock);
+
+	return ret;
+}
+
+static void hyp_trace_stop(void)
+{
+	struct hyp_trace_buffer *hyp_buffer = &hyp_trace_buffer;
+	int ret;
+
+	mutex_lock(&hyp_buffer->lock);
+
+	if (!hyp_buffer->tracing_on)
+		goto end;
+
+	ret = kvm_call_hyp_nvhe(__pkvm_enable_tracing, false);
+	if (!ret) {
+		ring_buffer_poll_remote(hyp_buffer->trace_buffer,
+					RING_BUFFER_ALL_CPUS);
+		hyp_buffer->tracing_on = false;
+		hyp_trace_buffer_teardown(hyp_buffer);
+	}
+
+end:
+	mutex_unlock(&hyp_buffer->lock);
+}
+
+static ssize_t hyp_tracing_on(struct file *filp, const char __user *ubuf,
+			      size_t cnt, loff_t *ppos)
+{
+	unsigned long val;
+	int ret;
+
+	ret = kstrtoul_from_user(ubuf, cnt, 10, &val);
+	if (ret)
+		return ret;
+
+	if (val)
+		ret = hyp_trace_start();
+	else
+		hyp_trace_stop();
+
+	return ret ? ret : cnt;
+}
+
+static ssize_t hyp_tracing_on_read(struct file *filp, char __user *ubuf,
+				   size_t cnt, loff_t *ppos)
+{
+	char buf[3];
+	int r;
+
+	mutex_lock(&hyp_trace_buffer.lock);
+	r = sprintf(buf, "%d\n", hyp_trace_buffer.tracing_on);
+	mutex_unlock(&hyp_trace_buffer.lock);
+
+	return simple_read_from_buffer(ubuf, cnt, ppos, buf, r);
+}
+
+static const struct file_operations hyp_tracing_on_fops = {
+	.write	= hyp_tracing_on,
+	.read	= hyp_tracing_on_read,
+};
+
+static ssize_t hyp_buffer_size(struct file *filp, const char __user *ubuf,
+			       size_t cnt, loff_t *ppos)
+{
+	unsigned long val;
+	int ret;
+
+	ret = kstrtoul_from_user(ubuf, cnt, 10, &val);
+	if (ret)
+		return ret;
+
+	if (!val)
+		return -EINVAL;
+
+	mutex_lock(&hyp_trace_buffer.lock);
+	hyp_trace_buffer_size = val << 10; /* KB to B */
+	mutex_unlock(&hyp_trace_buffer.lock);
+
+	return cnt;
+}
+
+static ssize_t hyp_buffer_size_read(struct file *filp, char __user *ubuf,
+				    size_t cnt, loff_t *ppos)
+{
+	char buf[64];
+	int r;
+
+	mutex_lock(&hyp_trace_buffer.lock);
+	r = sprintf(buf, "%lu (%s)\n", hyp_trace_buffer_size >> 10,
+		    hyp_trace_buffer_loaded(&hyp_trace_buffer) ?
+			"loaded" : "unloaded");
+	mutex_unlock(&hyp_trace_buffer.lock);
+
+	return simple_read_from_buffer(ubuf, cnt, ppos, buf, r);
+}
+
+static const struct file_operations hyp_buffer_size_fops = {
+	.write	= hyp_buffer_size,
+	.read	= hyp_buffer_size_read,
+};
+
+static void ht_print_trace_time(struct ht_iterator *iter)
+{
+	unsigned long usecs_rem;
+	u64 ts_ns = iter->ts;
+
+	do_div(ts_ns, 1000);
+	usecs_rem = do_div(ts_ns, USEC_PER_SEC);
+
+	trace_seq_printf(&iter->seq, "%5lu.%06lu: ",
+			 (unsigned long)ts_ns, usecs_rem);
+}
+
+static void ht_print_trace_cpu(struct ht_iterator *iter)
+{
+	trace_seq_printf(&iter->seq, "[%03d]\t", iter->ent_cpu);
+}
+
+static int ht_print_trace_fmt(struct ht_iterator *iter)
+{
+	if (iter->lost_events)
+		trace_seq_printf(&iter->seq, "CPU:%d [LOST %lu EVENTS]\n",
+				 iter->ent_cpu, iter->lost_events);
+
+	ht_print_trace_cpu(iter);
+	ht_print_trace_time(iter);
+
+	return trace_seq_has_overflowed(&iter->seq) ? -EOVERFLOW : 0;
+};
+
+static struct ring_buffer_event *__ht_next_pipe_event(struct ht_iterator *iter)
+{
+	struct ring_buffer_event *evt = NULL;
+	int cpu = iter->cpu;
+
+	if (cpu != RING_BUFFER_ALL_CPUS) {
+		if (ring_buffer_empty_cpu(iter->trace_buffer, cpu))
+			return NULL;
+
+		iter->ent_cpu = cpu;
+
+		return ring_buffer_peek(iter->trace_buffer, cpu, &iter->ts,
+					&iter->lost_events);
+	}
+
+	iter->ts = LLONG_MAX;
+	for_each_possible_cpu(cpu) {
+		struct ring_buffer_event *_evt;
+		unsigned long lost_events;
+		u64 ts;
+
+		if (ring_buffer_empty_cpu(iter->trace_buffer, cpu))
+			continue;
+
+		_evt = ring_buffer_peek(iter->trace_buffer, cpu, &ts,
+					&lost_events);
+		if (!_evt)
+			continue;
+
+		if (ts >= iter->ts)
+			continue;
+
+		iter->ts = ts;
+		iter->ent_cpu = cpu;
+		iter->lost_events = lost_events;
+		evt = _evt;
+	}
+
+	return evt;
+}
+
+static void *ht_next_pipe_event(struct ht_iterator *iter)
+{
+	struct ring_buffer_event *event;
+
+	event = __ht_next_pipe_event(iter);
+	if (!event)
+		return NULL;
+
+	iter->ent = (struct hyp_entry_hdr *)&event->array[1];
+	iter->ent_size = event->array[0];
+
+	return iter;
+}
+
+static ssize_t
+hyp_trace_pipe_read(struct file *file, char __user *ubuf,
+		    size_t cnt, loff_t *ppos)
+{
+	struct ht_iterator *iter = (struct ht_iterator *)file->private_data;
+	int ret;
+
+copy_to_user:
+	ret = trace_seq_to_user(&iter->seq, ubuf, cnt);
+	if (ret != -EBUSY)
+		return ret;
+
+	trace_seq_init(&iter->seq);
+
+	ret = ring_buffer_wait(iter->trace_buffer, iter->cpu, 0, NULL, NULL);
+	if (ret < 0)
+		return ret;
+
+	while (ht_next_pipe_event(iter)) {
+		int prev_len = iter->seq.seq.len;
+
+		if (ht_print_trace_fmt(iter)) {
+			iter->seq.seq.len = prev_len;
+			break;
+		}
+
+		ring_buffer_consume(iter->trace_buffer, iter->ent_cpu, NULL,
+				    NULL);
+	}
+
+	goto copy_to_user;
+}
+
+static void __poll_remote(struct work_struct *work)
+{
+	struct delayed_work *dwork = to_delayed_work(work);
+	struct ht_iterator *iter;
+
+	iter = container_of(dwork, struct ht_iterator, poll_work);
+
+	ring_buffer_poll_remote(iter->trace_buffer, iter->cpu);
+
+	schedule_delayed_work((struct delayed_work *)work,
+			      msecs_to_jiffies(RB_POLL_MS));
+}
+
+static int hyp_trace_pipe_open(struct inode *inode, struct file *file)
+{
+	struct hyp_trace_buffer *hyp_buffer = &hyp_trace_buffer;
+	int cpu = (s64)inode->i_private;
+	struct ht_iterator *iter = NULL;
+	int ret;
+
+	mutex_lock(&hyp_buffer->lock);
+
+	if (hyp_buffer->nr_readers == INT_MAX) {
+		ret = -EBUSY;
+		goto unlock;
+	}
+
+	ret = hyp_trace_buffer_load(hyp_buffer, hyp_trace_buffer_size);
+	if (ret)
+		goto unlock;
+
+	iter = kzalloc(sizeof(*iter), GFP_KERNEL);
+	if (!iter) {
+		ret = -ENOMEM;
+		goto unlock;
+	}
+	iter->trace_buffer = hyp_buffer->trace_buffer;
+	iter->cpu = cpu;
+	trace_seq_init(&iter->seq);
+	file->private_data = iter;
+
+	ret = ring_buffer_poll_remote(hyp_buffer->trace_buffer, cpu);
+	if (ret)
+		goto unlock;
+
+	INIT_DELAYED_WORK(&iter->poll_work, __poll_remote);
+	schedule_delayed_work(&iter->poll_work, msecs_to_jiffies(RB_POLL_MS));
+
+	hyp_buffer->nr_readers++;
+
+unlock:
+	if (ret) {
+		hyp_trace_buffer_teardown(hyp_buffer);
+		kfree(iter);
+	}
+
+	mutex_unlock(&hyp_buffer->lock);
+
+	return ret;
+}
+
+static int hyp_trace_pipe_release(struct inode *inode, struct file *file)
+{
+	struct hyp_trace_buffer *hyp_buffer = &hyp_trace_buffer;
+	struct ht_iterator *iter = file->private_data;
+
+	cancel_delayed_work_sync(&iter->poll_work);
+
+	mutex_lock(&hyp_buffer->lock);
+
+	WARN_ON(--hyp_buffer->nr_readers < 0);
+
+	hyp_trace_buffer_teardown(hyp_buffer);
+
+	mutex_unlock(&hyp_buffer->lock);
+
+	kfree(iter);
+
+	return 0;
+}
+
+static const struct file_operations hyp_trace_pipe_fops = {
+	.open           = hyp_trace_pipe_open,
+	.read           = hyp_trace_pipe_read,
+	.release        = hyp_trace_pipe_release,
+};
+
+int hyp_trace_init_tracefs(void)
+{
+	struct dentry *root, *per_cpu_root;
+	char per_cpu_name[16];
+	long cpu;
+
+	if (!is_protected_kvm_enabled())
+		return 0;
+
+	root = tracefs_create_dir(TRACEFS_DIR, NULL);
+	if (!root) {
+		pr_err("Failed to create tracefs "TRACEFS_DIR"/\n");
+		return -ENODEV;
+	}
+
+	tracefs_create_file("tracing_on", TRACEFS_MODE_WRITE, root, NULL,
+			    &hyp_tracing_on_fops);
+
+	tracefs_create_file("buffer_size_kb", TRACEFS_MODE_WRITE, root, NULL,
+			    &hyp_buffer_size_fops);
+
+	tracefs_create_file("trace_pipe", TRACEFS_MODE_WRITE, root,
+			    (void *)RING_BUFFER_ALL_CPUS, &hyp_trace_pipe_fops);
+
+	per_cpu_root = tracefs_create_dir("per_cpu", root);
+	if (!per_cpu_root) {
+		pr_err("Failed to create tracefs folder "TRACEFS_DIR"/per_cpu/\n");
+		return -ENODEV;
+	}
+
+	for_each_possible_cpu(cpu) {
+		struct dentry *per_cpu_dir;
+
+		snprintf(per_cpu_name, sizeof(per_cpu_name), "cpu%ld", cpu);
+		per_cpu_dir = tracefs_create_dir(per_cpu_name, per_cpu_root);
+		if (!per_cpu_dir) {
+			pr_warn("Failed to create tracefs "TRACEFS_DIR"/per_cpu/cpu%ld\n",
+				cpu);
+			continue;
+		}
+
+		tracefs_create_file("trace_pipe", TRACEFS_MODE_READ, per_cpu_dir,
+				    (void *)cpu, &hyp_trace_pipe_fops);
+	}
+
+	return 0;
+}
diff --git a/arch/arm64/kvm/hyp_trace.h b/arch/arm64/kvm/hyp_trace.h
new file mode 100644
index 000000000000..14fc06c625a6
--- /dev/null
+++ b/arch/arm64/kvm/hyp_trace.h
@@ -0,0 +1,28 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifndef __ARM64_KVM_HYP_TRACE_H__
+#define __ARM64_KVM_HYP_TRACE_H__
+
+#include <linux/trace_seq.h>
+#include <linux/workqueue.h>
+
+struct ht_iterator {
+	struct trace_buffer	*trace_buffer;
+	int			cpu;
+	struct hyp_entry_hdr	*ent;
+	unsigned long		lost_events;
+	int			ent_cpu;
+	size_t			ent_size;
+	u64			ts;
+	void			*spare;
+	size_t			copy_leftover;
+	struct trace_seq	seq;
+	struct delayed_work	poll_work;
+};
+
+#ifdef CONFIG_TRACING
+int hyp_trace_init_tracefs(void);
+#else
+static inline int hyp_trace_init_tracefs(void) { return 0; }
+#endif
+#endif
-- 
2.48.1.601.g30ceb7b040-goog

[PATCH 07/11] KVM: arm64: Add clock for hyp tracefs

[Vincent Donnefort]

Configure the hypervisor tracing clock before starting tracing. For
tracing purpose, the boot clock is interesting as it doesn't stop on
suspend. However, it is corrected on a regular basis, which implies we
need to re-evaluate it every once in a while.

Cc: John Stultz <jstultz@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Stephen Boyd <sboyd@kernel.org>
Cc: Christopher S. Hall <christopher.s.hall@intel.com>
Cc: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Vincent Donnefort <vdonnefort@google.com>

diff --git a/arch/arm64/include/asm/kvm_asm.h b/arch/arm64/include/asm/kvm_asm.h
index b5893e0afe8e..87d3e0e73b68 100644
--- a/arch/arm64/include/asm/kvm_asm.h
+++ b/arch/arm64/include/asm/kvm_asm.h
@@ -87,6 +87,7 @@ enum __kvm_host_smccc_func {
 	__KVM_HOST_SMCCC_FUNC___pkvm_vcpu_load,
 	__KVM_HOST_SMCCC_FUNC___pkvm_vcpu_put,
 	__KVM_HOST_SMCCC_FUNC___pkvm_tlb_flush_vmid,
+	__KVM_HOST_SMCCC_FUNC___pkvm_update_clock_tracing,
 	__KVM_HOST_SMCCC_FUNC___pkvm_load_tracing,
 	__KVM_HOST_SMCCC_FUNC___pkvm_teardown_tracing,
 	__KVM_HOST_SMCCC_FUNC___pkvm_enable_tracing,
diff --git a/arch/arm64/kvm/hyp/include/nvhe/trace.h b/arch/arm64/kvm/hyp/include/nvhe/trace.h
index bf74a6ee322d..6f1cc571b47a 100644
--- a/arch/arm64/kvm/hyp/include/nvhe/trace.h
+++ b/arch/arm64/kvm/hyp/include/nvhe/trace.h
@@ -16,6 +16,7 @@ struct hyp_buffer_page {
 void *tracing_reserve_entry(unsigned long length);
 void tracing_commit_entry(void);
 
+void __pkvm_update_clock_tracing(u32 mult, u32 shift, u64 epoch_ns, u64 epoch_cyc);
 int __pkvm_load_tracing(unsigned long desc_va, size_t desc_size);
 void __pkvm_teardown_tracing(void);
 int __pkvm_enable_tracing(bool enable);
@@ -24,6 +25,8 @@ int __pkvm_swap_reader_tracing(unsigned int cpu);
 static inline void *tracing_reserve_entry(unsigned long length) { return NULL; }
 static inline void tracing_commit_entry(void) { }
 
+static inline
+void __pkvm_update_clock_tracing(u32 mult, u32 shift, u64 epoch_ns, u64 epoch_cyc) { }
 static inline int __pkvm_load_tracing(unsigned long desc_va, size_t desc_size) { return -ENODEV; }
 static inline void __pkvm_teardown_tracing(void) { }
 static inline int __pkvm_enable_tracing(bool enable) { return -ENODEV; }
diff --git a/arch/arm64/kvm/hyp/nvhe/hyp-main.c b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
index ced0a161d56e..a8b497b22407 100644
--- a/arch/arm64/kvm/hyp/nvhe/hyp-main.c
+++ b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
@@ -571,6 +571,18 @@ static void handle___pkvm_teardown_vm(struct kvm_cpu_context *host_ctxt)
 	cpu_reg(host_ctxt, 1) = __pkvm_teardown_vm(handle);
 }
 
+static void handle___pkvm_update_clock_tracing(struct kvm_cpu_context *host_ctxt)
+{
+	DECLARE_REG(u32, mult, host_ctxt, 1);
+	DECLARE_REG(u32, shift, host_ctxt, 2);
+	DECLARE_REG(u64, epoch_ns, host_ctxt, 3);
+	DECLARE_REG(u64, epoch_cyc, host_ctxt, 4);
+
+	__pkvm_update_clock_tracing(mult, shift, epoch_ns, epoch_cyc);
+
+	cpu_reg(host_ctxt, 1) = 0;
+}
+
 static void handle___pkvm_load_tracing(struct kvm_cpu_context *host_ctxt)
 {
 	 DECLARE_REG(unsigned long, desc_hva, host_ctxt, 1);
@@ -639,6 +651,7 @@ static const hcall_t host_hcall[] = {
 	HANDLE_FUNC(__pkvm_vcpu_load),
 	HANDLE_FUNC(__pkvm_vcpu_put),
 	HANDLE_FUNC(__pkvm_tlb_flush_vmid),
+	HANDLE_FUNC(__pkvm_update_clock_tracing),
 	HANDLE_FUNC(__pkvm_load_tracing),
 	HANDLE_FUNC(__pkvm_teardown_tracing),
 	HANDLE_FUNC(__pkvm_enable_tracing),
diff --git a/arch/arm64/kvm/hyp/nvhe/trace.c b/arch/arm64/kvm/hyp/nvhe/trace.c
index 2f1e5005c5d4..d79b6539377e 100644
--- a/arch/arm64/kvm/hyp/nvhe/trace.c
+++ b/arch/arm64/kvm/hyp/nvhe/trace.c
@@ -430,6 +430,22 @@ static void hyp_teardown_bpage_backing(void)
 	hyp_buffer_pages_backing.size = 0;
 }
 
+void __pkvm_update_clock_tracing(u32 mult, u32 shift, u64 epoch_ns, u64 epoch_cyc)
+{
+	int cpu;
+
+	/* After this loop, all CPUs are observing the new bank... */
+	for (cpu = 0; cpu < hyp_nr_cpus; cpu++) {
+		struct hyp_rb_per_cpu *cpu_buffer = per_cpu_ptr(&trace_rb, cpu);
+
+		while (READ_ONCE(cpu_buffer->status) == HYP_RB_WRITING)
+			;
+	}
+
+	/* ...we can now override the old one and swap. */
+	trace_clock_update(mult, shift, epoch_ns, epoch_cyc);
+}
+
 int __pkvm_swap_reader_tracing(unsigned int cpu)
 {
 	int ret = 0;
diff --git a/arch/arm64/kvm/hyp_trace.c b/arch/arm64/kvm/hyp_trace.c
index c08ae8c33052..3f91ad69c25b 100644
--- a/arch/arm64/kvm/hyp_trace.c
+++ b/arch/arm64/kvm/hyp_trace.c
@@ -16,10 +16,33 @@
 
 #define RB_POLL_MS 100
 
+/* Same 10min used by clocksource when width is more than 32-bits */
+#define CLOCK_MAX_CONVERSION_S	600
+/*
+ * Time to give for the clock init. Long enough to get a good mult/shift
+ * estimation. Short enough to not delay the tracing start too much.
+ */
+#define CLOCK_INIT_MS		100
+/*
+ * Time between clock checks. Must be small enough to catch clock deviation when
+ * it is still tiny.
+ */
+#define CLOCK_UPDATE_MS		500
+
 #define TRACEFS_DIR "hypervisor"
 #define TRACEFS_MODE_WRITE 0640
 #define TRACEFS_MODE_READ 0440
 
+struct hyp_trace_clock {
+	u64			cycles;
+	u64			cyc_overflow64;
+	u64			boot;
+	u32			mult;
+	u32			shift;
+	struct delayed_work	work;
+	struct completion	ready;
+};
+
 static struct hyp_trace_buffer {
 	struct hyp_trace_desc		*desc;
 	struct ring_buffer_remote	remote;
@@ -28,6 +51,7 @@ static struct hyp_trace_buffer {
 	bool				tracing_on;
 	int				nr_readers;
 	struct mutex			lock;
+	struct hyp_trace_clock		clock;
 } hyp_trace_buffer = {
 	.lock		= __MUTEX_INITIALIZER(hyp_trace_buffer.lock),
 };
@@ -74,6 +98,103 @@ bpage_backing_free(struct hyp_buffer_pages_backing *bpage_backing)
 	free_pages_exact((void *)bpage_backing->start, bpage_backing->size);
 }
 
+static void __hyp_clock_work(struct work_struct *work)
+{
+	struct delayed_work *dwork = to_delayed_work(work);
+	struct hyp_trace_buffer *hyp_buffer;
+	struct hyp_trace_clock *hyp_clock;
+	struct system_time_snapshot snap;
+	u64 rate, delta_cycles;
+	u64 boot, delta_boot;
+
+	hyp_clock = container_of(dwork, struct hyp_trace_clock, work);
+	hyp_buffer = container_of(hyp_clock, struct hyp_trace_buffer, clock);
+
+	ktime_get_snapshot(&snap);
+	boot = ktime_to_ns(snap.boot);
+
+	delta_boot = boot - hyp_clock->boot;
+	delta_cycles = snap.cycles - hyp_clock->cycles;
+
+	/* Compare hyp clock with the kernel boot clock */
+	if (hyp_clock->mult) {
+		u64 err, cur = delta_cycles;
+
+		if (WARN_ON_ONCE(cur >= hyp_clock->cyc_overflow64)) {
+			__uint128_t tmp = (__uint128_t)cur * hyp_clock->mult;
+
+			cur = tmp >> hyp_clock->shift;
+		} else {
+			cur *= hyp_clock->mult;
+			cur >>= hyp_clock->shift;
+		}
+		cur += hyp_clock->boot;
+
+		err = abs_diff(cur, boot);
+		/* No deviation, only update epoch if necessary */
+		if (!err) {
+			if (delta_cycles >= (hyp_clock->cyc_overflow64 >> 1))
+				goto fast_forward;
+
+			goto resched;
+		}
+
+		/* Warn if the error is above tracing precision (1us) */
+		if (hyp_buffer->tracing_on && err > NSEC_PER_USEC)
+			pr_warn_ratelimited("hyp trace clock off by %lluus\n",
+					    err / NSEC_PER_USEC);
+	}
+
+	rate = div64_u64(delta_cycles * NSEC_PER_SEC, delta_boot);
+
+	clocks_calc_mult_shift(&hyp_clock->mult, &hyp_clock->shift,
+			       rate, NSEC_PER_SEC, CLOCK_MAX_CONVERSION_S);
+
+	/* Add a comfortable 50% margin */
+	hyp_clock->cyc_overflow64 = (U64_MAX / hyp_clock->mult) >> 1;
+
+fast_forward:
+	hyp_clock->cycles = snap.cycles;
+	hyp_clock->boot = boot;
+	kvm_call_hyp_nvhe(__pkvm_update_clock_tracing, hyp_clock->mult,
+			  hyp_clock->shift, hyp_clock->boot, hyp_clock->cycles);
+	complete(&hyp_clock->ready);
+
+resched:
+	schedule_delayed_work(&hyp_clock->work,
+			      msecs_to_jiffies(CLOCK_UPDATE_MS));
+}
+
+static void hyp_clock_start(struct hyp_trace_buffer *hyp_buffer)
+{
+	struct hyp_trace_clock *hyp_clock = &hyp_buffer->clock;
+	struct system_time_snapshot snap;
+
+	ktime_get_snapshot(&snap);
+
+	hyp_clock->boot = ktime_to_ns(snap.boot);
+	hyp_clock->cycles = snap.cycles;
+	hyp_clock->mult = 0;
+
+	init_completion(&hyp_clock->ready);
+	INIT_DELAYED_WORK(&hyp_clock->work, __hyp_clock_work);
+	schedule_delayed_work(&hyp_clock->work, msecs_to_jiffies(CLOCK_INIT_MS));
+}
+
+static void hyp_clock_stop(struct hyp_trace_buffer *hyp_buffer)
+{
+	struct hyp_trace_clock *hyp_clock = &hyp_buffer->clock;
+
+	cancel_delayed_work_sync(&hyp_clock->work);
+}
+
+static void hyp_clock_wait(struct hyp_trace_buffer *hyp_buffer)
+{
+	struct hyp_trace_clock *hyp_clock = &hyp_buffer->clock;
+
+	wait_for_completion(&hyp_clock->ready);
+}
+
 static int __get_reader_page(int cpu)
 {
 	return kvm_call_hyp_nvhe(__pkvm_swap_reader_tracing, cpu);
@@ -297,10 +418,14 @@ static int hyp_trace_start(void)
 	if (hyp_buffer->tracing_on)
 		goto out;
 
+	hyp_clock_start(hyp_buffer);
+
 	ret = hyp_trace_buffer_load(hyp_buffer, hyp_trace_buffer_size);
 	if (ret)
 		goto out;
 
+	hyp_clock_wait(hyp_buffer);
+
 	ret = kvm_call_hyp_nvhe(__pkvm_enable_tracing, true);
 	if (ret) {
 		hyp_trace_buffer_teardown(hyp_buffer);
@@ -310,6 +435,9 @@ static int hyp_trace_start(void)
 	hyp_buffer->tracing_on = true;
 
 out:
+	if (!hyp_buffer->tracing_on)
+		hyp_clock_stop(hyp_buffer);
+
 	mutex_unlock(&hyp_buffer->lock);
 
 	return ret;
@@ -329,6 +457,7 @@ static void hyp_trace_stop(void)
 	if (!ret) {
 		ring_buffer_poll_remote(hyp_buffer->trace_buffer,
 					RING_BUFFER_ALL_CPUS);
+		hyp_clock_stop(hyp_buffer);
 		hyp_buffer->tracing_on = false;
 		hyp_trace_buffer_teardown(hyp_buffer);
 	}
@@ -617,6 +746,14 @@ static const struct file_operations hyp_trace_pipe_fops = {
 	.release        = hyp_trace_pipe_release,
 };
 
+static int hyp_trace_clock_show(struct seq_file *m, void *v)
+{
+	seq_puts(m, "[boot]\n");
+
+	return 0;
+}
+DEFINE_SHOW_ATTRIBUTE(hyp_trace_clock);
+
 int hyp_trace_init_tracefs(void)
 {
 	struct dentry *root, *per_cpu_root;
@@ -641,6 +778,9 @@ int hyp_trace_init_tracefs(void)
 	tracefs_create_file("trace_pipe", TRACEFS_MODE_WRITE, root,
 			    (void *)RING_BUFFER_ALL_CPUS, &hyp_trace_pipe_fops);
 
+	tracefs_create_file("trace_clock", TRACEFS_MODE_READ, root, NULL,
+			    &hyp_trace_clock_fops);
+
 	per_cpu_root = tracefs_create_dir("per_cpu", root);
 	if (!per_cpu_root) {
 		pr_err("Failed to create tracefs folder "TRACEFS_DIR"/per_cpu/\n");
-- 
2.48.1.601.g30ceb7b040-goog

[PATCH 08/11] KVM: arm64: Add raw interface for hyp tracefs

[Vincent Donnefort]

The raw interface enables userspace tools such as trace-cmd to directly
read the ring-buffer without any decoding by the kernel.

Signed-off-by: Vincent Donnefort <vdonnefort@google.com>

diff --git a/arch/arm64/kvm/hyp_trace.c b/arch/arm64/kvm/hyp_trace.c
index 3f91ad69c25b..38d97e34eada 100644
--- a/arch/arm64/kvm/hyp_trace.c
+++ b/arch/arm64/kvm/hyp_trace.c
@@ -746,6 +746,85 @@ static const struct file_operations hyp_trace_pipe_fops = {
 	.release        = hyp_trace_pipe_release,
 };
 
+static ssize_t
+hyp_trace_raw_read(struct file *file, char __user *ubuf,
+		   size_t cnt, loff_t *ppos)
+{
+	struct ht_iterator *iter = (struct ht_iterator *)file->private_data;
+	size_t size;
+	int ret;
+
+	if (iter->copy_leftover)
+		goto read;
+
+again:
+	ret = ring_buffer_read_page(iter->trace_buffer,
+				    (struct buffer_data_read_page *)iter->spare,
+				    cnt, iter->cpu, 0);
+	if (ret < 0) {
+		if (!ring_buffer_empty_cpu(iter->trace_buffer, iter->cpu))
+			return 0;
+
+		ret = ring_buffer_wait(iter->trace_buffer, iter->cpu, 0, NULL,
+				       NULL);
+		if (ret < 0)
+			return ret;
+
+		goto again;
+	}
+
+	iter->copy_leftover = 0;
+
+read:
+	size = PAGE_SIZE - iter->copy_leftover;
+	if (size > cnt)
+		size = cnt;
+
+	ret = copy_to_user(ubuf, iter->spare + PAGE_SIZE - size, size);
+	if (ret == size)
+		return -EFAULT;
+
+	size -= ret;
+	*ppos += size;
+	iter->copy_leftover = ret;
+
+	return size;
+}
+
+static int hyp_trace_raw_open(struct inode *inode, struct file *file)
+{
+	int ret = hyp_trace_pipe_open(inode, file);
+	struct ht_iterator *iter;
+
+	if (ret)
+		return ret;
+
+	iter = file->private_data;
+	iter->spare = ring_buffer_alloc_read_page(iter->trace_buffer, iter->cpu);
+	if (IS_ERR(iter->spare)) {
+		ret = PTR_ERR(iter->spare);
+		iter->spare = NULL;
+		return ret;
+	}
+
+	return 0;
+}
+
+static int hyp_trace_raw_release(struct inode *inode, struct file *file)
+{
+	struct ht_iterator *iter = file->private_data;
+
+	ring_buffer_free_read_page(iter->trace_buffer, iter->cpu, iter->spare);
+
+	return hyp_trace_pipe_release(inode, file);
+}
+
+static const struct file_operations hyp_trace_raw_fops = {
+	.open           = hyp_trace_raw_open,
+	.read           = hyp_trace_raw_read,
+	.release        = hyp_trace_raw_release,
+};
+
 static int hyp_trace_clock_show(struct seq_file *m, void *v)
 {
 	seq_puts(m, "[boot]\n");
@@ -800,6 +879,9 @@ int hyp_trace_init_tracefs(void)
 
 		tracefs_create_file("trace_pipe", TRACEFS_MODE_READ, per_cpu_dir,
 				    (void *)cpu, &hyp_trace_pipe_fops);
+
+		tracefs_create_file("trace_pipe_raw", TRACEFS_MODE_READ, per_cpu_dir,
+				    (void *)cpu, &hyp_trace_pipe_fops);
 	}
 
 	return 0;
-- 
2.48.1.601.g30ceb7b040-goog

[PATCH 09/11] KVM: arm64: Add trace interface for hyp tracefs

[Vincent Donnefort]

The trace interface is solely here to reset tracing. Non-consuming read
is not yet supported due to the lack of support in the ring-buffer
meta page.

Signed-off-by: Vincent Donnefort <vdonnefort@google.com>

diff --git a/arch/arm64/include/asm/kvm_asm.h b/arch/arm64/include/asm/kvm_asm.h
index 87d3e0e73b68..74f10847a55e 100644
--- a/arch/arm64/include/asm/kvm_asm.h
+++ b/arch/arm64/include/asm/kvm_asm.h
@@ -91,6 +91,7 @@ enum __kvm_host_smccc_func {
 	__KVM_HOST_SMCCC_FUNC___pkvm_load_tracing,
 	__KVM_HOST_SMCCC_FUNC___pkvm_teardown_tracing,
 	__KVM_HOST_SMCCC_FUNC___pkvm_enable_tracing,
+	__KVM_HOST_SMCCC_FUNC___pkvm_reset_tracing,
 	__KVM_HOST_SMCCC_FUNC___pkvm_swap_reader_tracing,
 };
 
diff --git a/arch/arm64/kvm/hyp/include/nvhe/trace.h b/arch/arm64/kvm/hyp/include/nvhe/trace.h
index 6f1cc571b47a..28bbb54b7a0b 100644
--- a/arch/arm64/kvm/hyp/include/nvhe/trace.h
+++ b/arch/arm64/kvm/hyp/include/nvhe/trace.h
@@ -20,6 +20,7 @@ void __pkvm_update_clock_tracing(u32 mult, u32 shift, u64 epoch_ns, u64 epoch_cy
 int __pkvm_load_tracing(unsigned long desc_va, size_t desc_size);
 void __pkvm_teardown_tracing(void);
 int __pkvm_enable_tracing(bool enable);
+int __pkvm_reset_tracing(unsigned int cpu);
 int __pkvm_swap_reader_tracing(unsigned int cpu);
 #else
 static inline void *tracing_reserve_entry(unsigned long length) { return NULL; }
@@ -30,6 +31,7 @@ void __pkvm_update_clock_tracing(u32 mult, u32 shift, u64 epoch_ns, u64 epoch_cy
 static inline int __pkvm_load_tracing(unsigned long desc_va, size_t desc_size) { return -ENODEV; }
 static inline void __pkvm_teardown_tracing(void) { }
 static inline int __pkvm_enable_tracing(bool enable) { return -ENODEV; }
+static inline int __pkvm_reset_tracing(unsigned int cpu) { return -ENODEV; }
 static inline int __pkvm_swap_reader_tracing(unsigned int cpu) { return -ENODEV; }
 #endif
 #endif
diff --git a/arch/arm64/kvm/hyp/nvhe/hyp-main.c b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
index a8b497b22407..e2419c97c57d 100644
--- a/arch/arm64/kvm/hyp/nvhe/hyp-main.c
+++ b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
@@ -605,6 +605,13 @@ static void handle___pkvm_enable_tracing(struct kvm_cpu_context *host_ctxt)
 	cpu_reg(host_ctxt, 1) = __pkvm_enable_tracing(enable);
 }
 
+static void handle___pkvm_reset_tracing(struct kvm_cpu_context *host_ctxt)
+{
+	DECLARE_REG(unsigned int, cpu, host_ctxt, 1);
+
+	cpu_reg(host_ctxt, 1) = __pkvm_reset_tracing(cpu);
+}
+
 static void handle___pkvm_swap_reader_tracing(struct kvm_cpu_context *host_ctxt)
 {
 	DECLARE_REG(unsigned int, cpu, host_ctxt, 1);
@@ -655,6 +662,7 @@ static const hcall_t host_hcall[] = {
 	HANDLE_FUNC(__pkvm_load_tracing),
 	HANDLE_FUNC(__pkvm_teardown_tracing),
 	HANDLE_FUNC(__pkvm_enable_tracing),
+	HANDLE_FUNC(__pkvm_reset_tracing),
 	HANDLE_FUNC(__pkvm_swap_reader_tracing),
 };
 
diff --git a/arch/arm64/kvm/hyp/nvhe/trace.c b/arch/arm64/kvm/hyp/nvhe/trace.c
index d79b6539377e..bf935645ed91 100644
--- a/arch/arm64/kvm/hyp/nvhe/trace.c
+++ b/arch/arm64/kvm/hyp/nvhe/trace.c
@@ -262,7 +262,7 @@ void tracing_commit_entry(void)
 	smp_store_release(&cpu_buffer->status, HYP_RB_READY);
 }
 
-static void hyp_rb_disable_writing(struct hyp_rb_per_cpu *cpu_buffer)
+static u32 hyp_rb_disable_writing(struct hyp_rb_per_cpu *cpu_buffer)
 {
 	u32 prev_status;
 
@@ -272,6 +272,8 @@ static void hyp_rb_disable_writing(struct hyp_rb_per_cpu *cpu_buffer)
 					      HYP_RB_READY,
 					      HYP_RB_UNAVAILABLE);
 	} while (prev_status == HYP_RB_WRITING);
+
+	return prev_status;
 }
 
 static int hyp_rb_enable_writing(struct hyp_rb_per_cpu *cpu_buffer)
@@ -284,6 +286,44 @@ static int hyp_rb_enable_writing(struct hyp_rb_per_cpu *cpu_buffer)
 	return 0;
 }
 
+static int hyp_rb_reset(struct hyp_rb_per_cpu *cpu_buffer)
+{
+	struct hyp_buffer_page *bpage;
+	u32 prev_status;
+
+	if (!hyp_rb_loaded(cpu_buffer))
+		return -ENODEV;
+
+	prev_status = hyp_rb_disable_writing(cpu_buffer);
+
+	while (!hyp_bpage_is_head(cpu_buffer->head_page))
+		cpu_buffer->head_page = hyp_bpage_next_page(cpu_buffer->head_page);
+
+	bpage = cpu_buffer->tail_page = cpu_buffer->head_page;
+	do {
+		hyp_bpage_reset(bpage);
+		bpage = hyp_bpage_next_page(bpage);
+	} while (bpage != cpu_buffer->head_page);
+
+	hyp_bpage_reset(cpu_buffer->reader_page);
+
+	cpu_buffer->last_overrun = 0;
+	cpu_buffer->write_stamp = 0;
+
+	cpu_buffer->meta->reader.read = 0;
+	cpu_buffer->meta->reader.lost_events = 0;
+	cpu_buffer->meta->entries = 0;
+	cpu_buffer->meta->overrun = 0;
+	cpu_buffer->meta->read = 0;
+	meta_pages_lost(cpu_buffer->meta) = 0;
+	meta_pages_touched(cpu_buffer->meta) = 0;
+
+	if (prev_status == HYP_RB_READY)
+		hyp_rb_enable_writing(cpu_buffer);
+
+	return 0;
+}
+
 static void hyp_rb_teardown(struct hyp_rb_per_cpu *cpu_buffer)
 {
 	int i;
@@ -572,3 +612,17 @@ int __pkvm_enable_tracing(bool enable)
 
 	return ret;
 }
+
+int __pkvm_reset_tracing(unsigned int cpu)
+{
+	int ret = 0;
+
+	if (cpu >= hyp_nr_cpus)
+		return -EINVAL;
+
+	hyp_spin_lock(&trace_rb_lock);
+	ret = hyp_rb_reset(per_cpu_ptr(&trace_rb, cpu));
+	hyp_spin_unlock(&trace_rb_lock);
+
+	return ret;
+}
diff --git a/arch/arm64/kvm/hyp_trace.c b/arch/arm64/kvm/hyp_trace.c
index 38d97e34eada..03a6813cbe66 100644
--- a/arch/arm64/kvm/hyp_trace.c
+++ b/arch/arm64/kvm/hyp_trace.c
@@ -200,6 +200,11 @@ static int __get_reader_page(int cpu)
 	return kvm_call_hyp_nvhe(__pkvm_swap_reader_tracing, cpu);
 }
 
+static int __reset(int cpu)
+{
+	return kvm_call_hyp_nvhe(__pkvm_reset_tracing, cpu);
+}
+
 static void hyp_trace_free_pages(struct hyp_trace_desc *desc)
 {
 	struct rb_page_desc *rb_desc;
@@ -361,6 +366,7 @@ static int hyp_trace_buffer_load(struct hyp_trace_buffer *hyp_buffer, size_t siz
 
 	hyp_buffer->remote.pdesc = &desc->page_desc;
 	hyp_buffer->remote.get_reader_page = __get_reader_page;
+	hyp_buffer->remote.reset = __reset;
 	hyp_buffer->trace_buffer = ring_buffer_remote(&hyp_buffer->remote);
 	if (!hyp_buffer->trace_buffer) {
 		ret = -ENOMEM;
@@ -825,6 +831,49 @@ static const struct file_operations hyp_trace_raw_fops = {
 	.release        = hyp_trace_raw_release,
 };
 
+static void hyp_trace_reset(int cpu)
+{
+	struct hyp_trace_buffer *hyp_buffer = &hyp_trace_buffer;
+
+	mutex_lock(&hyp_buffer->lock);
+
+	if (!hyp_trace_buffer_loaded(hyp_buffer))
+		goto out;
+
+	if (cpu == RING_BUFFER_ALL_CPUS)
+		ring_buffer_reset(hyp_buffer->trace_buffer);
+	else
+		ring_buffer_reset_cpu(hyp_buffer->trace_buffer, cpu);
+
+out:
+	mutex_unlock(&hyp_buffer->lock);
+}
+
+static int hyp_trace_open(struct inode *inode, struct file *file)
+{
+	int cpu = (s64)inode->i_private;
+
+	if (file->f_mode & FMODE_WRITE) {
+		hyp_trace_reset(cpu);
+
+		return 0;
+	}
+
+	return -EPERM;
+}
+
+static ssize_t hyp_trace_write(struct file *filp, const char __user *ubuf,
+			       size_t count, loff_t *ppos)
+{
+	return count;
+}
+
+static const struct file_operations hyp_trace_fops = {
+	.open           = hyp_trace_open,
+	.write          = hyp_trace_write,
+	.release        = NULL,
+};
+
 static int hyp_trace_clock_show(struct seq_file *m, void *v)
 {
 	seq_puts(m, "[boot]\n");
@@ -857,6 +906,9 @@ int hyp_trace_init_tracefs(void)
 	tracefs_create_file("trace_pipe", TRACEFS_MODE_WRITE, root,
 			    (void *)RING_BUFFER_ALL_CPUS, &hyp_trace_pipe_fops);
 
+	tracefs_create_file("trace", TRACEFS_MODE_WRITE, root,
+			    (void *)RING_BUFFER_ALL_CPUS, &hyp_trace_fops);
+
 	tracefs_create_file("trace_clock", TRACEFS_MODE_READ, root, NULL,
 			    &hyp_trace_clock_fops);
 
@@ -882,6 +934,9 @@ int hyp_trace_init_tracefs(void)
 
 		tracefs_create_file("trace_pipe_raw", TRACEFS_MODE_READ, per_cpu_dir,
 				    (void *)cpu, &hyp_trace_pipe_fops);
+
+		tracefs_create_file("trace", TRACEFS_MODE_READ, per_cpu_dir,
+				    (void *)cpu, &hyp_trace_fops);
 	}
 
 	return 0;
-- 
2.48.1.601.g30ceb7b040-goog

[PATCH 10/11] KVM: arm64: Add support for hyp events

[Vincent Donnefort]

Following the introduction of hyp tracing for pKVM, add the ability to
describe and emit events into the hypervisor ring-buffers.

Hypervisor events are declared into kvm_hypevents.h and can be called
with trace_<event_name>() in a similar fashion to the kernel tracefs
events.

hyp_enter and hyp_exit events are provided as an example.

Signed-off-by: Vincent Donnefort <vdonnefort@google.com>

diff --git a/arch/arm64/include/asm/kvm_asm.h b/arch/arm64/include/asm/kvm_asm.h
index 74f10847a55e..afeb983ca97b 100644
--- a/arch/arm64/include/asm/kvm_asm.h
+++ b/arch/arm64/include/asm/kvm_asm.h
@@ -93,6 +93,7 @@ enum __kvm_host_smccc_func {
 	__KVM_HOST_SMCCC_FUNC___pkvm_enable_tracing,
 	__KVM_HOST_SMCCC_FUNC___pkvm_reset_tracing,
 	__KVM_HOST_SMCCC_FUNC___pkvm_swap_reader_tracing,
+	__KVM_HOST_SMCCC_FUNC___pkvm_enable_event,
 };
 
 #define DECLARE_KVM_VHE_SYM(sym)	extern char sym[]
diff --git a/arch/arm64/include/asm/kvm_define_hypevents.h b/arch/arm64/include/asm/kvm_define_hypevents.h
new file mode 100644
index 000000000000..cda7dc27dba7
--- /dev/null
+++ b/arch/arm64/include/asm/kvm_define_hypevents.h
@@ -0,0 +1,61 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#include <linux/trace_events.h>
+
+#include <asm/kvm_hyptrace.h>
+#include <asm/kvm_hypevents_defs.h>
+
+#ifndef HYP_EVENT_FILE
+# undef __ARM64_KVM_HYPEVENTS_H_
+# define  __HYP_EVENT_FILE <asm/kvm_hypevents.h>
+#else
+# define __HYP_EVENT_FILE __stringify(HYP_EVENT_FILE)
+#endif
+
+#define HYP_EVENT(__name, __proto, __struct, __assign, __printk)		\
+	HYP_EVENT_FORMAT(__name, __struct);					\
+	static void hyp_event_trace_##__name(struct ht_iterator *iter)		\
+	{									\
+		struct trace_hyp_format_##__name __maybe_unused *__entry =	\
+			(struct trace_hyp_format_##__name *)iter->ent;		\
+		trace_seq_puts(&iter->seq, #__name);				\
+		trace_seq_putc(&iter->seq, ' ');				\
+		trace_seq_printf(&iter->seq, __printk);				\
+		trace_seq_putc(&iter->seq, '\n');				\
+	}
+#define HYP_EVENT_MULTI_READ
+#include __HYP_EVENT_FILE
+
+#undef he_field
+#define he_field(_type, _item)						\
+	{								\
+		.type = #_type, .name = #_item,				\
+		.size = sizeof(_type), .align = __alignof__(_type),	\
+		.is_signed = is_signed_type(_type),			\
+	},
+#undef HYP_EVENT
+#define HYP_EVENT(__name, __proto, __struct, __assign, __printk)		\
+	static struct trace_event_fields hyp_event_fields_##__name[] = {	\
+		__struct							\
+		{}								\
+	};
+#include __HYP_EVENT_FILE
+
+#undef HYP_EVENT
+#undef HE_PRINTK
+#define __entry REC
+#define HE_PRINTK(fmt, args...) "\"" fmt "\", " __stringify(args)
+#define HYP_EVENT(__name, __proto, __struct, __assign, __printk)	\
+	static char hyp_event_print_fmt_##__name[] = __printk;		\
+	static bool hyp_event_enabled_##__name;				\
+	struct hyp_event __section("_hyp_events."#__name)		\
+	hyp_event_##__name = {						\
+		.name		= #__name,				\
+		.enabled	= &hyp_event_enabled_##__name,		\
+		.fields		= hyp_event_fields_##__name,		\
+		.print_fmt	= hyp_event_print_fmt_##__name,		\
+		.trace_func	= hyp_event_trace_##__name,		\
+	}
+#include __HYP_EVENT_FILE
+
+#undef HYP_EVENT_MULTI_READ
diff --git a/arch/arm64/include/asm/kvm_hypevents.h b/arch/arm64/include/asm/kvm_hypevents.h
new file mode 100644
index 000000000000..0b98a87a1250
--- /dev/null
+++ b/arch/arm64/include/asm/kvm_hypevents.h
@@ -0,0 +1,31 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#if !defined(__ARM64_KVM_HYPEVENTS_H_) || defined(HYP_EVENT_MULTI_READ)
+#define __ARM64_KVM_HYPEVENTS_H_
+
+#ifdef __KVM_NVHE_HYPERVISOR__
+#include <nvhe/trace.h>
+#endif
+
+/*
+ * Hypervisor events definitions.
+ */
+
+HYP_EVENT(hyp_enter,
+	HE_PROTO(void),
+	HE_STRUCT(
+	),
+	HE_ASSIGN(
+	),
+	HE_PRINTK(" ")
+);
+
+HYP_EVENT(hyp_exit,
+	HE_PROTO(void),
+	HE_STRUCT(
+	),
+	HE_ASSIGN(
+	),
+	HE_PRINTK(" ")
+);
+#endif
diff --git a/arch/arm64/include/asm/kvm_hypevents_defs.h b/arch/arm64/include/asm/kvm_hypevents_defs.h
new file mode 100644
index 000000000000..473bf4363d82
--- /dev/null
+++ b/arch/arm64/include/asm/kvm_hypevents_defs.h
@@ -0,0 +1,41 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifndef __ARM64_KVM_HYPEVENTS_DEFS_H
+#define __ARM64_KVM_HYPEVENTS_DEFS_H
+
+struct hyp_event_id {
+	unsigned short id;
+	void *data;
+};
+
+#define HYP_EVENT_NAME_MAX 32
+
+struct hyp_event {
+	char				name[HYP_EVENT_NAME_MAX];
+	bool				*enabled;
+	char				*print_fmt;
+	struct trace_event_fields	*fields;
+	void (*trace_func)(struct ht_iterator *iter);
+	int				id;
+};
+
+struct hyp_entry_hdr {
+	unsigned short id;
+};
+
+/*
+ * Hyp events definitions common to the hyp and the host
+ */
+#define HYP_EVENT_FORMAT(__name, __struct)		\
+	struct __packed trace_hyp_format_##__name {	\
+		struct hyp_entry_hdr hdr;		\
+		__struct				\
+	}
+
+#define HE_PROTO(args...)	args
+#define HE_STRUCT(args...)	args
+#define HE_ASSIGN(args...)	args
+#define HE_PRINTK(args...)	args
+
+#define he_field(type, item)	type item;
+#endif
diff --git a/arch/arm64/include/asm/kvm_hyptrace.h b/arch/arm64/include/asm/kvm_hyptrace.h
index 7da6a248c7fa..7b66bd06537f 100644
--- a/arch/arm64/include/asm/kvm_hyptrace.h
+++ b/arch/arm64/include/asm/kvm_hyptrace.h
@@ -4,6 +4,22 @@
 #include <asm/kvm_hyp.h>
 
 #include <linux/ring_buffer.h>
+#include <linux/trace_seq.h>
+#include <linux/workqueue.h>
+
+struct ht_iterator {
+	struct trace_buffer	*trace_buffer;
+	int			cpu;
+	struct hyp_entry_hdr	*ent;
+	unsigned long		lost_events;
+	int			ent_cpu;
+	size_t			ent_size;
+	u64			ts;
+	void			*spare;
+	size_t			copy_leftover;
+	struct trace_seq        seq;
+	struct delayed_work     poll_work;
+};
 
 /*
  * Host donations to the hypervisor to store the struct hyp_buffer_page.
diff --git a/arch/arm64/kernel/image-vars.h b/arch/arm64/kernel/image-vars.h
index ef3a69cc398e..3b8bf8ded48c 100644
--- a/arch/arm64/kernel/image-vars.h
+++ b/arch/arm64/kernel/image-vars.h
@@ -137,6 +137,10 @@ KVM_NVHE_ALIAS(__hyp_bss_start);
 KVM_NVHE_ALIAS(__hyp_bss_end);
 KVM_NVHE_ALIAS(__hyp_rodata_start);
 KVM_NVHE_ALIAS(__hyp_rodata_end);
+#ifdef CONFIG_TRACING
+KVM_NVHE_ALIAS(__hyp_event_ids_start);
+KVM_NVHE_ALIAS(__hyp_event_ids_end);
+#endif
 
 /* pKVM static key */
 KVM_NVHE_ALIAS(kvm_protected_mode_initialized);
diff --git a/arch/arm64/kernel/vmlinux.lds.S b/arch/arm64/kernel/vmlinux.lds.S
index e73326bd3ff7..14e52049c4e5 100644
--- a/arch/arm64/kernel/vmlinux.lds.S
+++ b/arch/arm64/kernel/vmlinux.lds.S
@@ -13,12 +13,23 @@
 	*(__kvm_ex_table)					\
 	__stop___kvm_ex_table = .;
 
+#ifdef CONFIG_TRACING
+#define HYPERVISOR_EVENT_IDS 					\
+	. = ALIGN(PAGE_SIZE);					\
+	__hyp_event_ids_start = .;				\
+	*(HYP_SECTION_NAME(.event_ids))				\
+	__hyp_event_ids_end = .;
+#else
+#define HYPERVISOR_EVENT_IDS
+#endif
+
 #define HYPERVISOR_DATA_SECTIONS				\
 	HYP_SECTION_NAME(.rodata) : {				\
 		. = ALIGN(PAGE_SIZE);				\
 		__hyp_rodata_start = .;				\
 		*(HYP_SECTION_NAME(.data..ro_after_init))	\
 		*(HYP_SECTION_NAME(.rodata))			\
+		HYPERVISOR_EVENT_IDS				\
 		. = ALIGN(PAGE_SIZE);				\
 		__hyp_rodata_end = .;				\
 	}
@@ -201,6 +212,13 @@ SECTIONS
 	ASSERT(SIZEOF(.got.plt) == 0 || SIZEOF(.got.plt) == 0x18,
 	       "Unexpected GOT/PLT entries detected!")
 
+#ifdef CONFIG_TRACING
+	.rodata.hyp_events : {
+		__hyp_events_start = .;
+		*(SORT(_hyp_events.*))
+		__hyp_events_end = .;
+	}
+#endif
 	/* code sections that are never executed via the kernel mapping */
 	.rodata.text : {
 		TRAMP_TEXT
diff --git a/arch/arm64/kvm/Makefile b/arch/arm64/kvm/Makefile
index 865971bb8905..f9e208273031 100644
--- a/arch/arm64/kvm/Makefile
+++ b/arch/arm64/kvm/Makefile
@@ -29,7 +29,7 @@ kvm-$(CONFIG_HW_PERF_EVENTS)  += pmu-emul.o pmu.o
 kvm-$(CONFIG_ARM64_PTR_AUTH)  += pauth.o
 kvm-$(CONFIG_PTDUMP_STAGE2_DEBUGFS) += ptdump.o
 
-kvm-$(CONFIG_TRACING) += hyp_trace.o
+kvm-$(CONFIG_TRACING) += hyp_events.o hyp_trace.o
 
 always-y := hyp_constants.h hyp-constants.s
 
diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c
index f3951d36b9c1..2f1b869efc80 100644
--- a/arch/arm64/kvm/arm.c
+++ b/arch/arm64/kvm/arm.c
@@ -2644,6 +2644,8 @@ static int __init init_hyp_mode(void)
 
 	kvm_hyp_init_symbols();
 
+	hyp_trace_init_events();
+
 	if (is_protected_kvm_enabled()) {
 		if (IS_ENABLED(CONFIG_ARM64_PTR_AUTH_KERNEL) &&
 		    cpus_have_final_cap(ARM64_HAS_ADDRESS_AUTH))
diff --git a/arch/arm64/kvm/hyp/include/nvhe/arm-smccc.h b/arch/arm64/kvm/hyp/include/nvhe/arm-smccc.h
new file mode 100644
index 000000000000..4b69d33e4f2d
--- /dev/null
+++ b/arch/arm64/kvm/hyp/include/nvhe/arm-smccc.h
@@ -0,0 +1,13 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+
+#include <asm/kvm_hypevents.h>
+
+#include <linux/arm-smccc.h>
+
+#undef arm_smccc_1_1_smc
+#define arm_smccc_1_1_smc(...)					\
+	do {							\
+		trace_hyp_exit();				\
+		__arm_smccc_1_1(SMCCC_SMC_INST, __VA_ARGS__);	\
+		trace_hyp_enter();				\
+	} while (0)
diff --git a/arch/arm64/kvm/hyp/include/nvhe/define_events.h b/arch/arm64/kvm/hyp/include/nvhe/define_events.h
new file mode 100644
index 000000000000..2298b49cb355
--- /dev/null
+++ b/arch/arm64/kvm/hyp/include/nvhe/define_events.h
@@ -0,0 +1,21 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifndef HYP_EVENT_FILE
+# define __HYP_EVENT_FILE <asm/kvm_hypevents.h>
+#else
+# define __HYP_EVENT_FILE __stringify(HYP_EVENT_FILE)
+#endif
+
+#undef HYP_EVENT
+#define HYP_EVENT(__name, __proto, __struct, __assign, __printk)	\
+	atomic_t __ro_after_init __name##_enabled = ATOMIC_INIT(0);	\
+	struct hyp_event_id hyp_event_id_##__name			\
+	__section(".hyp.event_ids."#__name) = {				\
+		.data = (void *)&__name##_enabled,			\
+	}
+
+#define HYP_EVENT_MULTI_READ
+#include __HYP_EVENT_FILE
+#undef HYP_EVENT_MULTI_READ
+
+#undef HYP_EVENT
diff --git a/arch/arm64/kvm/hyp/include/nvhe/trace.h b/arch/arm64/kvm/hyp/include/nvhe/trace.h
index 28bbb54b7a0b..1bf9c5e61aee 100644
--- a/arch/arm64/kvm/hyp/include/nvhe/trace.h
+++ b/arch/arm64/kvm/hyp/include/nvhe/trace.h
@@ -2,6 +2,7 @@
 #ifndef __ARM64_KVM_HYP_NVHE_TRACE_H
 #define __ARM64_KVM_HYP_NVHE_TRACE_H
 #include <asm/kvm_hyptrace.h>
+#include <asm/kvm_hypevents_defs.h>
 
 /* Internal struct exported for hyp-constants.c */
 struct hyp_buffer_page {
@@ -15,6 +16,24 @@ struct hyp_buffer_page {
 #ifdef CONFIG_TRACING
 void *tracing_reserve_entry(unsigned long length);
 void tracing_commit_entry(void);
+#define HYP_EVENT(__name, __proto, __struct, __assign, __printk)		\
+	HYP_EVENT_FORMAT(__name, __struct);					\
+	extern atomic_t __name##_enabled;					\
+	extern struct hyp_event_id hyp_event_id_##__name;			\
+	static __always_inline void trace_##__name(__proto)			\
+	{									\
+		size_t length = sizeof(struct trace_hyp_format_##__name);	\
+		struct trace_hyp_format_##__name *__entry;			\
+										\
+		if (!atomic_read(&__name##_enabled))				\
+			return;							\
+		__entry = tracing_reserve_entry(length);			\
+		if (!__entry)							\
+			return;							\
+		__entry->hdr.id = hyp_event_id_##__name.id;			\
+		__assign							\
+		tracing_commit_entry();						\
+	}
 
 void __pkvm_update_clock_tracing(u32 mult, u32 shift, u64 epoch_ns, u64 epoch_cyc);
 int __pkvm_load_tracing(unsigned long desc_va, size_t desc_size);
@@ -22,9 +41,12 @@ void __pkvm_teardown_tracing(void);
 int __pkvm_enable_tracing(bool enable);
 int __pkvm_reset_tracing(unsigned int cpu);
 int __pkvm_swap_reader_tracing(unsigned int cpu);
+int __pkvm_enable_event(unsigned short id, bool enable);
 #else
 static inline void *tracing_reserve_entry(unsigned long length) { return NULL; }
 static inline void tracing_commit_entry(void) { }
+#define HYP_EVENT(__name, __proto, __struct, __assign, __printk)      \
+	static inline void trace_##__name(__proto) {}
 
 static inline
 void __pkvm_update_clock_tracing(u32 mult, u32 shift, u64 epoch_ns, u64 epoch_cyc) { }
@@ -33,5 +55,6 @@ static inline void __pkvm_teardown_tracing(void) { }
 static inline int __pkvm_enable_tracing(bool enable) { return -ENODEV; }
 static inline int __pkvm_reset_tracing(unsigned int cpu) { return -ENODEV; }
 static inline int __pkvm_swap_reader_tracing(unsigned int cpu) { return -ENODEV; }
+static inline int __pkvm_enable_event(unsigned short id, bool enable)  { return -ENODEV; }
 #endif
 #endif
diff --git a/arch/arm64/kvm/hyp/nvhe/Makefile b/arch/arm64/kvm/hyp/nvhe/Makefile
index 40f243c44cf5..fc11e47a1e90 100644
--- a/arch/arm64/kvm/hyp/nvhe/Makefile
+++ b/arch/arm64/kvm/hyp/nvhe/Makefile
@@ -28,7 +28,7 @@ hyp-obj-y := timer-sr.o sysreg-sr.o debug-sr.o switch.o tlb.o hyp-init.o host.o
 hyp-obj-y += ../vgic-v3-sr.o ../aarch32.o ../vgic-v2-cpuif-proxy.o ../entry.o \
 	 ../fpsimd.o ../hyp-entry.o ../exception.o ../pgtable.o
 hyp-obj-$(CONFIG_LIST_HARDENED) += list_debug.o
-hyp-obj-$(CONFIG_TRACING) += clock.o trace.o
+hyp-obj-$(CONFIG_TRACING) += clock.o events.o trace.o
 hyp-obj-y += $(lib-objs)
 
 ##
diff --git a/arch/arm64/kvm/hyp/nvhe/events.c b/arch/arm64/kvm/hyp/nvhe/events.c
new file mode 100644
index 000000000000..5905b42cb0d0
--- /dev/null
+++ b/arch/arm64/kvm/hyp/nvhe/events.c
@@ -0,0 +1,36 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2025 Google LLC
+ * Author: Vincent Donnefort <vdonnefort@google.com>
+ */
+
+#include <nvhe/mm.h>
+#include <nvhe/trace.h>
+
+#include <nvhe/define_events.h>
+
+extern struct hyp_event_id __hyp_event_ids_start[];
+extern struct hyp_event_id __hyp_event_ids_end[];
+
+int __pkvm_enable_event(unsigned short id, bool enable)
+{
+	struct hyp_event_id *event_id = __hyp_event_ids_start;
+	atomic_t *enable_key;
+
+	for (; (unsigned long)event_id < (unsigned long)__hyp_event_ids_end;
+	     event_id++) {
+		if (event_id->id != id)
+			continue;
+
+		enable_key = (atomic_t *)event_id->data;
+		enable_key = hyp_fixmap_map(__hyp_pa(enable_key));
+
+		atomic_set(enable_key, enable);
+
+		hyp_fixmap_unmap();
+
+		return 0;
+	}
+
+	return -EINVAL;
+}
diff --git a/arch/arm64/kvm/hyp/nvhe/ffa.c b/arch/arm64/kvm/hyp/nvhe/ffa.c
index e433dfab882a..6c740f8dcf82 100644
--- a/arch/arm64/kvm/hyp/nvhe/ffa.c
+++ b/arch/arm64/kvm/hyp/nvhe/ffa.c
@@ -26,10 +26,10 @@
  * the duration and are therefore serialised.
  */
 
-#include <linux/arm-smccc.h>
 #include <linux/arm_ffa.h>
 #include <asm/kvm_pkvm.h>
 
+#include <nvhe/arm-smccc.h>
 #include <nvhe/ffa.h>
 #include <nvhe/mem_protect.h>
 #include <nvhe/memory.h>
diff --git a/arch/arm64/kvm/hyp/nvhe/hyp-main.c b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
index e2419c97c57d..96dde58f4984 100644
--- a/arch/arm64/kvm/hyp/nvhe/hyp-main.c
+++ b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
@@ -12,6 +12,7 @@
 #include <asm/kvm_emulate.h>
 #include <asm/kvm_host.h>
 #include <asm/kvm_hyp.h>
+#include <asm/kvm_hypevents.h>
 #include <asm/kvm_mmu.h>
 
 #include <nvhe/ffa.h>
@@ -619,6 +620,14 @@ static void handle___pkvm_swap_reader_tracing(struct kvm_cpu_context *host_ctxt)
 	cpu_reg(host_ctxt, 1) = __pkvm_swap_reader_tracing(cpu);
 }
 
+static void handle___pkvm_enable_event(struct kvm_cpu_context *host_ctxt)
+{
+	DECLARE_REG(unsigned short, id, host_ctxt, 1);
+	DECLARE_REG(bool, enable, host_ctxt, 2);
+
+	cpu_reg(host_ctxt, 1) = __pkvm_enable_event(id, enable);
+}
+
 typedef void (*hcall_t)(struct kvm_cpu_context *);
 
 #define HANDLE_FUNC(x)	[__KVM_HOST_SMCCC_FUNC_##x] = (hcall_t)handle_##x
@@ -664,6 +673,7 @@ static const hcall_t host_hcall[] = {
 	HANDLE_FUNC(__pkvm_enable_tracing),
 	HANDLE_FUNC(__pkvm_reset_tracing),
 	HANDLE_FUNC(__pkvm_swap_reader_tracing),
+	HANDLE_FUNC(__pkvm_enable_event),
 };
 
 static void handle_host_hcall(struct kvm_cpu_context *host_ctxt)
@@ -704,7 +714,9 @@ static void handle_host_hcall(struct kvm_cpu_context *host_ctxt)
 
 static void default_host_smc_handler(struct kvm_cpu_context *host_ctxt)
 {
+	trace_hyp_exit();
 	__kvm_hyp_host_forward_smc(host_ctxt);
+	trace_hyp_enter();
 }
 
 static void handle_host_smc(struct kvm_cpu_context *host_ctxt)
@@ -728,6 +740,8 @@ void handle_trap(struct kvm_cpu_context *host_ctxt)
 {
 	u64 esr = read_sysreg_el2(SYS_ESR);
 
+	trace_hyp_enter();
+
 	switch (ESR_ELx_EC(esr)) {
 	case ESR_ELx_EC_HVC64:
 		handle_host_hcall(host_ctxt);
@@ -742,4 +756,6 @@ void handle_trap(struct kvm_cpu_context *host_ctxt)
 	default:
 		BUG();
 	}
+
+	trace_hyp_exit();
 }
diff --git a/arch/arm64/kvm/hyp/nvhe/hyp.lds.S b/arch/arm64/kvm/hyp/nvhe/hyp.lds.S
index f4562f417d3f..2f9262057bac 100644
--- a/arch/arm64/kvm/hyp/nvhe/hyp.lds.S
+++ b/arch/arm64/kvm/hyp/nvhe/hyp.lds.S
@@ -16,6 +16,12 @@ SECTIONS {
 	HYP_SECTION(.text)
 	HYP_SECTION(.data..ro_after_init)
 	HYP_SECTION(.rodata)
+#ifdef CONFIG_TRACING
+	. = ALIGN(PAGE_SIZE);
+	BEGIN_HYP_SECTION(.event_ids)
+		*(SORT(.hyp.event_ids.*))
+	END_HYP_SECTION
+#endif
 
 	/*
 	 * .hyp..data..percpu needs to be page aligned to maintain the same
diff --git a/arch/arm64/kvm/hyp/nvhe/psci-relay.c b/arch/arm64/kvm/hyp/nvhe/psci-relay.c
index 9c2ce1e0e99a..00bc2ab94d59 100644
--- a/arch/arm64/kvm/hyp/nvhe/psci-relay.c
+++ b/arch/arm64/kvm/hyp/nvhe/psci-relay.c
@@ -6,11 +6,12 @@
 
 #include <asm/kvm_asm.h>
 #include <asm/kvm_hyp.h>
+#include <asm/kvm_hypevents.h>
 #include <asm/kvm_mmu.h>
-#include <linux/arm-smccc.h>
 #include <linux/kvm_host.h>
 #include <uapi/linux/psci.h>
 
+#include <nvhe/arm-smccc.h>
 #include <nvhe/memory.h>
 #include <nvhe/trap_handler.h>
 
@@ -153,6 +154,7 @@ static int psci_cpu_suspend(u64 func_id, struct kvm_cpu_context *host_ctxt)
 	DECLARE_REG(u64, power_state, host_ctxt, 1);
 	DECLARE_REG(unsigned long, pc, host_ctxt, 2);
 	DECLARE_REG(unsigned long, r0, host_ctxt, 3);
+	int ret;
 
 	struct psci_boot_args *boot_args;
 	struct kvm_nvhe_init_params *init_params;
@@ -171,9 +173,11 @@ static int psci_cpu_suspend(u64 func_id, struct kvm_cpu_context *host_ctxt)
 	 * Will either return if shallow sleep state, or wake up into the entry
 	 * point if it is a deep sleep state.
 	 */
-	return psci_call(func_id, power_state,
-			 __hyp_pa(&kvm_hyp_cpu_resume),
-			 __hyp_pa(init_params));
+	ret = psci_call(func_id, power_state,
+			__hyp_pa(&kvm_hyp_cpu_resume),
+			__hyp_pa(init_params));
+
+	return ret;
 }
 
 static int psci_system_suspend(u64 func_id, struct kvm_cpu_context *host_ctxt)
@@ -205,6 +209,7 @@ asmlinkage void __noreturn __kvm_host_psci_cpu_entry(bool is_cpu_on)
 	struct psci_boot_args *boot_args;
 	struct kvm_cpu_context *host_ctxt;
 
+	trace_hyp_enter();
 	host_ctxt = host_data_ptr(host_ctxt);
 
 	if (is_cpu_on)
@@ -218,6 +223,7 @@ asmlinkage void __noreturn __kvm_host_psci_cpu_entry(bool is_cpu_on)
 	if (is_cpu_on)
 		release_boot_args(boot_args);
 
+	trace_hyp_exit();
 	__host_enter(host_ctxt);
 }
 
diff --git a/arch/arm64/kvm/hyp/nvhe/switch.c b/arch/arm64/kvm/hyp/nvhe/switch.c
index 7d2ba6ef0261..bbe035acda89 100644
--- a/arch/arm64/kvm/hyp/nvhe/switch.c
+++ b/arch/arm64/kvm/hyp/nvhe/switch.c
@@ -7,7 +7,6 @@
 #include <hyp/switch.h>
 #include <hyp/sysreg-sr.h>
 
-#include <linux/arm-smccc.h>
 #include <linux/kvm_host.h>
 #include <linux/types.h>
 #include <linux/jump_label.h>
@@ -21,6 +20,7 @@
 #include <asm/kvm_asm.h>
 #include <asm/kvm_emulate.h>
 #include <asm/kvm_hyp.h>
+#include <asm/kvm_hypevents.h>
 #include <asm/kvm_mmu.h>
 #include <asm/fpsimd.h>
 #include <asm/debug-monitors.h>
@@ -349,10 +349,13 @@ int __kvm_vcpu_run(struct kvm_vcpu *vcpu)
 	__debug_switch_to_guest(vcpu);
 
 	do {
+		trace_hyp_exit();
+
 		/* Jump in the fire! */
 		exit_code = __guest_enter(vcpu);
 
 		/* And we're baaack! */
+		trace_hyp_enter();
 	} while (fixup_guest_exit(vcpu, &exit_code));
 
 	__sysreg_save_state_nvhe(guest_ctxt);
diff --git a/arch/arm64/kvm/hyp_events.c b/arch/arm64/kvm/hyp_events.c
new file mode 100644
index 000000000000..ec56a63f3451
--- /dev/null
+++ b/arch/arm64/kvm/hyp_events.c
@@ -0,0 +1,159 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2023 Google LLC
+ */
+
+#include <linux/tracefs.h>
+
+#include <asm/kvm_host.h>
+#include <asm/kvm_define_hypevents.h>
+#include <asm/setup.h>
+
+#include "hyp_trace.h"
+
+extern struct hyp_event __hyp_events_start[];
+extern struct hyp_event __hyp_events_end[];
+
+/* hyp_event section used by the hypervisor */
+extern struct hyp_event_id __hyp_event_ids_start[];
+extern struct hyp_event_id __hyp_event_ids_end[];
+
+static ssize_t
+hyp_event_write(struct file *filp, const char __user *ubuf, size_t cnt, loff_t *ppos)
+{
+	struct seq_file *seq_file = (struct seq_file *)filp->private_data;
+	struct hyp_event *evt = (struct hyp_event *)seq_file->private;
+	unsigned short id = evt->id;
+	bool enabling;
+	int ret;
+	char c;
+
+	if (!cnt || cnt > 2)
+		return -EINVAL;
+
+	if (get_user(c, ubuf))
+		return -EFAULT;
+
+	switch (c) {
+	case '1':
+		enabling = true;
+		break;
+	case '0':
+		enabling = false;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	if (enabling != *evt->enabled) {
+		ret = kvm_call_hyp_nvhe(__pkvm_enable_event, id, enabling);
+		if (ret)
+			return ret;
+	}
+
+	*evt->enabled = enabling;
+
+	return cnt;
+}
+
+static int hyp_event_show(struct seq_file *m, void *v)
+{
+	struct hyp_event *evt = (struct hyp_event *)m->private;
+
+	seq_printf(m, "%d\n", *evt->enabled);
+
+	return 0;
+}
+
+static int hyp_event_open(struct inode *inode, struct file *filp)
+{
+	return single_open(filp, hyp_event_show, inode->i_private);
+}
+
+static const struct file_operations hyp_event_fops = {
+	.open		= hyp_event_open,
+	.write		= hyp_event_write,
+	.read		= seq_read,
+	.llseek		= seq_lseek,
+	.release	= single_release,
+};
+
+static int hyp_event_id_show(struct seq_file *m, void *v)
+{
+	struct hyp_event *evt = (struct hyp_event *)m->private;
+
+	seq_printf(m, "%d\n", evt->id);
+
+	return 0;
+}
+
+static int hyp_event_id_open(struct inode *inode, struct file *filp)
+{
+	return single_open(filp, hyp_event_id_show, inode->i_private);
+}
+
+static const struct file_operations hyp_event_id_fops = {
+	.open = hyp_event_id_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = single_release,
+};
+
+void hyp_trace_init_event_tracefs(struct dentry *parent)
+{
+	struct hyp_event *event = __hyp_events_start;
+
+	parent = tracefs_create_dir("events", parent);
+	if (!parent) {
+		pr_err("Failed to create tracefs folder for hyp events\n");
+		return;
+	}
+
+	parent = tracefs_create_dir("hypervisor", parent);
+	if (!parent) {
+		pr_err("Failed to create tracefs folder for hyp events\n");
+		return;
+	}
+
+	for (; (unsigned long)event < (unsigned long)__hyp_events_end; event++) {
+		struct dentry *event_dir = tracefs_create_dir(event->name, parent);
+
+		if (!event_dir) {
+			pr_err("Failed to create events/hypervisor/%s\n",
+			       event->name);
+			continue;
+		}
+
+		tracefs_create_file("enable", 0640, event_dir, (void *)event,
+				    &hyp_event_fops);
+		tracefs_create_file("id", 0440, event_dir, (void *)event,
+				    &hyp_event_id_fops);
+	}
+}
+
+struct hyp_event *hyp_trace_find_event(int id)
+{
+	struct hyp_event *event = __hyp_events_start + id;
+
+	if ((unsigned long)event >= (unsigned long)__hyp_events_end)
+		return NULL;
+
+	return event;
+}
+
+/*
+ * Register hyp events and write their id into the hyp section _hyp_event_ids.
+ */
+int hyp_trace_init_events(void)
+{
+	struct hyp_event_id *hyp_event_id = __hyp_event_ids_start;
+	struct hyp_event *event = __hyp_events_start;
+	int id = 0;
+
+	/* Events on both sides hypervisor are sorted */
+	for (; (unsigned long)event < (unsigned long)__hyp_events_end;
+		event++, hyp_event_id++, id++)
+		event->id = hyp_event_id->id = id;
+
+	return 0;
+}
diff --git a/arch/arm64/kvm/hyp_trace.c b/arch/arm64/kvm/hyp_trace.c
index 03a6813cbe66..cb63af69c38d 100644
--- a/arch/arm64/kvm/hyp_trace.c
+++ b/arch/arm64/kvm/hyp_trace.c
@@ -6,10 +6,12 @@
 
 #include <linux/arm-smccc.h>
 #include <linux/percpu-defs.h>
+#include <linux/trace_events.h>
 #include <linux/tracefs.h>
 
 #include <asm/kvm_host.h>
 #include <asm/kvm_hyptrace.h>
+#include <asm/kvm_hypevents_defs.h>
 
 #include "hyp_constants.h"
 #include "hyp_trace.h"
@@ -567,6 +569,8 @@ static void ht_print_trace_cpu(struct ht_iterator *iter)
 
 static int ht_print_trace_fmt(struct ht_iterator *iter)
 {
+	struct hyp_event *e;
+
 	if (iter->lost_events)
 		trace_seq_printf(&iter->seq, "CPU:%d [LOST %lu EVENTS]\n",
 				 iter->ent_cpu, iter->lost_events);
@@ -574,6 +578,12 @@ static int ht_print_trace_fmt(struct ht_iterator *iter)
 	ht_print_trace_cpu(iter);
 	ht_print_trace_time(iter);
 
+	e = hyp_trace_find_event(iter->ent->id);
+	if (e)
+		e->trace_func(iter);
+	else
+		trace_seq_printf(&iter->seq, "Unknown event id %d\n", iter->ent->id);
+
 	return trace_seq_has_overflowed(&iter->seq) ? -EOVERFLOW : 0;
 };
 
@@ -939,5 +949,7 @@ int hyp_trace_init_tracefs(void)
 				    (void *)cpu, &hyp_trace_fops);
 	}
 
+	hyp_trace_init_event_tracefs(root);
+
 	return 0;
 }
diff --git a/arch/arm64/kvm/hyp_trace.h b/arch/arm64/kvm/hyp_trace.h
index 14fc06c625a6..3ac648415bf9 100644
--- a/arch/arm64/kvm/hyp_trace.h
+++ b/arch/arm64/kvm/hyp_trace.h
@@ -3,26 +3,13 @@
 #ifndef __ARM64_KVM_HYP_TRACE_H__
 #define __ARM64_KVM_HYP_TRACE_H__
 
-#include <linux/trace_seq.h>
-#include <linux/workqueue.h>
-
-struct ht_iterator {
-	struct trace_buffer	*trace_buffer;
-	int			cpu;
-	struct hyp_entry_hdr	*ent;
-	unsigned long		lost_events;
-	int			ent_cpu;
-	size_t			ent_size;
-	u64			ts;
-	void			*spare;
-	size_t			copy_leftover;
-	struct trace_seq	seq;
-	struct delayed_work	poll_work;
-};
-
 #ifdef CONFIG_TRACING
 int hyp_trace_init_tracefs(void);
+int hyp_trace_init_events(void);
+struct hyp_event *hyp_trace_find_event(int id);
+void hyp_trace_init_event_tracefs(struct dentry *parent);
 #else
 static inline int hyp_trace_init_tracefs(void) { return 0; }
+static inline int hyp_trace_init_events(void) { return 0; }
 #endif
 #endif
-- 
2.48.1.601.g30ceb7b040-goog

[PATCH 11/11] KVM: arm64: Add kselftest for tracefs hyp tracefs

[Vincent Donnefort]

Add a test to validate the newly introduced tracefs interface for the
pKVM hypervisor. Covers buffer reset, loading/unloading, extended
timestamp and coherence of the tracing clock.

Signed-off-by: Vincent Donnefort <vdonnefort@google.com>

diff --git a/arch/arm64/include/asm/kvm_asm.h b/arch/arm64/include/asm/kvm_asm.h
index afeb983ca97b..c798bd6ad27e 100644
--- a/arch/arm64/include/asm/kvm_asm.h
+++ b/arch/arm64/include/asm/kvm_asm.h
@@ -94,6 +94,7 @@ enum __kvm_host_smccc_func {
 	__KVM_HOST_SMCCC_FUNC___pkvm_reset_tracing,
 	__KVM_HOST_SMCCC_FUNC___pkvm_swap_reader_tracing,
 	__KVM_HOST_SMCCC_FUNC___pkvm_enable_event,
+	__KVM_HOST_SMCCC_FUNC___pkvm_selftest_event,
 };
 
 #define DECLARE_KVM_VHE_SYM(sym)	extern char sym[]
diff --git a/arch/arm64/include/asm/kvm_hypevents.h b/arch/arm64/include/asm/kvm_hypevents.h
index 0b98a87a1250..25d8e23a3cc6 100644
--- a/arch/arm64/include/asm/kvm_hypevents.h
+++ b/arch/arm64/include/asm/kvm_hypevents.h
@@ -28,4 +28,14 @@ HYP_EVENT(hyp_exit,
 	),
 	HE_PRINTK(" ")
 );
+
+#ifdef CONFIG_PKVM_SELFTESTS
+HYP_EVENT(selftest,
+	  HE_PROTO(void),
+	  HE_STRUCT(),
+	  HE_ASSIGN(),
+	  HE_PRINTK(" ")
+);
 #endif
+
+#endif /* __ARM64_KVM_HYPEVENTS_H_ */
diff --git a/arch/arm64/kvm/Kconfig b/arch/arm64/kvm/Kconfig
index ead632ad01b4..fff9d24d7771 100644
--- a/arch/arm64/kvm/Kconfig
+++ b/arch/arm64/kvm/Kconfig
@@ -46,6 +46,7 @@ menuconfig KVM
 config NVHE_EL2_DEBUG
 	bool "Debug mode for non-VHE EL2 object"
 	depends on KVM
+	select PKVM_SELFTESTS
 	help
 	  Say Y here to enable the debug mode for the non-VHE KVM EL2 object.
 	  Failure reports will BUG() in the hypervisor. This is intended for
@@ -83,4 +84,13 @@ config PTDUMP_STAGE2_DEBUGFS
 
 	  If in doubt, say N.
 
+config PKVM_SELFTESTS
+	bool "Protected KVM hypervisor selftests"
+	depends on KVM
+	default n
+	help
+	  Say Y here to enable pKVM hypervisor testing infrastructure.
+
+	  If unsure, say N.
+
 endif # VIRTUALIZATION
diff --git a/arch/arm64/kvm/hyp/nvhe/hyp-main.c b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
index 96dde58f4984..8f0dafaab568 100644
--- a/arch/arm64/kvm/hyp/nvhe/hyp-main.c
+++ b/arch/arm64/kvm/hyp/nvhe/hyp-main.c
@@ -628,6 +628,19 @@ static void handle___pkvm_enable_event(struct kvm_cpu_context *host_ctxt)
 	cpu_reg(host_ctxt, 1) = __pkvm_enable_event(id, enable);
 }
 
+static void handle___pkvm_selftest_event(struct kvm_cpu_context *host_ctxt)
+{
+	int smc_ret = SMCCC_RET_NOT_SUPPORTED, ret = -EOPNOTSUPP;
+
+#ifdef CONFIG_PKVM_SELFTESTS
+	trace_selftest();
+	smc_ret = SMCCC_RET_SUCCESS;
+	ret = 0;
+#endif
+	cpu_reg(host_ctxt, 0) = smc_ret;
+	cpu_reg(host_ctxt, 1) = ret;
+}
+
 typedef void (*hcall_t)(struct kvm_cpu_context *);
 
 #define HANDLE_FUNC(x)	[__KVM_HOST_SMCCC_FUNC_##x] = (hcall_t)handle_##x
@@ -674,6 +687,7 @@ static const hcall_t host_hcall[] = {
 	HANDLE_FUNC(__pkvm_reset_tracing),
 	HANDLE_FUNC(__pkvm_swap_reader_tracing),
 	HANDLE_FUNC(__pkvm_enable_event),
+	HANDLE_FUNC(__pkvm_selftest_event),
 };
 
 static void handle_host_hcall(struct kvm_cpu_context *host_ctxt)
diff --git a/arch/arm64/kvm/hyp_trace.c b/arch/arm64/kvm/hyp_trace.c
index cb63af69c38d..1e887e430c42 100644
--- a/arch/arm64/kvm/hyp_trace.c
+++ b/arch/arm64/kvm/hyp_trace.c
@@ -892,6 +892,35 @@ static int hyp_trace_clock_show(struct seq_file *m, void *v)
 }
 DEFINE_SHOW_ATTRIBUTE(hyp_trace_clock);
 
+#ifdef CONFIG_PKVM_SELFTESTS
+static int selftest_event_open(struct inode *inode, struct file *file)
+{
+	if (file->f_mode & FMODE_WRITE)
+		return kvm_call_hyp_nvhe(__pkvm_selftest_event);
+
+	return 0;
+}
+
+static ssize_t selftest_event_write(struct file *f, const char __user *buf,
+				    size_t cnt, loff_t *pos)
+{
+	return cnt;
+}
+
+static const struct file_operations selftest_event_fops = {
+	.open	= selftest_event_open,
+	.write	= selftest_event_write,
+};
+
+static void hyp_trace_init_testing_tracefs(struct dentry *root)
+{
+	tracefs_create_file("selftest_event", TRACEFS_MODE_WRITE, root, NULL,
+			    &selftest_event_fops);
+}
+#else
+static void hyp_trace_init_testing_tracefs(struct dentry *root) { }
+#endif
+
 int hyp_trace_init_tracefs(void)
 {
 	struct dentry *root, *per_cpu_root;
@@ -950,6 +979,7 @@ int hyp_trace_init_tracefs(void)
 	}
 
 	hyp_trace_init_event_tracefs(root);
+	hyp_trace_init_testing_tracefs(root);
 
 	return 0;
 }
diff --git a/tools/testing/selftests/hyp-trace/Makefile b/tools/testing/selftests/hyp-trace/Makefile
new file mode 100644
index 000000000000..2a5b2e29667e
--- /dev/null
+++ b/tools/testing/selftests/hyp-trace/Makefile
@@ -0,0 +1,6 @@
+# SPDX-License-Identifier: GPL-2.0
+all:
+
+TEST_PROGS := hyp-trace-test
+
+include ../lib.mk
diff --git a/tools/testing/selftests/hyp-trace/config b/tools/testing/selftests/hyp-trace/config
new file mode 100644
index 000000000000..135657ef550d
--- /dev/null
+++ b/tools/testing/selftests/hyp-trace/config
@@ -0,0 +1,4 @@
+CONFIG_FTRACE=y
+CONFIG_ARM64=y
+CONFIG_KVM=y
+CONFIG_PKVM_SELFTESTS=y
diff --git a/tools/testing/selftests/hyp-trace/hyp-trace-test b/tools/testing/selftests/hyp-trace/hyp-trace-test
new file mode 100755
index 000000000000..9be0c5f57160
--- /dev/null
+++ b/tools/testing/selftests/hyp-trace/hyp-trace-test
@@ -0,0 +1,264 @@
+#!/bin/sh -e
+# SPDX-License-Identifier: GPL-2.0-only
+
+# hyp-trace-test - Tracefs for pKVM hypervisor test
+#
+# Copyright (C) 2024 - Google LLC
+# Author: Vincent Donnefort <vdonnefort@google.com>
+#
+
+log_and_die()
+{
+    echo "$1"
+
+    exit 1
+}
+
+host_clock()
+{
+    # BOOTTIME clock
+    awk '/now/ { printf "%.6f\n", $3 / 1000000000 }' /proc/timer_list
+}
+
+page_size()
+{
+    echo "$(awk '/KernelPageSize/ {print $2; exit}' /proc/self/smaps) * 1024" | bc
+}
+
+goto_hyp_trace()
+{
+    if [ -d "/sys/kernel/debug/tracing/hypervisor" ]; then
+        cd /sys/kernel/debug/tracing/hypervisor
+        return
+    fi
+
+    if [ -d "/sys/kernel/tracing/hypervisor" ]; then
+        cd /sys/kernel/tracing/hypervisor
+        return
+    fi
+
+    echo "ERROR: hyp tracing folder not found!"
+
+    exit 1
+}
+
+reset_hyp_trace()
+{
+    echo 0 > tracing_on
+    echo 0 > trace
+    for event in events/hypervisor/*; do
+        echo 0 > $event/enable
+    done
+
+    assert_unloaded
+}
+
+setup_hyp_trace()
+{
+    reset_hyp_trace
+
+    echo 16 > buffer_size_kb
+    echo 1 > events/hypervisor/selftest/enable
+    echo 1 > tracing_on
+}
+
+stop_hyp_trace()
+{
+    echo 0 > tracing_on
+}
+
+hyp_trace_loaded()
+{
+    grep -q "(loaded)" buffer_size_kb
+}
+
+write_events()
+{
+    local num="$1"
+    local func="$2"
+
+    for i in $(seq 1 $num); do
+        echo 1 > selftest_event
+        [ -z "$func" -o $i -eq $num ] || eval $func
+    done
+}
+
+consuming_read()
+{
+    local output=$1
+
+    cat trace_pipe > $output &
+
+    echo $!
+}
+
+run_test_consuming()
+{
+    local nr_events=$1
+    local func=$2
+    local tmp="$(mktemp)"
+    local start_ts=0
+    local end_ts=0
+    local pid=0
+
+    echo "Output trace file: $tmp"
+
+    setup_hyp_trace
+    pid=$(consuming_read $tmp)
+
+    start_ts=$(host_clock)
+    write_events $nr_events $func
+    stop_hyp_trace
+    end_ts=$(host_clock)
+
+    kill $pid
+    validate_test $tmp $nr_events $start_ts $end_ts
+
+    rm $tmp
+}
+
+validate_test()
+{
+    local output=$1
+    local expected_events=$2
+    local start_ts=$3
+    local end_ts=$4
+    local prev_ts=$3
+    local ts=0
+    local num_events=0
+
+    IFS=$'\n'
+    for line in $(cat $output); do
+        echo "$line" | grep -q -E "^# " && continue
+        ts=$(echo "$line" | awk '{print $2}' | cut -d ':' -f1)
+        if [ $(echo "$ts<$prev_ts" | bc) -eq 1 ]; then
+            log_and_die "Error event @$ts < $prev_ts"
+        fi
+        prev_ts=$ts
+        num_events=$((num_events + 1))
+    done
+
+    if [ $(echo "$ts>$end_ts" | bc) -eq 1 ]; then
+        log_and_die "Error event @$ts > $end_ts"
+    fi
+
+    if [ $num_events -ne $expected_events ]; then
+        log_and_die "Expected $expected_events events, got $num_events"
+    fi
+}
+
+test_ts()
+{
+    echo "Test Timestamps..."
+
+    run_test_consuming 1000
+
+    echo "done."
+}
+
+test_extended_ts()
+{
+    echo "Test Extended Timestamps..."
+
+    run_test_consuming 1000 "sleep 0.1"
+
+    echo "done."
+}
+
+assert_loaded()
+{
+    hyp_trace_loaded || log_and_die "Expected loaded buffer"
+}
+
+assert_unloaded()
+{
+    ! hyp_trace_loaded || log_and_die "Expected unloaded buffer"
+}
+
+test_unloading()
+{
+    local tmp="$(mktemp)"
+
+    echo "Test unloading..."
+
+    setup_hyp_trace
+    assert_loaded
+
+    echo 0 > tracing_on
+    assert_unloaded
+
+    pid=$(consuming_read $tmp)
+    sleep 1
+    assert_loaded
+    kill $pid
+    assert_unloaded
+
+    echo 1 > tracing_on
+    write_events 1
+    echo 0 > trace
+    assert_loaded
+    echo 0 > tracing_on
+    assert_unloaded
+
+    echo "done."
+}
+
+test_reset()
+{
+    local tmp="$(mktemp)"
+
+    echo "Test Reset..."
+
+    setup_hyp_trace
+    write_events 1000
+    echo 0 > trace
+    clock_before=$(host_clock)
+    write_events 5
+
+    pid=$(consuming_read $tmp)
+    sleep 1
+    stop_hyp_trace
+    kill $pid
+
+    validate_test $tmp 5 $clock_before $(host_clock)
+
+    rm $tmp
+
+    echo "done."
+}
+
+test_big_bpacking()
+{
+    local hyp_buffer_page_size=40
+    local page_size=$(page_size)
+    local min_buf_size
+
+    # Number of ring-buffer pages stored in a single bpacking page
+    min_buf_size=$(echo "$page_size / ($hyp_buffer_page_size * $(nproc))" | bc)
+    # Size of the ring-buffer to fill a single bpacking page
+    min_buf_size=$(echo "$page_size * $min_buf_size" | bc)
+    # Size in kiB of the ring-buffer to fill an order-1 bpacking page
+    min_buf_size=$(echo "$min_buf_size * 2 / 1024" | bc)
+
+    echo "Test loading $min_buf_size kB buffer..."
+
+    reset_hyp_trace
+
+    echo $min_buf_size > buffer_size_kb
+    echo 1 > tracing_on
+    assert_loaded
+
+    stop_hyp_trace
+
+    echo "done."
+}
+
+goto_hyp_trace
+
+test_reset
+test_unloading
+test_big_bpacking
+test_ts
+test_extended_ts
+
+exit 0
-- 
2.48.1.601.g30ceb7b040-goog

Re: [PATCH 02/11] ring-buffer: Expose buffer_data_page material

[Steven Rostedt]

On Mon, 24 Feb 2025 12:13:44 +0000
Vincent Donnefort <vdonnefort@google.com> wrote:

> @@ -61,11 +65,50 @@ enum ring_buffer_type {
>  	RINGBUF_TYPE_TIME_STAMP,
>  };
>  
> +#define TS_SHIFT        27
> +#define TS_MASK         ((1ULL << TS_SHIFT) - 1)
> +#define TS_DELTA_TEST   (~TS_MASK)

By moving this to include/linux you need to add a prefix to prevent name
space conflicts.

 RB_TS_SHIFT
 RB_TS_MASK
 RB_TS_DELTA_TEST

-- Steve


> +
> +/*
> + * We need to fit the time_stamp delta into 27 bits.
> + */
> +static inline bool test_time_stamp(u64 delta)
> +{
> +	return !!(delta & TS_DELTA_TEST);
> +}
> +