[PATCH v16 00/10] unwind_user: Deferred unwinding infrastructure

[PATCH v16 00/10] unwind_user: Deferred unwinding infrastructure

[Steven Rostedt]

Jen Remus suggested some updates from v15:

  https://lore.kernel.org/linux-trace-kernel/20250725185512.673587297@kernel.org/

Those were:

- Make fp_frame into a constant

- Removed useless initializing ra variable to zero
    
  git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace.git
unwind/core

Head SHA1: cf079f0176cc16e937f0fd868f0ec2d649ad53dd


Josh Poimboeuf (3):
      unwind_user: Add user space unwinding API with frame pointer support
      unwind_user/deferred: Add unwind cache
      unwind_user/deferred: Add deferred unwinding interface

Steven Rostedt (7):
      unwind_user/deferred: Add unwind_user_faultable()
      unwind_user/deferred: Make unwind deferral requests NMI-safe
      unwind deferred: Use bitmask to determine which callbacks to call
      unwind deferred: Add unwind_completed mask to stop spurious callbacks
      unwind: Add USED bit to only have one conditional on way back to user space
      unwind deferred: Use SRCU unwind_deferred_task_work()
      unwind: Finish up unwind when a task exits

----
 MAINTAINERS                           |   8 +
 arch/Kconfig                          |   7 +
 include/asm-generic/Kbuild            |   1 +
 include/asm-generic/unwind_user.h     |   5 +
 include/linux/entry-common.h          |   2 +
 include/linux/sched.h                 |   5 +
 include/linux/unwind_deferred.h       |  81 ++++++++
 include/linux/unwind_deferred_types.h |  39 ++++
 include/linux/unwind_user.h           |  14 ++
 include/linux/unwind_user_types.h     |  44 +++++
 kernel/Makefile                       |   1 +
 kernel/exit.c                         |   2 +
 kernel/fork.c                         |   4 +
 kernel/unwind/Makefile                |   1 +
 kernel/unwind/deferred.c              | 362 ++++++++++++++++++++++++++++++++++
 kernel/unwind/user.c                  | 128 ++++++++++++
 16 files changed, 704 insertions(+)
 create mode 100644 include/asm-generic/unwind_user.h
 create mode 100644 include/linux/unwind_deferred.h
 create mode 100644 include/linux/unwind_deferred_types.h
 create mode 100644 include/linux/unwind_user.h
 create mode 100644 include/linux/unwind_user_types.h
 create mode 100644 kernel/unwind/Makefile
 create mode 100644 kernel/unwind/deferred.c
 create mode 100644 kernel/unwind/user.c

[PATCH v16 01/10] unwind_user: Add user space unwinding API with frame pointer support

[Steven Rostedt]

From: Josh Poimboeuf <jpoimboe@kernel.org>

Introduce a generic API for unwinding user stacks.

In order to expand user space unwinding to be able to handle more complex
scenarios, such as deferred unwinding and reading user space information,
create a generic interface that all architectures can use that support the
various unwinding methods.

This is an alternative method for handling user space stack traces from
the simple stack_trace_save_user() API. This does not replace that
interface, but this interface will be used to expand the functionality of
user space stack walking.

None of the structures introduced will be exposed to user space tooling.

Support for frame pointer unwinding is added. For an architecture to
support frame pointer unwinding it needs to enable
CONFIG_HAVE_UNWIND_USER_FP and define ARCH_INIT_USER_FP_FRAME.

By encoding the frame offsets in struct unwind_user_frame, much of this
code can also be reused for future unwinder implementations like sframe.

Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org>
Co-developed-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Link: https://lore.kernel.org/all/20250710164301.3094-2-mathieu.desnoyers@efficios.com/
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Co-developed-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
---
Changes since v15: https://lore.kernel.org/20250725185739.233988371@kernel.org

- Made fp_frame into a constant (Jens Remus)

- Removed useless initializing ra variable to zero (Jens Remus)

 MAINTAINERS                       |   8 ++
 arch/Kconfig                      |   7 ++
 include/asm-generic/Kbuild        |   1 +
 include/asm-generic/unwind_user.h |   5 ++
 include/linux/unwind_user.h       |  14 ++++
 include/linux/unwind_user_types.h |  44 ++++++++++
 kernel/Makefile                   |   1 +
 kernel/unwind/Makefile            |   1 +
 kernel/unwind/user.c              | 128 ++++++++++++++++++++++++++++++
 9 files changed, 209 insertions(+)
 create mode 100644 include/asm-generic/unwind_user.h
 create mode 100644 include/linux/unwind_user.h
 create mode 100644 include/linux/unwind_user_types.h
 create mode 100644 kernel/unwind/Makefile
 create mode 100644 kernel/unwind/user.c

diff --git a/MAINTAINERS b/MAINTAINERS
index fad6cb025a19..370d780fd5f8 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -25928,6 +25928,14 @@ F:	Documentation/driver-api/uio-howto.rst
 F:	drivers/uio/
 F:	include/linux/uio_driver.h
 
+USERSPACE STACK UNWINDING
+M:	Josh Poimboeuf <jpoimboe@kernel.org>
+M:	Steven Rostedt <rostedt@goodmis.org>
+S:	Maintained
+F:	include/linux/unwind*.h
+F:	kernel/unwind/
+
+
 UTIL-LINUX PACKAGE
 M:	Karel Zak <kzak@redhat.com>
 L:	util-linux@vger.kernel.org
diff --git a/arch/Kconfig b/arch/Kconfig
index a3308a220f86..8e3fd723bd74 100644
--- a/arch/Kconfig
+++ b/arch/Kconfig
@@ -435,6 +435,13 @@ config HAVE_HARDLOCKUP_DETECTOR_ARCH
 	  It uses the same command line parameters, and sysctl interface,
 	  as the generic hardlockup detectors.
 
+config UNWIND_USER
+	bool
+
+config HAVE_UNWIND_USER_FP
+	bool
+	select UNWIND_USER
+
 config HAVE_PERF_REGS
 	bool
 	help
diff --git a/include/asm-generic/Kbuild b/include/asm-generic/Kbuild
index 8675b7b4ad23..295c94a3ccc1 100644
--- a/include/asm-generic/Kbuild
+++ b/include/asm-generic/Kbuild
@@ -59,6 +59,7 @@ mandatory-y += tlbflush.h
 mandatory-y += topology.h
 mandatory-y += trace_clock.h
 mandatory-y += uaccess.h
+mandatory-y += unwind_user.h
 mandatory-y += vermagic.h
 mandatory-y += vga.h
 mandatory-y += video.h
diff --git a/include/asm-generic/unwind_user.h b/include/asm-generic/unwind_user.h
new file mode 100644
index 000000000000..b8882b909944
--- /dev/null
+++ b/include/asm-generic/unwind_user.h
@@ -0,0 +1,5 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _ASM_GENERIC_UNWIND_USER_H
+#define _ASM_GENERIC_UNWIND_USER_H
+
+#endif /* _ASM_GENERIC_UNWIND_USER_H */
diff --git a/include/linux/unwind_user.h b/include/linux/unwind_user.h
new file mode 100644
index 000000000000..7f7282516bf5
--- /dev/null
+++ b/include/linux/unwind_user.h
@@ -0,0 +1,14 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_UNWIND_USER_H
+#define _LINUX_UNWIND_USER_H
+
+#include <linux/unwind_user_types.h>
+#include <asm/unwind_user.h>
+
+#ifndef ARCH_INIT_USER_FP_FRAME
+ #define ARCH_INIT_USER_FP_FRAME
+#endif
+
+int unwind_user(struct unwind_stacktrace *trace, unsigned int max_entries);
+
+#endif /* _LINUX_UNWIND_USER_H */
diff --git a/include/linux/unwind_user_types.h b/include/linux/unwind_user_types.h
new file mode 100644
index 000000000000..a449f15be890
--- /dev/null
+++ b/include/linux/unwind_user_types.h
@@ -0,0 +1,44 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_UNWIND_USER_TYPES_H
+#define _LINUX_UNWIND_USER_TYPES_H
+
+#include <linux/types.h>
+
+/*
+ * Unwind types, listed in priority order: lower numbers are attempted first if
+ * available.
+ */
+enum unwind_user_type_bits {
+	UNWIND_USER_TYPE_FP_BIT =		0,
+
+	NR_UNWIND_USER_TYPE_BITS,
+};
+
+enum unwind_user_type {
+	/* Type "none" for the start of stack walk iteration. */
+	UNWIND_USER_TYPE_NONE =			0,
+	UNWIND_USER_TYPE_FP =			BIT(UNWIND_USER_TYPE_FP_BIT),
+};
+
+struct unwind_stacktrace {
+	unsigned int	nr;
+	unsigned long	*entries;
+};
+
+struct unwind_user_frame {
+	s32 cfa_off;
+	s32 ra_off;
+	s32 fp_off;
+	bool use_fp;
+};
+
+struct unwind_user_state {
+	unsigned long				ip;
+	unsigned long				sp;
+	unsigned long				fp;
+	enum unwind_user_type			current_type;
+	unsigned int				available_types;
+	bool					done;
+};
+
+#endif /* _LINUX_UNWIND_USER_TYPES_H */
diff --git a/kernel/Makefile b/kernel/Makefile
index 32e80dd626af..541186050251 100644
--- a/kernel/Makefile
+++ b/kernel/Makefile
@@ -55,6 +55,7 @@ obj-y += rcu/
 obj-y += livepatch/
 obj-y += dma/
 obj-y += entry/
+obj-y += unwind/
 obj-$(CONFIG_MODULES) += module/
 
 obj-$(CONFIG_KCMP) += kcmp.o
diff --git a/kernel/unwind/Makefile b/kernel/unwind/Makefile
new file mode 100644
index 000000000000..349ce3677526
--- /dev/null
+++ b/kernel/unwind/Makefile
@@ -0,0 +1 @@
+ obj-$(CONFIG_UNWIND_USER) += user.o
diff --git a/kernel/unwind/user.c b/kernel/unwind/user.c
new file mode 100644
index 000000000000..97a8415e3216
--- /dev/null
+++ b/kernel/unwind/user.c
@@ -0,0 +1,128 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+* Generic interfaces for unwinding user space
+*/
+#include <linux/kernel.h>
+#include <linux/sched.h>
+#include <linux/sched/task_stack.h>
+#include <linux/unwind_user.h>
+#include <linux/uaccess.h>
+
+static const struct unwind_user_frame fp_frame = {
+	ARCH_INIT_USER_FP_FRAME
+};
+
+#define for_each_user_frame(state) \
+	for (unwind_user_start(state); !(state)->done; unwind_user_next(state))
+
+static int unwind_user_next_fp(struct unwind_user_state *state)
+{
+	const struct unwind_user_frame *frame = &fp_frame;
+	unsigned long cfa, fp, ra;
+	unsigned int shift;
+
+	if (frame->use_fp) {
+		if (state->fp < state->sp)
+			return -EINVAL;
+		cfa = state->fp;
+	} else {
+		cfa = state->sp;
+	}
+
+	/* Get the Canonical Frame Address (CFA) */
+	cfa += frame->cfa_off;
+
+	/* stack going in wrong direction? */
+	if (cfa <= state->sp)
+		return -EINVAL;
+
+	/* Make sure that the address is word aligned */
+	shift = sizeof(long) == 4 ? 2 : 3;
+	if (cfa & ((1 << shift) - 1))
+		return -EINVAL;
+
+	/* Find the Return Address (RA) */
+	if (get_user(ra, (unsigned long *)(cfa + frame->ra_off)))
+		return -EINVAL;
+
+	if (frame->fp_off && get_user(fp, (unsigned long __user *)(cfa + frame->fp_off)))
+		return -EINVAL;
+
+	state->ip = ra;
+	state->sp = cfa;
+	if (frame->fp_off)
+		state->fp = fp;
+	return 0;
+}
+
+static int unwind_user_next(struct unwind_user_state *state)
+{
+	unsigned long iter_mask = state->available_types;
+	unsigned int bit;
+
+	if (state->done)
+		return -EINVAL;
+
+	for_each_set_bit(bit, &iter_mask, NR_UNWIND_USER_TYPE_BITS) {
+		enum unwind_user_type type = BIT(bit);
+
+		state->current_type = type;
+		switch (type) {
+		case UNWIND_USER_TYPE_FP:
+			if (!unwind_user_next_fp(state))
+				return 0;
+			continue;
+		default:
+			WARN_ONCE(1, "Undefined unwind bit %d", bit);
+			break;
+		}
+		break;
+	}
+
+	/* No successful unwind method. */
+	state->current_type = UNWIND_USER_TYPE_NONE;
+	state->done = true;
+	return -EINVAL;
+}
+
+static int unwind_user_start(struct unwind_user_state *state)
+{
+	struct pt_regs *regs = task_pt_regs(current);
+
+	memset(state, 0, sizeof(*state));
+
+	if ((current->flags & PF_KTHREAD) || !user_mode(regs)) {
+		state->done = true;
+		return -EINVAL;
+	}
+
+	if (IS_ENABLED(CONFIG_HAVE_UNWIND_USER_FP))
+		state->available_types |= UNWIND_USER_TYPE_FP;
+
+	state->ip = instruction_pointer(regs);
+	state->sp = user_stack_pointer(regs);
+	state->fp = frame_pointer(regs);
+
+	return 0;
+}
+
+int unwind_user(struct unwind_stacktrace *trace, unsigned int max_entries)
+{
+	struct unwind_user_state state;
+
+	trace->nr = 0;
+
+	if (!max_entries)
+		return -EINVAL;
+
+	if (current->flags & PF_KTHREAD)
+		return 0;
+
+	for_each_user_frame(&state) {
+		trace->entries[trace->nr++] = state.ip;
+		if (trace->nr >= max_entries)
+			break;
+	}
+
+	return 0;
+}
-- 
2.47.2

[PATCH v16 02/10] unwind_user/deferred: Add unwind_user_faultable()

[Steven Rostedt]

From: Steven Rostedt <rostedt@goodmis.org>

Add a new API to retrieve a user space callstack called
unwind_user_faultable(). The difference between this user space stack
tracer from the current user space stack tracer is that this must be
called from faultable context as it may use routines to access user space
data that needs to be faulted in.

It can be safely called from entering or exiting a system call as the code
can still be faulted in there.

This code is based on work by Josh Poimboeuf's deferred unwinding code:

Link: https://lore.kernel.org/all/6052e8487746603bdb29b65f4033e739092d9925.1737511963.git.jpoimboe@kernel.org/

Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
---
 include/linux/sched.h                 |  5 +++
 include/linux/unwind_deferred.h       | 24 +++++++++++
 include/linux/unwind_deferred_types.h |  9 ++++
 kernel/fork.c                         |  4 ++
 kernel/unwind/Makefile                |  2 +-
 kernel/unwind/deferred.c              | 60 +++++++++++++++++++++++++++
 6 files changed, 103 insertions(+), 1 deletion(-)
 create mode 100644 include/linux/unwind_deferred.h
 create mode 100644 include/linux/unwind_deferred_types.h
 create mode 100644 kernel/unwind/deferred.c

diff --git a/include/linux/sched.h b/include/linux/sched.h
index 4f78a64beb52..59fdf7d9bb1e 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -46,6 +46,7 @@
 #include <linux/rv.h>
 #include <linux/uidgid_types.h>
 #include <linux/tracepoint-defs.h>
+#include <linux/unwind_deferred_types.h>
 #include <asm/kmap_size.h>
 
 /* task_struct member predeclarations (sorted alphabetically): */
@@ -1654,6 +1655,10 @@ struct task_struct {
 	struct user_event_mm		*user_event_mm;
 #endif
 
+#ifdef CONFIG_UNWIND_USER
+	struct unwind_task_info		unwind_info;
+#endif
+
 	/* CPU-specific state of this task: */
 	struct thread_struct		thread;
 
diff --git a/include/linux/unwind_deferred.h b/include/linux/unwind_deferred.h
new file mode 100644
index 000000000000..a5f6e8f8a1a2
--- /dev/null
+++ b/include/linux/unwind_deferred.h
@@ -0,0 +1,24 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_UNWIND_USER_DEFERRED_H
+#define _LINUX_UNWIND_USER_DEFERRED_H
+
+#include <linux/unwind_user.h>
+#include <linux/unwind_deferred_types.h>
+
+#ifdef CONFIG_UNWIND_USER
+
+void unwind_task_init(struct task_struct *task);
+void unwind_task_free(struct task_struct *task);
+
+int unwind_user_faultable(struct unwind_stacktrace *trace);
+
+#else /* !CONFIG_UNWIND_USER */
+
+static inline void unwind_task_init(struct task_struct *task) {}
+static inline void unwind_task_free(struct task_struct *task) {}
+
+static inline int unwind_user_faultable(struct unwind_stacktrace *trace) { return -ENOSYS; }
+
+#endif /* !CONFIG_UNWIND_USER */
+
+#endif /* _LINUX_UNWIND_USER_DEFERRED_H */
diff --git a/include/linux/unwind_deferred_types.h b/include/linux/unwind_deferred_types.h
new file mode 100644
index 000000000000..aa32db574e43
--- /dev/null
+++ b/include/linux/unwind_deferred_types.h
@@ -0,0 +1,9 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_UNWIND_USER_DEFERRED_TYPES_H
+#define _LINUX_UNWIND_USER_DEFERRED_TYPES_H
+
+struct unwind_task_info {
+	unsigned long		*entries;
+};
+
+#endif /* _LINUX_UNWIND_USER_DEFERRED_TYPES_H */
diff --git a/kernel/fork.c b/kernel/fork.c
index 1ee8eb11f38b..3341d50c61f2 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -105,6 +105,7 @@
 #include <uapi/linux/pidfd.h>
 #include <linux/pidfs.h>
 #include <linux/tick.h>
+#include <linux/unwind_deferred.h>
 
 #include <asm/pgalloc.h>
 #include <linux/uaccess.h>
@@ -732,6 +733,7 @@ void __put_task_struct(struct task_struct *tsk)
 	WARN_ON(refcount_read(&tsk->usage));
 	WARN_ON(tsk == current);
 
+	unwind_task_free(tsk);
 	sched_ext_free(tsk);
 	io_uring_free(tsk);
 	cgroup_free(tsk);
@@ -2135,6 +2137,8 @@ __latent_entropy struct task_struct *copy_process(
 	p->bpf_ctx = NULL;
 #endif
 
+	unwind_task_init(p);
+
 	/* Perform scheduler related setup. Assign this task to a CPU. */
 	retval = sched_fork(clone_flags, p);
 	if (retval)
diff --git a/kernel/unwind/Makefile b/kernel/unwind/Makefile
index 349ce3677526..eae37bea54fd 100644
--- a/kernel/unwind/Makefile
+++ b/kernel/unwind/Makefile
@@ -1 +1 @@
- obj-$(CONFIG_UNWIND_USER) += user.o
+ obj-$(CONFIG_UNWIND_USER)	+= user.o deferred.o
diff --git a/kernel/unwind/deferred.c b/kernel/unwind/deferred.c
new file mode 100644
index 000000000000..a0badbeb3cc1
--- /dev/null
+++ b/kernel/unwind/deferred.c
@@ -0,0 +1,60 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Deferred user space unwinding
+ */
+#include <linux/kernel.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/unwind_deferred.h>
+
+#define UNWIND_MAX_ENTRIES 512
+
+/**
+ * unwind_user_faultable - Produce a user stacktrace in faultable context
+ * @trace: The descriptor that will store the user stacktrace
+ *
+ * This must be called in a known faultable context (usually when entering
+ * or exiting user space). Depending on the available implementations
+ * the @trace will be loaded with the addresses of the user space stacktrace
+ * if it can be found.
+ *
+ * Return: 0 on success and negative on error
+ *         On success @trace will contain the user space stacktrace
+ */
+int unwind_user_faultable(struct unwind_stacktrace *trace)
+{
+	struct unwind_task_info *info = &current->unwind_info;
+
+	/* Should always be called from faultable context */
+	might_fault();
+
+	if (current->flags & PF_EXITING)
+		return -EINVAL;
+
+	if (!info->entries) {
+		info->entries = kmalloc_array(UNWIND_MAX_ENTRIES, sizeof(long),
+					      GFP_KERNEL);
+		if (!info->entries)
+			return -ENOMEM;
+	}
+
+	trace->nr = 0;
+	trace->entries = info->entries;
+	unwind_user(trace, UNWIND_MAX_ENTRIES);
+
+	return 0;
+}
+
+void unwind_task_init(struct task_struct *task)
+{
+	struct unwind_task_info *info = &task->unwind_info;
+
+	memset(info, 0, sizeof(*info));
+}
+
+void unwind_task_free(struct task_struct *task)
+{
+	struct unwind_task_info *info = &task->unwind_info;
+
+	kfree(info->entries);
+}
-- 
2.47.2

[PATCH v16 03/10] unwind_user/deferred: Add unwind cache

[Steven Rostedt]

From: Josh Poimboeuf <jpoimboe@kernel.org>

Cache the results of the unwind to ensure the unwind is only performed
once, even when called by multiple tracers.

The cache nr_entries gets cleared every time the task exits the kernel.
When a stacktrace is requested, nr_entries gets set to the number of
entries in the stacktrace. If another stacktrace is requested, if
nr_entries is not zero, then it contains the same stacktrace that would be
retrieved so it is not processed again and the entries is given to the
caller.

Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Co-developed-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
---
 include/linux/entry-common.h          |  2 ++
 include/linux/unwind_deferred.h       |  8 +++++++
 include/linux/unwind_deferred_types.h |  7 +++++-
 kernel/unwind/deferred.c              | 31 +++++++++++++++++++++------
 4 files changed, 40 insertions(+), 8 deletions(-)

diff --git a/include/linux/entry-common.h b/include/linux/entry-common.h
index f94f3fdf15fc..8908b8eeb99b 100644
--- a/include/linux/entry-common.h
+++ b/include/linux/entry-common.h
@@ -12,6 +12,7 @@
 #include <linux/resume_user_mode.h>
 #include <linux/tick.h>
 #include <linux/kmsan.h>
+#include <linux/unwind_deferred.h>
 
 #include <asm/entry-common.h>
 #include <asm/syscall.h>
@@ -362,6 +363,7 @@ static __always_inline void exit_to_user_mode(void)
 	lockdep_hardirqs_on_prepare();
 	instrumentation_end();
 
+	unwind_reset_info();
 	user_enter_irqoff();
 	arch_exit_to_user_mode();
 	lockdep_hardirqs_on(CALLER_ADDR0);
diff --git a/include/linux/unwind_deferred.h b/include/linux/unwind_deferred.h
index a5f6e8f8a1a2..baacf4a1eb4c 100644
--- a/include/linux/unwind_deferred.h
+++ b/include/linux/unwind_deferred.h
@@ -12,6 +12,12 @@ void unwind_task_free(struct task_struct *task);
 
 int unwind_user_faultable(struct unwind_stacktrace *trace);
 
+static __always_inline void unwind_reset_info(void)
+{
+	if (unlikely(current->unwind_info.cache))
+		current->unwind_info.cache->nr_entries = 0;
+}
+
 #else /* !CONFIG_UNWIND_USER */
 
 static inline void unwind_task_init(struct task_struct *task) {}
@@ -19,6 +25,8 @@ static inline void unwind_task_free(struct task_struct *task) {}
 
 static inline int unwind_user_faultable(struct unwind_stacktrace *trace) { return -ENOSYS; }
 
+static inline void unwind_reset_info(void) {}
+
 #endif /* !CONFIG_UNWIND_USER */
 
 #endif /* _LINUX_UNWIND_USER_DEFERRED_H */
diff --git a/include/linux/unwind_deferred_types.h b/include/linux/unwind_deferred_types.h
index aa32db574e43..db5b54b18828 100644
--- a/include/linux/unwind_deferred_types.h
+++ b/include/linux/unwind_deferred_types.h
@@ -2,8 +2,13 @@
 #ifndef _LINUX_UNWIND_USER_DEFERRED_TYPES_H
 #define _LINUX_UNWIND_USER_DEFERRED_TYPES_H
 
+struct unwind_cache {
+	unsigned int		nr_entries;
+	unsigned long		entries[];
+};
+
 struct unwind_task_info {
-	unsigned long		*entries;
+	struct unwind_cache	*cache;
 };
 
 #endif /* _LINUX_UNWIND_USER_DEFERRED_TYPES_H */
diff --git a/kernel/unwind/deferred.c b/kernel/unwind/deferred.c
index a0badbeb3cc1..96368a5aa522 100644
--- a/kernel/unwind/deferred.c
+++ b/kernel/unwind/deferred.c
@@ -4,10 +4,13 @@
  */
 #include <linux/kernel.h>
 #include <linux/sched.h>
+#include <linux/sizes.h>
 #include <linux/slab.h>
 #include <linux/unwind_deferred.h>
 
-#define UNWIND_MAX_ENTRIES 512
+/* Make the cache fit in a 4K page */
+#define UNWIND_MAX_ENTRIES					\
+	((SZ_4K - sizeof(struct unwind_cache)) / sizeof(long))
 
 /**
  * unwind_user_faultable - Produce a user stacktrace in faultable context
@@ -24,6 +27,7 @@
 int unwind_user_faultable(struct unwind_stacktrace *trace)
 {
 	struct unwind_task_info *info = &current->unwind_info;
+	struct unwind_cache *cache;
 
 	/* Should always be called from faultable context */
 	might_fault();
@@ -31,17 +35,30 @@ int unwind_user_faultable(struct unwind_stacktrace *trace)
 	if (current->flags & PF_EXITING)
 		return -EINVAL;
 
-	if (!info->entries) {
-		info->entries = kmalloc_array(UNWIND_MAX_ENTRIES, sizeof(long),
-					      GFP_KERNEL);
-		if (!info->entries)
+	if (!info->cache) {
+		info->cache = kzalloc(struct_size(cache, entries, UNWIND_MAX_ENTRIES),
+				      GFP_KERNEL);
+		if (!info->cache)
 			return -ENOMEM;
 	}
 
+	cache = info->cache;
+	trace->entries = cache->entries;
+
+	if (cache->nr_entries) {
+		/*
+		 * The user stack has already been previously unwound in this
+		 * entry context.  Skip the unwind and use the cache.
+		 */
+		trace->nr = cache->nr_entries;
+		return 0;
+	}
+
 	trace->nr = 0;
-	trace->entries = info->entries;
 	unwind_user(trace, UNWIND_MAX_ENTRIES);
 
+	cache->nr_entries = trace->nr;
+
 	return 0;
 }
 
@@ -56,5 +73,5 @@ void unwind_task_free(struct task_struct *task)
 {
 	struct unwind_task_info *info = &task->unwind_info;
 
-	kfree(info->entries);
+	kfree(info->cache);
 }
-- 
2.47.2

Re: [PATCH v16 03/10] unwind_user/deferred: Add unwind cache

[Indu Bhagat]

On 7/29/25 11:23 AM, Steven Rostedt wrote:
> From: Josh Poimboeuf <jpoimboe@kernel.org>
> 
> Cache the results of the unwind to ensure the unwind is only performed
> once, even when called by multiple tracers.
> 
> The cache nr_entries gets cleared every time the task exits the kernel.
> When a stacktrace is requested, nr_entries gets set to the number of
> entries in the stacktrace. If another stacktrace is requested, if
> nr_entries is not zero, then it contains the same stacktrace that would be
> retrieved so it is not processed again and the entries is given to the
> caller.
> 
> Reviewed-by: Jens Remus <jremus@linux.ibm.com>
> Co-developed-by: Steven Rostedt (Google) <rostedt@goodmis.org>
> Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org>
> Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
> Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
> ---
>   include/linux/entry-common.h          |  2 ++
>   include/linux/unwind_deferred.h       |  8 +++++++
>   include/linux/unwind_deferred_types.h |  7 +++++-
>   kernel/unwind/deferred.c              | 31 +++++++++++++++++++++------
>   4 files changed, 40 insertions(+), 8 deletions(-)
> 
> diff --git a/include/linux/entry-common.h b/include/linux/entry-common.h
> index f94f3fdf15fc..8908b8eeb99b 100644
> --- a/include/linux/entry-common.h
> +++ b/include/linux/entry-common.h
> @@ -12,6 +12,7 @@
>   #include <linux/resume_user_mode.h>
>   #include <linux/tick.h>
>   #include <linux/kmsan.h>
> +#include <linux/unwind_deferred.h>
>   
>   #include <asm/entry-common.h>
>   #include <asm/syscall.h>
> @@ -362,6 +363,7 @@ static __always_inline void exit_to_user_mode(void)
>   	lockdep_hardirqs_on_prepare();
>   	instrumentation_end();
>   
> +	unwind_reset_info();
>   	user_enter_irqoff();
>   	arch_exit_to_user_mode();
>   	lockdep_hardirqs_on(CALLER_ADDR0);
> diff --git a/include/linux/unwind_deferred.h b/include/linux/unwind_deferred.h
> index a5f6e8f8a1a2..baacf4a1eb4c 100644
> --- a/include/linux/unwind_deferred.h
> +++ b/include/linux/unwind_deferred.h
> @@ -12,6 +12,12 @@ void unwind_task_free(struct task_struct *task);
>   
>   int unwind_user_faultable(struct unwind_stacktrace *trace);
>   
> +static __always_inline void unwind_reset_info(void)
> +{
> +	if (unlikely(current->unwind_info.cache))
> +		current->unwind_info.cache->nr_entries = 0;
> +}

Should the entries[] items upto nr_entries (stack trace info from the 
previous request) also be reset to 0 here ?

> +
>   #else /* !CONFIG_UNWIND_USER */
>   
>   static inline void unwind_task_init(struct task_struct *task) {}
> @@ -19,6 +25,8 @@ static inline void unwind_task_free(struct task_struct *task) {}
>   
>   static inline int unwind_user_faultable(struct unwind_stacktrace *trace) { return -ENOSYS; }
>   
> +static inline void unwind_reset_info(void) {}
> +
>   #endif /* !CONFIG_UNWIND_USER */
>   
>   #endif /* _LINUX_UNWIND_USER_DEFERRED_H */
> diff --git a/include/linux/unwind_deferred_types.h b/include/linux/unwind_deferred_types.h
> index aa32db574e43..db5b54b18828 100644
> --- a/include/linux/unwind_deferred_types.h
> +++ b/include/linux/unwind_deferred_types.h
> @@ -2,8 +2,13 @@
>   #ifndef _LINUX_UNWIND_USER_DEFERRED_TYPES_H
>   #define _LINUX_UNWIND_USER_DEFERRED_TYPES_H
>   
> +struct unwind_cache {
> +	unsigned int		nr_entries;
> +	unsigned long		entries[];
> +};
> +

Should we use __counted_by ?

>   struct unwind_task_info {
> -	unsigned long		*entries;
> +	struct unwind_cache	*cache;
>   };
>   
>   #endif /* _LINUX_UNWIND_USER_DEFERRED_TYPES_H */
> diff --git a/kernel/unwind/deferred.c b/kernel/unwind/deferred.c
> index a0badbeb3cc1..96368a5aa522 100644
> --- a/kernel/unwind/deferred.c
> +++ b/kernel/unwind/deferred.c
> @@ -4,10 +4,13 @@
>    */
>   #include <linux/kernel.h>
>   #include <linux/sched.h>
> +#include <linux/sizes.h>
>   #include <linux/slab.h>
>   #include <linux/unwind_deferred.h>
>   
> -#define UNWIND_MAX_ENTRIES 512
> +/* Make the cache fit in a 4K page */
> +#define UNWIND_MAX_ENTRIES					\
> +	((SZ_4K - sizeof(struct unwind_cache)) / sizeof(long))
>   
>   /**
>    * unwind_user_faultable - Produce a user stacktrace in faultable context
> @@ -24,6 +27,7 @@
>   int unwind_user_faultable(struct unwind_stacktrace *trace)
>   {
>   	struct unwind_task_info *info = &current->unwind_info;
> +	struct unwind_cache *cache;
>   
>   	/* Should always be called from faultable context */
>   	might_fault();
> @@ -31,17 +35,30 @@ int unwind_user_faultable(struct unwind_stacktrace *trace)
>   	if (current->flags & PF_EXITING)
>   		return -EINVAL;
>   
> -	if (!info->entries) {
> -		info->entries = kmalloc_array(UNWIND_MAX_ENTRIES, sizeof(long),
> -					      GFP_KERNEL);
> -		if (!info->entries)
> +	if (!info->cache) {
> +		info->cache = kzalloc(struct_size(cache, entries, UNWIND_MAX_ENTRIES),
> +				      GFP_KERNEL);
> +		if (!info->cache)
>   			return -ENOMEM;
>   	}
>   
> +	cache = info->cache;
> +	trace->entries = cache->entries;
> +
> +	if (cache->nr_entries) {
> +		/*
> +		 * The user stack has already been previously unwound in this
> +		 * entry context.  Skip the unwind and use the cache.
> +		 */
> +		trace->nr = cache->nr_entries;
> +		return 0;
> +	}
> +
>   	trace->nr = 0;
> -	trace->entries = info->entries;
>   	unwind_user(trace, UNWIND_MAX_ENTRIES);
>   
> +	cache->nr_entries = trace->nr;
> +
>   	return 0;
>   }
>   
> @@ -56,5 +73,5 @@ void unwind_task_free(struct task_struct *task)
>   {
>   	struct unwind_task_info *info = &task->unwind_info;
>   
> -	kfree(info->entries);
> +	kfree(info->cache);
>   }

Re: [PATCH v16 03/10] unwind_user/deferred: Add unwind cache

[Steven Rostedt]

On Tue, 29 Jul 2025 21:55:39 -0700
Indu Bhagat <indu.bhagat@oracle.com> wrote:

> > diff --git a/include/linux/unwind_deferred.h b/include/linux/unwind_deferred.h
> > index a5f6e8f8a1a2..baacf4a1eb4c 100644
> > --- a/include/linux/unwind_deferred.h
> > +++ b/include/linux/unwind_deferred.h
> > @@ -12,6 +12,12 @@ void unwind_task_free(struct task_struct *task);
> >   
> >   int unwind_user_faultable(struct unwind_stacktrace *trace);
> >   
> > +static __always_inline void unwind_reset_info(void)
> > +{
> > +	if (unlikely(current->unwind_info.cache))
> > +		current->unwind_info.cache->nr_entries = 0;
> > +}  
> 
> Should the entries[] items upto nr_entries (stack trace info from the 
> previous request) also be reset to 0 here ?

This is in a critical path, there's no reason to reset to zero. The data will
just be stale. Nothing should care about anything over nr_entries.

> > diff --git a/include/linux/unwind_deferred_types.h b/include/linux/unwind_deferred_types.h
> > index aa32db574e43..db5b54b18828 100644
> > --- a/include/linux/unwind_deferred_types.h
> > +++ b/include/linux/unwind_deferred_types.h
> > @@ -2,8 +2,13 @@
> >   #ifndef _LINUX_UNWIND_USER_DEFERRED_TYPES_H
> >   #define _LINUX_UNWIND_USER_DEFERRED_TYPES_H
> >   
> > +struct unwind_cache {
> > +	unsigned int		nr_entries;
> > +	unsigned long		entries[];
> > +};
> > +  
> 
> Should we use __counted_by ?

The size of entries[] is not determined by nr_entries. It is allocated on
the first use, and not freed until the task exits. It's a fixed size
defined by:

/* Make the cache fit in a 4K page */
#define UNWIND_MAX_ENTRIES					\
	((SZ_4K - sizeof(struct unwind_cache)) / sizeof(long))


-- Steve

	

Re: [PATCH v16 03/10] unwind_user/deferred: Add unwind cache

[Indu Bhagat]

On 7/30/25 6:32 AM, Steven Rostedt wrote:
> On Tue, 29 Jul 2025 21:55:39 -0700
> Indu Bhagat <indu.bhagat@oracle.com> wrote:
> 
>>> diff --git a/include/linux/unwind_deferred.h b/include/linux/unwind_deferred.h
>>> index a5f6e8f8a1a2..baacf4a1eb4c 100644
>>> --- a/include/linux/unwind_deferred.h
>>> +++ b/include/linux/unwind_deferred.h
>>> @@ -12,6 +12,12 @@ void unwind_task_free(struct task_struct *task);
>>>    
>>>    int unwind_user_faultable(struct unwind_stacktrace *trace);
>>>    
>>> +static __always_inline void unwind_reset_info(void)
>>> +{
>>> +	if (unlikely(current->unwind_info.cache))
>>> +		current->unwind_info.cache->nr_entries = 0;
>>> +}
>>
>> Should the entries[] items upto nr_entries (stack trace info from the
>> previous request) also be reset to 0 here ?
> 
> This is in a critical path, there's no reason to reset to zero. The data will
> just be stale. Nothing should care about anything over nr_entries.
> 

OK.

Reviewed-By: Indu Bhagat <indu.bhagat@oracle.com>

[PATCH v16 04/10] unwind_user/deferred: Add deferred unwinding interface

[Steven Rostedt]

From: Josh Poimboeuf <jpoimboe@kernel.org>

Add an interface for scheduling task work to unwind the user space stack
before returning to user space. This solves several problems for its
callers:

  - Ensure the unwind happens in task context even if the caller may be
    running in interrupt context.

  - Avoid duplicate unwinds, whether called multiple times by the same
    caller or by different callers.

  - Create a "context cookie" which allows trace post-processing to
    correlate kernel unwinds/traces with the user unwind.

A concept of a "cookie" is created to detect when the stacktrace is the
same. A cookie is generated the first time a user space stacktrace is
requested after the task enters the kernel. As the stacktrace is saved on
the task_struct while the task is in the kernel, if another request comes
in, if the cookie is still the same, it will use the saved stacktrace,
and not have to regenerate one.

The cookie is passed to the caller on request, and when the stacktrace is
generated upon returning to user space, it calls the requester's callback
with the cookie as well as the stacktrace. The cookie is cleared
when it goes back to user space. Note, this currently adds another
conditional to the unwind_reset_info() path that is always called
returning to user space, but future changes will put this back to a single
conditional.

A global list is created and protected by a global mutex that holds
tracers that register with the unwind infrastructure. The number of
registered tracers will be limited in future changes. Each perf program or
ftrace instance will register its own descriptor to use for deferred
unwind stack traces.

Note, in the function unwind_deferred_task_work() that gets called when
returning to user space, it uses a global mutex for synchronization which
will cause a big bottleneck. This will be replaced by SRCU, but that
change adds some complex synchronization that deservers its own commit.

Co-developed-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
---
 include/linux/unwind_deferred.h       |  24 ++++
 include/linux/unwind_deferred_types.h |  24 ++++
 kernel/unwind/deferred.c              | 156 +++++++++++++++++++++++++-
 3 files changed, 203 insertions(+), 1 deletion(-)

diff --git a/include/linux/unwind_deferred.h b/include/linux/unwind_deferred.h
index baacf4a1eb4c..14efd8c027aa 100644
--- a/include/linux/unwind_deferred.h
+++ b/include/linux/unwind_deferred.h
@@ -2,9 +2,19 @@
 #ifndef _LINUX_UNWIND_USER_DEFERRED_H
 #define _LINUX_UNWIND_USER_DEFERRED_H
 
+#include <linux/task_work.h>
 #include <linux/unwind_user.h>
 #include <linux/unwind_deferred_types.h>
 
+struct unwind_work;
+
+typedef void (*unwind_callback_t)(struct unwind_work *work, struct unwind_stacktrace *trace, u64 cookie);
+
+struct unwind_work {
+	struct list_head		list;
+	unwind_callback_t		func;
+};
+
 #ifdef CONFIG_UNWIND_USER
 
 void unwind_task_init(struct task_struct *task);
@@ -12,8 +22,19 @@ void unwind_task_free(struct task_struct *task);
 
 int unwind_user_faultable(struct unwind_stacktrace *trace);
 
+int unwind_deferred_init(struct unwind_work *work, unwind_callback_t func);
+int unwind_deferred_request(struct unwind_work *work, u64 *cookie);
+void unwind_deferred_cancel(struct unwind_work *work);
+
 static __always_inline void unwind_reset_info(void)
 {
+	if (unlikely(current->unwind_info.id.id))
+		current->unwind_info.id.id = 0;
+	/*
+	 * As unwind_user_faultable() can be called directly and
+	 * depends on nr_entries being cleared on exit to user,
+	 * this needs to be a separate conditional.
+	 */
 	if (unlikely(current->unwind_info.cache))
 		current->unwind_info.cache->nr_entries = 0;
 }
@@ -24,6 +45,9 @@ static inline void unwind_task_init(struct task_struct *task) {}
 static inline void unwind_task_free(struct task_struct *task) {}
 
 static inline int unwind_user_faultable(struct unwind_stacktrace *trace) { return -ENOSYS; }
+static inline int unwind_deferred_init(struct unwind_work *work, unwind_callback_t func) { return -ENOSYS; }
+static inline int unwind_deferred_request(struct unwind_work *work, u64 *timestamp) { return -ENOSYS; }
+static inline void unwind_deferred_cancel(struct unwind_work *work) {}
 
 static inline void unwind_reset_info(void) {}
 
diff --git a/include/linux/unwind_deferred_types.h b/include/linux/unwind_deferred_types.h
index db5b54b18828..104c477d5609 100644
--- a/include/linux/unwind_deferred_types.h
+++ b/include/linux/unwind_deferred_types.h
@@ -7,8 +7,32 @@ struct unwind_cache {
 	unsigned long		entries[];
 };
 
+/*
+ * The unwind_task_id is a unique identifier that maps to a user space
+ * stacktrace. It is generated the first time a deferred user space
+ * stacktrace is requested after a task has entered the kerenl and
+ * is cleared to zero when it exits. The mapped id will be a non-zero
+ * number.
+ *
+ * To simplify the generation of the 64 bit number, 32 bits will be
+ * the CPU it was generated on, and the other 32 bits will be a per
+ * cpu counter that gets incremented by two every time a new identifier
+ * is generated. The LSB will always be set to keep the value
+ * from being zero.
+ */
+union unwind_task_id {
+	struct {
+		u32		cpu;
+		u32		cnt;
+	};
+	u64			id;
+};
+
 struct unwind_task_info {
 	struct unwind_cache	*cache;
+	struct callback_head	work;
+	union unwind_task_id	id;
+	int			pending;
 };
 
 #endif /* _LINUX_UNWIND_USER_DEFERRED_TYPES_H */
diff --git a/kernel/unwind/deferred.c b/kernel/unwind/deferred.c
index 96368a5aa522..2cbae2ada309 100644
--- a/kernel/unwind/deferred.c
+++ b/kernel/unwind/deferred.c
@@ -2,16 +2,63 @@
 /*
  * Deferred user space unwinding
  */
+#include <linux/sched/task_stack.h>
+#include <linux/unwind_deferred.h>
+#include <linux/sched/clock.h>
+#include <linux/task_work.h>
 #include <linux/kernel.h>
 #include <linux/sched.h>
 #include <linux/sizes.h>
 #include <linux/slab.h>
-#include <linux/unwind_deferred.h>
+#include <linux/mm.h>
 
 /* Make the cache fit in a 4K page */
 #define UNWIND_MAX_ENTRIES					\
 	((SZ_4K - sizeof(struct unwind_cache)) / sizeof(long))
 
+/* Guards adding to and reading the list of callbacks */
+static DEFINE_MUTEX(callback_mutex);
+static LIST_HEAD(callbacks);
+
+/*
+ * This is a unique percpu identifier for a given task entry context.
+ * Conceptually, it's incremented every time the CPU enters the kernel from
+ * user space, so that each "entry context" on the CPU gets a unique ID.  In
+ * reality, as an optimization, it's only incremented on demand for the first
+ * deferred unwind request after a given entry-from-user.
+ *
+ * It's combined with the CPU id to make a systemwide-unique "context cookie".
+ */
+static DEFINE_PER_CPU(u32, unwind_ctx_ctr);
+
+/*
+ * The context cookie is a unique identifier that is assigned to a user
+ * space stacktrace. As the user space stacktrace remains the same while
+ * the task is in the kernel, the cookie is an identifier for the stacktrace.
+ * Although it is possible for the stacktrace to get another cookie if another
+ * request is made after the cookie was cleared and before reentering user
+ * space.
+ */
+static u64 get_cookie(struct unwind_task_info *info)
+{
+	u32 cnt = 1;
+	u32 old = 0;
+
+	if (info->id.cpu)
+		return info->id.id;
+
+	/* LSB is always set to ensure 0 is an invalid value */
+	cnt |= __this_cpu_read(unwind_ctx_ctr) + 2;
+	if (try_cmpxchg(&info->id.cnt, &old, cnt)) {
+		/* Update the per cpu counter */
+		__this_cpu_write(unwind_ctx_ctr, cnt);
+	}
+	/* Interrupts are disabled, the CPU will always be same */
+	info->id.cpu = smp_processor_id() + 1; /* Must be non zero */
+
+	return info->id.id;
+}
+
 /**
  * unwind_user_faultable - Produce a user stacktrace in faultable context
  * @trace: The descriptor that will store the user stacktrace
@@ -62,11 +109,117 @@ int unwind_user_faultable(struct unwind_stacktrace *trace)
 	return 0;
 }
 
+static void unwind_deferred_task_work(struct callback_head *head)
+{
+	struct unwind_task_info *info = container_of(head, struct unwind_task_info, work);
+	struct unwind_stacktrace trace;
+	struct unwind_work *work;
+	u64 cookie;
+
+	if (WARN_ON_ONCE(!info->pending))
+		return;
+
+	/* Allow work to come in again */
+	WRITE_ONCE(info->pending, 0);
+
+	/*
+	 * From here on out, the callback must always be called, even if it's
+	 * just an empty trace.
+	 */
+	trace.nr = 0;
+	trace.entries = NULL;
+
+	unwind_user_faultable(&trace);
+
+	cookie = info->id.id;
+
+	guard(mutex)(&callback_mutex);
+	list_for_each_entry(work, &callbacks, list) {
+		work->func(work, &trace, cookie);
+	}
+}
+
+/**
+ * unwind_deferred_request - Request a user stacktrace on task kernel exit
+ * @work: Unwind descriptor requesting the trace
+ * @cookie: The cookie of the first request made for this task
+ *
+ * Schedule a user space unwind to be done in task work before exiting the
+ * kernel.
+ *
+ * The returned @cookie output is the generated cookie of the very first
+ * request for a user space stacktrace for this task since it entered the
+ * kernel. It can be from a request by any caller of this infrastructure.
+ * Its value will also be passed to the callback function.  It can be
+ * used to stitch kernel and user stack traces together in post-processing.
+ *
+ * It's valid to call this function multiple times for the same @work within
+ * the same task entry context.  Each call will return the same cookie
+ * while the task hasn't left the kernel. If the callback is not pending
+ * because it has already been previously called for the same entry context,
+ * it will be called again with the same stack trace and cookie.
+ *
+ * Return: 1 if the the callback was already queued.
+ *         0 if the callback successfully was queued.
+ *         Negative if there's an error.
+ *         @cookie holds the cookie of the first request by any user
+ */
+int unwind_deferred_request(struct unwind_work *work, u64 *cookie)
+{
+	struct unwind_task_info *info = &current->unwind_info;
+	int ret;
+
+	*cookie = 0;
+
+	if (WARN_ON_ONCE(in_nmi()))
+		return -EINVAL;
+
+	if ((current->flags & (PF_KTHREAD | PF_EXITING)) ||
+	    !user_mode(task_pt_regs(current)))
+		return -EINVAL;
+
+	guard(irqsave)();
+
+	*cookie = get_cookie(info);
+
+	/* callback already pending? */
+	if (info->pending)
+		return 1;
+
+	/* The work has been claimed, now schedule it. */
+	ret = task_work_add(current, &info->work, TWA_RESUME);
+	if (WARN_ON_ONCE(ret))
+		return ret;
+
+	info->pending = 1;
+	return 0;
+}
+
+void unwind_deferred_cancel(struct unwind_work *work)
+{
+	if (!work)
+		return;
+
+	guard(mutex)(&callback_mutex);
+	list_del(&work->list);
+}
+
+int unwind_deferred_init(struct unwind_work *work, unwind_callback_t func)
+{
+	memset(work, 0, sizeof(*work));
+
+	guard(mutex)(&callback_mutex);
+	list_add(&work->list, &callbacks);
+	work->func = func;
+	return 0;
+}
+
 void unwind_task_init(struct task_struct *task)
 {
 	struct unwind_task_info *info = &task->unwind_info;
 
 	memset(info, 0, sizeof(*info));
+	init_task_work(&info->work, unwind_deferred_task_work);
 }
 
 void unwind_task_free(struct task_struct *task)
@@ -74,4 +227,5 @@ void unwind_task_free(struct task_struct *task)
 	struct unwind_task_info *info = &task->unwind_info;
 
 	kfree(info->cache);
+	task_work_cancel(task, &info->work);
 }
-- 
2.47.2

[PATCH v16 05/10] unwind_user/deferred: Make unwind deferral requests NMI-safe

[Steven Rostedt]

From: Steven Rostedt <rostedt@goodmis.org>

Make unwind_deferred_request() NMI-safe so tracers in NMI context can
call it and safely request a user space stacktrace when the task exits.

Note, this is only allowed for architectures that implement a safe
cmpxchg. If an architecture requests a deferred stack trace from NMI
context that does not support a safe NMI cmpxchg, it will get an -EINVAL
and trigger a warning. For those architectures, they would need another
method (perhaps an irqwork), to request a deferred user space stack trace.
That can be dealt with later if one of theses architectures require this
feature.

Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
---
 kernel/unwind/deferred.c | 52 +++++++++++++++++++++++++++++++++-------
 1 file changed, 44 insertions(+), 8 deletions(-)

diff --git a/kernel/unwind/deferred.c b/kernel/unwind/deferred.c
index 2cbae2ada309..c5ac087d2396 100644
--- a/kernel/unwind/deferred.c
+++ b/kernel/unwind/deferred.c
@@ -12,6 +12,31 @@
 #include <linux/slab.h>
 #include <linux/mm.h>
 
+/*
+ * For requesting a deferred user space stack trace from NMI context
+ * the architecture must support a safe cmpxchg in NMI context.
+ * For those architectures that do not have that, then it cannot ask
+ * for a deferred user space stack trace from an NMI context. If it
+ * does, then it will get -EINVAL.
+ */
+#if defined(CONFIG_ARCH_HAVE_NMI_SAFE_CMPXCHG)
+# define CAN_USE_IN_NMI		1
+static inline bool try_assign_cnt(struct unwind_task_info *info, u32 cnt)
+{
+	u32 old = 0;
+
+	return try_cmpxchg(&info->id.cnt, &old, cnt);
+}
+#else
+# define CAN_USE_IN_NMI		0
+/* When NMIs are not allowed, this always succeeds */
+static inline bool try_assign_cnt(struct unwind_task_info *info, u32 cnt)
+{
+	info->id.cnt = cnt;
+	return true;
+}
+#endif
+
 /* Make the cache fit in a 4K page */
 #define UNWIND_MAX_ENTRIES					\
 	((SZ_4K - sizeof(struct unwind_cache)) / sizeof(long))
@@ -42,14 +67,13 @@ static DEFINE_PER_CPU(u32, unwind_ctx_ctr);
 static u64 get_cookie(struct unwind_task_info *info)
 {
 	u32 cnt = 1;
-	u32 old = 0;
 
 	if (info->id.cpu)
 		return info->id.id;
 
 	/* LSB is always set to ensure 0 is an invalid value */
 	cnt |= __this_cpu_read(unwind_ctx_ctr) + 2;
-	if (try_cmpxchg(&info->id.cnt, &old, cnt)) {
+	if (try_assign_cnt(info, cnt)) {
 		/* Update the per cpu counter */
 		__this_cpu_write(unwind_ctx_ctr, cnt);
 	}
@@ -167,31 +191,43 @@ static void unwind_deferred_task_work(struct callback_head *head)
 int unwind_deferred_request(struct unwind_work *work, u64 *cookie)
 {
 	struct unwind_task_info *info = &current->unwind_info;
+	long pending;
 	int ret;
 
 	*cookie = 0;
 
-	if (WARN_ON_ONCE(in_nmi()))
-		return -EINVAL;
-
 	if ((current->flags & (PF_KTHREAD | PF_EXITING)) ||
 	    !user_mode(task_pt_regs(current)))
 		return -EINVAL;
 
+	/*
+	 * NMI requires having safe cmpxchg operations.
+	 * Trigger a warning to make it obvious that an architecture
+	 * is using this in NMI when it should not be.
+	 */
+	if (WARN_ON_ONCE(!CAN_USE_IN_NMI && in_nmi()))
+		return -EINVAL;
+
 	guard(irqsave)();
 
 	*cookie = get_cookie(info);
 
 	/* callback already pending? */
-	if (info->pending)
+	pending = READ_ONCE(info->pending);
+	if (pending)
+		return 1;
+
+	/* Claim the work unless an NMI just now swooped in to do so. */
+	if (!try_cmpxchg(&info->pending, &pending, 1))
 		return 1;
 
 	/* The work has been claimed, now schedule it. */
 	ret = task_work_add(current, &info->work, TWA_RESUME);
-	if (WARN_ON_ONCE(ret))
+	if (WARN_ON_ONCE(ret)) {
+		WRITE_ONCE(info->pending, 0);
 		return ret;
+	}
 
-	info->pending = 1;
 	return 0;
 }
 
-- 
2.47.2

[PATCH v16 06/10] unwind deferred: Use bitmask to determine which callbacks to call

[Steven Rostedt]

From: Steven Rostedt <rostedt@goodmis.org>

In order to know which registered callback requested a stacktrace for when
the task goes back to user space, add a bitmask to keep track of all
registered tracers. The bitmask is the size of long, which means that on a
32 bit machine, it can have at most 32 registered tracers, and on 64 bit,
it can have at most 64 registered tracers. This should not be an issue as
there should not be more than 10 (unless BPF can abuse this?).

When a tracer registers with unwind_deferred_init() it will get a bit
number assigned to it. When a tracer requests a stacktrace, it will have
its bit set within the task_struct. When the task returns back to user
space, it will call the callbacks for all the registered tracers where
their bits are set in the task's mask.

When a tracer is removed by the unwind_deferred_cancel() all current tasks
will clear the associated bit, just in case another tracer gets registered
immediately afterward and then gets their callback called unexpectedly.

To prevent live locks from happening if an event that happens between the
task_work and when the task goes back to user space, triggers the deferred
unwind, have the unwind_mask get cleared on exit to user space and not
after the callback is made.

Move the pending bit from a value on the task_struct to bit zero of the
unwind_mask (saves space on the task_struct). This will allow modifying
the pending bit along with the work bits atomically.

Instead of clearing a work's bit after its callback is called, it is
delayed until exit. If the work is requested again, the task_work is not
queued again and the request will be notified that the task has already been
called by returning a positive number (the same as if it was already
pending).

The pending bit is cleared before calling the callback functions but the
current work bits remain. If one of the called works registers again, it
will not trigger a task_work if its bit is still present in the task's
unwind_mask.

If a new work requests a deferred unwind, then it will set both the
pending bit and its own bit. Note this will also cause any work that was
previously queued and had their callback already executed to be executed
again. Future work will remove these spurious callbacks.

The use of atomic_long bit operations were suggested by Peter Zijlstra:
Link: https://lore.kernel.org/all/20250715102912.GQ1613200@noisy.programming.kicks-ass.net/
The unwind_mask could not be converted to atomic_long_t do to atomic_long
not having all the bit operations needed by unwind_mask. Instead it
follows other use cases in the kernel and just typecasts the unwind_mask
to atomic_long_t when using the two atomic_long functions.

Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
---
 include/linux/unwind_deferred.h       | 26 +++++++-
 include/linux/unwind_deferred_types.h |  2 +-
 kernel/unwind/deferred.c              | 87 +++++++++++++++++++++------
 3 files changed, 92 insertions(+), 23 deletions(-)

diff --git a/include/linux/unwind_deferred.h b/include/linux/unwind_deferred.h
index 14efd8c027aa..337ead927d4d 100644
--- a/include/linux/unwind_deferred.h
+++ b/include/linux/unwind_deferred.h
@@ -13,10 +13,19 @@ typedef void (*unwind_callback_t)(struct unwind_work *work, struct unwind_stackt
 struct unwind_work {
 	struct list_head		list;
 	unwind_callback_t		func;
+	int				bit;
 };
 
 #ifdef CONFIG_UNWIND_USER
 
+enum {
+	UNWIND_PENDING_BIT = 0,
+};
+
+enum {
+	UNWIND_PENDING		= BIT(UNWIND_PENDING_BIT),
+};
+
 void unwind_task_init(struct task_struct *task);
 void unwind_task_free(struct task_struct *task);
 
@@ -28,15 +37,26 @@ void unwind_deferred_cancel(struct unwind_work *work);
 
 static __always_inline void unwind_reset_info(void)
 {
-	if (unlikely(current->unwind_info.id.id))
+	struct unwind_task_info *info = &current->unwind_info;
+	unsigned long bits;
+
+	/* Was there any unwinding? */
+	if (unlikely(info->unwind_mask)) {
+		bits = info->unwind_mask;
+		do {
+			/* Is a task_work going to run again before going back */
+			if (bits & UNWIND_PENDING)
+				return;
+		} while (!try_cmpxchg(&info->unwind_mask, &bits, 0UL));
 		current->unwind_info.id.id = 0;
+	}
 	/*
 	 * As unwind_user_faultable() can be called directly and
 	 * depends on nr_entries being cleared on exit to user,
 	 * this needs to be a separate conditional.
 	 */
-	if (unlikely(current->unwind_info.cache))
-		current->unwind_info.cache->nr_entries = 0;
+	if (unlikely(info->cache))
+		info->cache->nr_entries = 0;
 }
 
 #else /* !CONFIG_UNWIND_USER */
diff --git a/include/linux/unwind_deferred_types.h b/include/linux/unwind_deferred_types.h
index 104c477d5609..5dc9cda141ff 100644
--- a/include/linux/unwind_deferred_types.h
+++ b/include/linux/unwind_deferred_types.h
@@ -29,10 +29,10 @@ union unwind_task_id {
 };
 
 struct unwind_task_info {
+	unsigned long		unwind_mask;
 	struct unwind_cache	*cache;
 	struct callback_head	work;
 	union unwind_task_id	id;
-	int			pending;
 };
 
 #endif /* _LINUX_UNWIND_USER_DEFERRED_TYPES_H */
diff --git a/kernel/unwind/deferred.c b/kernel/unwind/deferred.c
index c5ac087d2396..e19f02ef416d 100644
--- a/kernel/unwind/deferred.c
+++ b/kernel/unwind/deferred.c
@@ -45,6 +45,16 @@ static inline bool try_assign_cnt(struct unwind_task_info *info, u32 cnt)
 static DEFINE_MUTEX(callback_mutex);
 static LIST_HEAD(callbacks);
 
+#define RESERVED_BITS	(UNWIND_PENDING)
+
+/* Zero'd bits are available for assigning callback users */
+static unsigned long unwind_mask = RESERVED_BITS;
+
+static inline bool unwind_pending(struct unwind_task_info *info)
+{
+	return test_bit(UNWIND_PENDING_BIT, &info->unwind_mask);
+}
+
 /*
  * This is a unique percpu identifier for a given task entry context.
  * Conceptually, it's incremented every time the CPU enters the kernel from
@@ -138,14 +148,15 @@ static void unwind_deferred_task_work(struct callback_head *head)
 	struct unwind_task_info *info = container_of(head, struct unwind_task_info, work);
 	struct unwind_stacktrace trace;
 	struct unwind_work *work;
+	unsigned long bits;
 	u64 cookie;
 
-	if (WARN_ON_ONCE(!info->pending))
+	if (WARN_ON_ONCE(!unwind_pending(info)))
 		return;
 
-	/* Allow work to come in again */
-	WRITE_ONCE(info->pending, 0);
-
+	/* Clear pending bit but make sure to have the current bits */
+	bits = atomic_long_fetch_andnot(UNWIND_PENDING,
+				  (atomic_long_t *)&info->unwind_mask);
 	/*
 	 * From here on out, the callback must always be called, even if it's
 	 * just an empty trace.
@@ -159,7 +170,8 @@ static void unwind_deferred_task_work(struct callback_head *head)
 
 	guard(mutex)(&callback_mutex);
 	list_for_each_entry(work, &callbacks, list) {
-		work->func(work, &trace, cookie);
+		if (test_bit(work->bit, &bits))
+			work->func(work, &trace, cookie);
 	}
 }
 
@@ -183,15 +195,16 @@ static void unwind_deferred_task_work(struct callback_head *head)
  * because it has already been previously called for the same entry context,
  * it will be called again with the same stack trace and cookie.
  *
- * Return: 1 if the the callback was already queued.
- *         0 if the callback successfully was queued.
+ * Return: 0 if the callback successfully was queued.
+ *         1 if the callback is pending or was already executed.
  *         Negative if there's an error.
  *         @cookie holds the cookie of the first request by any user
  */
 int unwind_deferred_request(struct unwind_work *work, u64 *cookie)
 {
 	struct unwind_task_info *info = &current->unwind_info;
-	long pending;
+	unsigned long old, bits;
+	unsigned long bit = BIT(work->bit);
 	int ret;
 
 	*cookie = 0;
@@ -212,32 +225,59 @@ int unwind_deferred_request(struct unwind_work *work, u64 *cookie)
 
 	*cookie = get_cookie(info);
 
-	/* callback already pending? */
-	pending = READ_ONCE(info->pending);
-	if (pending)
-		return 1;
+	old = READ_ONCE(info->unwind_mask);
 
-	/* Claim the work unless an NMI just now swooped in to do so. */
-	if (!try_cmpxchg(&info->pending, &pending, 1))
+	/* Is this already queued or executed */
+	if (old & bit)
 		return 1;
 
+	/*
+	 * This work's bit hasn't been set yet. Now set it with the PENDING
+	 * bit and fetch the current value of unwind_mask. If ether the
+	 * work's bit or PENDING was already set, then this is already queued
+	 * to have a callback.
+	 */
+	bits = UNWIND_PENDING | bit;
+	old = atomic_long_fetch_or(bits, (atomic_long_t *)&info->unwind_mask);
+	if (old & bits) {
+		/*
+		 * If the work's bit was set, whatever set it had better
+		 * have also set pending and queued a callback.
+		 */
+		WARN_ON_ONCE(!(old & UNWIND_PENDING));
+		return old & bit;
+	}
+
 	/* The work has been claimed, now schedule it. */
 	ret = task_work_add(current, &info->work, TWA_RESUME);
-	if (WARN_ON_ONCE(ret)) {
-		WRITE_ONCE(info->pending, 0);
-		return ret;
-	}
 
-	return 0;
+	if (WARN_ON_ONCE(ret))
+		WRITE_ONCE(info->unwind_mask, 0);
+
+	return ret;
 }
 
 void unwind_deferred_cancel(struct unwind_work *work)
 {
+	struct task_struct *g, *t;
+
 	if (!work)
 		return;
 
+	/* No work should be using a reserved bit */
+	if (WARN_ON_ONCE(BIT(work->bit) & RESERVED_BITS))
+		return;
+
 	guard(mutex)(&callback_mutex);
 	list_del(&work->list);
+
+	__clear_bit(work->bit, &unwind_mask);
+
+	guard(rcu)();
+	/* Clear this bit from all threads */
+	for_each_process_thread(g, t) {
+		clear_bit(work->bit, &t->unwind_info.unwind_mask);
+	}
 }
 
 int unwind_deferred_init(struct unwind_work *work, unwind_callback_t func)
@@ -245,6 +285,14 @@ int unwind_deferred_init(struct unwind_work *work, unwind_callback_t func)
 	memset(work, 0, sizeof(*work));
 
 	guard(mutex)(&callback_mutex);
+
+	/* See if there's a bit in the mask available */
+	if (unwind_mask == ~0UL)
+		return -EBUSY;
+
+	work->bit = ffz(unwind_mask);
+	__set_bit(work->bit, &unwind_mask);
+
 	list_add(&work->list, &callbacks);
 	work->func = func;
 	return 0;
@@ -256,6 +304,7 @@ void unwind_task_init(struct task_struct *task)
 
 	memset(info, 0, sizeof(*info));
 	init_task_work(&info->work, unwind_deferred_task_work);
+	info->unwind_mask = 0;
 }
 
 void unwind_task_free(struct task_struct *task)
-- 
2.47.2

[PATCH v16 07/10] unwind deferred: Add unwind_completed mask to stop spurious callbacks

[Steven Rostedt]

From: Steven Rostedt <rostedt@goodmis.org>

If there's more than one registered tracer to the unwind deferred
infrastructure, it is currently possible that one tracer could cause extra
callbacks to happen for another tracer if the former requests a deferred
stacktrace after the latter's callback was executed and before the task
went back to user space.

Here's an example of how this could occur:

  [Task enters kernel]
    tracer 1 request -> add cookie to its buffer
    tracer 1 request -> add cookie to its buffer
    <..>
    [ task work executes ]
    tracer 1 callback -> add trace + cookie to its buffer

    [tracer 2 requests and triggers the task work again]
    [ task work executes again ]
    tracer 1 callback -> add trace + cookie to its buffer
    tracer 2 callback -> add trace + cookie to its buffer
 [Task exits back to user space]

This is because the bit for tracer 1 gets set in the task's unwind_mask
when it did its request and does not get cleared until the task returns
back to user space. But if another tracer were to request another deferred
stacktrace, then the next task work will executed all tracer's callbacks
that have their bits set in the task's unwind_mask.

To fix this issue, add another mask called unwind_completed and place it
into the task's info->cache structure. The cache structure is allocated
on the first occurrence of a deferred stacktrace and this unwind_completed
mask is not needed until then. It's better to have it in the cache than to
permanently waste space in the task_struct.

After a tracer's callback is executed, it's bit gets set in this
unwind_completed mask. When the task_work enters, it will AND the task's
unwind_mask with the inverse of the unwind_completed which will eliminate
any work that already had its callback executed since the task entered the
kernel.

When the task leaves the kernel, it will reset this unwind_completed mask
just like it resets the other values as it enters user space.

Link: https://lore.kernel.org/all/20250716142609.47f0e4a5@batman.local.home/

Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
---
 include/linux/unwind_deferred.h       |  4 +++-
 include/linux/unwind_deferred_types.h |  1 +
 kernel/unwind/deferred.c              | 19 +++++++++++++++----
 3 files changed, 19 insertions(+), 5 deletions(-)

diff --git a/include/linux/unwind_deferred.h b/include/linux/unwind_deferred.h
index 337ead927d4d..b9ec4c8515c7 100644
--- a/include/linux/unwind_deferred.h
+++ b/include/linux/unwind_deferred.h
@@ -55,8 +55,10 @@ static __always_inline void unwind_reset_info(void)
 	 * depends on nr_entries being cleared on exit to user,
 	 * this needs to be a separate conditional.
 	 */
-	if (unlikely(info->cache))
+	if (unlikely(info->cache)) {
 		info->cache->nr_entries = 0;
+		info->cache->unwind_completed = 0;
+	}
 }
 
 #else /* !CONFIG_UNWIND_USER */
diff --git a/include/linux/unwind_deferred_types.h b/include/linux/unwind_deferred_types.h
index 5dc9cda141ff..33b62ac25c86 100644
--- a/include/linux/unwind_deferred_types.h
+++ b/include/linux/unwind_deferred_types.h
@@ -3,6 +3,7 @@
 #define _LINUX_UNWIND_USER_DEFERRED_TYPES_H
 
 struct unwind_cache {
+	unsigned long		unwind_completed;
 	unsigned int		nr_entries;
 	unsigned long		entries[];
 };
diff --git a/kernel/unwind/deferred.c b/kernel/unwind/deferred.c
index e19f02ef416d..a3d26014a2e6 100644
--- a/kernel/unwind/deferred.c
+++ b/kernel/unwind/deferred.c
@@ -166,12 +166,18 @@ static void unwind_deferred_task_work(struct callback_head *head)
 
 	unwind_user_faultable(&trace);
 
+	if (info->cache)
+		bits &= ~(info->cache->unwind_completed);
+
 	cookie = info->id.id;
 
 	guard(mutex)(&callback_mutex);
 	list_for_each_entry(work, &callbacks, list) {
-		if (test_bit(work->bit, &bits))
+		if (test_bit(work->bit, &bits)) {
 			work->func(work, &trace, cookie);
+			if (info->cache)
+				info->cache->unwind_completed |= BIT(work->bit);
+		}
 	}
 }
 
@@ -260,23 +266,28 @@ int unwind_deferred_request(struct unwind_work *work, u64 *cookie)
 void unwind_deferred_cancel(struct unwind_work *work)
 {
 	struct task_struct *g, *t;
+	int bit;
 
 	if (!work)
 		return;
 
+	bit = work->bit;
+
 	/* No work should be using a reserved bit */
-	if (WARN_ON_ONCE(BIT(work->bit) & RESERVED_BITS))
+	if (WARN_ON_ONCE(BIT(bit) & RESERVED_BITS))
 		return;
 
 	guard(mutex)(&callback_mutex);
 	list_del(&work->list);
 
-	__clear_bit(work->bit, &unwind_mask);
+	__clear_bit(bit, &unwind_mask);
 
 	guard(rcu)();
 	/* Clear this bit from all threads */
 	for_each_process_thread(g, t) {
-		clear_bit(work->bit, &t->unwind_info.unwind_mask);
+		clear_bit(bit, &t->unwind_info.unwind_mask);
+		if (t->unwind_info.cache)
+			clear_bit(bit, &t->unwind_info.cache->unwind_completed);
 	}
 }
 
-- 
2.47.2

[PATCH v16 08/10] unwind: Add USED bit to only have one conditional on way back to user space

[Steven Rostedt]

From: Steven Rostedt <rostedt@goodmis.org>

On the way back to user space, the function unwind_reset_info() is called
unconditionally (but always inlined). It currently has two conditionals.
One that checks the unwind_mask which is set whenever a deferred trace is
called and is used to know that the mask needs to be cleared. The other
checks if the cache has been allocated, and if so, it resets the
nr_entries so that the unwinder knows it needs to do the work to get a new
user space stack trace again (it only does it once per entering the
kernel).

Use one of the bits in the unwind mask as a "USED" bit that gets set
whenever a trace is created. This will make it possible to only check the
unwind_mask in the unwind_reset_info() to know if it needs to do work or
not and eliminates a conditional that happens every time the task goes
back to user space.

Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
---
 include/linux/unwind_deferred.h | 18 +++++++++---------
 kernel/unwind/deferred.c        |  5 ++++-
 2 files changed, 13 insertions(+), 10 deletions(-)

diff --git a/include/linux/unwind_deferred.h b/include/linux/unwind_deferred.h
index b9ec4c8515c7..2efbda01e959 100644
--- a/include/linux/unwind_deferred.h
+++ b/include/linux/unwind_deferred.h
@@ -20,10 +20,14 @@ struct unwind_work {
 
 enum {
 	UNWIND_PENDING_BIT = 0,
+	UNWIND_USED_BIT,
 };
 
 enum {
 	UNWIND_PENDING		= BIT(UNWIND_PENDING_BIT),
+
+	/* Set if the unwinding was used (directly or deferred) */
+	UNWIND_USED		= BIT(UNWIND_USED_BIT)
 };
 
 void unwind_task_init(struct task_struct *task);
@@ -49,15 +53,11 @@ static __always_inline void unwind_reset_info(void)
 				return;
 		} while (!try_cmpxchg(&info->unwind_mask, &bits, 0UL));
 		current->unwind_info.id.id = 0;
-	}
-	/*
-	 * As unwind_user_faultable() can be called directly and
-	 * depends on nr_entries being cleared on exit to user,
-	 * this needs to be a separate conditional.
-	 */
-	if (unlikely(info->cache)) {
-		info->cache->nr_entries = 0;
-		info->cache->unwind_completed = 0;
+
+		if (unlikely(info->cache)) {
+			info->cache->nr_entries = 0;
+			info->cache->unwind_completed = 0;
+		}
 	}
 }
 
diff --git a/kernel/unwind/deferred.c b/kernel/unwind/deferred.c
index a3d26014a2e6..2311b725d691 100644
--- a/kernel/unwind/deferred.c
+++ b/kernel/unwind/deferred.c
@@ -45,7 +45,7 @@ static inline bool try_assign_cnt(struct unwind_task_info *info, u32 cnt)
 static DEFINE_MUTEX(callback_mutex);
 static LIST_HEAD(callbacks);
 
-#define RESERVED_BITS	(UNWIND_PENDING)
+#define RESERVED_BITS	(UNWIND_PENDING | UNWIND_USED)
 
 /* Zero'd bits are available for assigning callback users */
 static unsigned long unwind_mask = RESERVED_BITS;
@@ -140,6 +140,9 @@ int unwind_user_faultable(struct unwind_stacktrace *trace)
 
 	cache->nr_entries = trace->nr;
 
+	/* Clear nr_entries on way back to user space */
+	set_bit(UNWIND_USED_BIT, &info->unwind_mask);
+
 	return 0;
 }
 
-- 
2.47.2

[PATCH v16 09/10] unwind deferred: Use SRCU unwind_deferred_task_work()

[Steven Rostedt]

From: Steven Rostedt <rostedt@goodmis.org>

Instead of using the callback_mutex to protect the link list of callbacks
in unwind_deferred_task_work(), use SRCU instead. This gets called every
time a task exits that has to record a stack trace that was requested.
This can happen for many tasks on several CPUs at the same time. A mutex
is a bottleneck and can cause a bit of contention and slow down performance.

As the callbacks themselves are allowed to sleep, regular RCU cannot be
used to protect the list. Instead use SRCU, as that still allows the
callbacks to sleep and the list can be read without needing to hold the
callback_mutex.

Link: https://lore.kernel.org/all/ca9bd83a-6c80-4ee0-a83c-224b9d60b755@efficios.com/

Cc: "Paul E. McKenney" <paulmck@kernel.org>
Suggested-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
---
 kernel/unwind/deferred.c | 27 +++++++++++++++++++++------
 1 file changed, 21 insertions(+), 6 deletions(-)

diff --git a/kernel/unwind/deferred.c b/kernel/unwind/deferred.c
index 2311b725d691..a5ef1c1f915e 100644
--- a/kernel/unwind/deferred.c
+++ b/kernel/unwind/deferred.c
@@ -41,7 +41,7 @@ static inline bool try_assign_cnt(struct unwind_task_info *info, u32 cnt)
 #define UNWIND_MAX_ENTRIES					\
 	((SZ_4K - sizeof(struct unwind_cache)) / sizeof(long))
 
-/* Guards adding to and reading the list of callbacks */
+/* Guards adding to or removing from the list of callbacks */
 static DEFINE_MUTEX(callback_mutex);
 static LIST_HEAD(callbacks);
 
@@ -49,6 +49,7 @@ static LIST_HEAD(callbacks);
 
 /* Zero'd bits are available for assigning callback users */
 static unsigned long unwind_mask = RESERVED_BITS;
+DEFINE_STATIC_SRCU(unwind_srcu);
 
 static inline bool unwind_pending(struct unwind_task_info *info)
 {
@@ -174,8 +175,9 @@ static void unwind_deferred_task_work(struct callback_head *head)
 
 	cookie = info->id.id;
 
-	guard(mutex)(&callback_mutex);
-	list_for_each_entry(work, &callbacks, list) {
+	guard(srcu)(&unwind_srcu);
+	list_for_each_entry_srcu(work, &callbacks, list,
+				 srcu_read_lock_held(&unwind_srcu)) {
 		if (test_bit(work->bit, &bits)) {
 			work->func(work, &trace, cookie);
 			if (info->cache)
@@ -213,7 +215,7 @@ int unwind_deferred_request(struct unwind_work *work, u64 *cookie)
 {
 	struct unwind_task_info *info = &current->unwind_info;
 	unsigned long old, bits;
-	unsigned long bit = BIT(work->bit);
+	unsigned long bit;
 	int ret;
 
 	*cookie = 0;
@@ -230,6 +232,14 @@ int unwind_deferred_request(struct unwind_work *work, u64 *cookie)
 	if (WARN_ON_ONCE(!CAN_USE_IN_NMI && in_nmi()))
 		return -EINVAL;
 
+	/* Do not allow cancelled works to request again */
+	bit = READ_ONCE(work->bit);
+	if (WARN_ON_ONCE(bit < 0))
+		return -EINVAL;
+
+	/* Only need the mask now */
+	bit = BIT(bit);
+
 	guard(irqsave)();
 
 	*cookie = get_cookie(info);
@@ -281,10 +291,15 @@ void unwind_deferred_cancel(struct unwind_work *work)
 		return;
 
 	guard(mutex)(&callback_mutex);
-	list_del(&work->list);
+	list_del_rcu(&work->list);
+
+	/* Do not allow any more requests and prevent callbacks */
+	work->bit = -1;
 
 	__clear_bit(bit, &unwind_mask);
 
+	synchronize_srcu(&unwind_srcu);
+
 	guard(rcu)();
 	/* Clear this bit from all threads */
 	for_each_process_thread(g, t) {
@@ -307,7 +322,7 @@ int unwind_deferred_init(struct unwind_work *work, unwind_callback_t func)
 	work->bit = ffz(unwind_mask);
 	__set_bit(work->bit, &unwind_mask);
 
-	list_add(&work->list, &callbacks);
+	list_add_rcu(&work->list, &callbacks);
 	work->func = func;
 	return 0;
 }
-- 
2.47.2

Re: [PATCH v16 09/10] unwind deferred: Use SRCU unwind_deferred_task_work()

[Paul E. McKenney]

On Tue, Jul 29, 2025 at 02:23:13PM -0400, Steven Rostedt wrote:
> From: Steven Rostedt <rostedt@goodmis.org>
> 
> Instead of using the callback_mutex to protect the link list of callbacks
> in unwind_deferred_task_work(), use SRCU instead. This gets called every
> time a task exits that has to record a stack trace that was requested.
> This can happen for many tasks on several CPUs at the same time. A mutex
> is a bottleneck and can cause a bit of contention and slow down performance.
> 
> As the callbacks themselves are allowed to sleep, regular RCU cannot be
> used to protect the list. Instead use SRCU, as that still allows the
> callbacks to sleep and the list can be read without needing to hold the
> callback_mutex.
> 
> Link: https://lore.kernel.org/all/ca9bd83a-6c80-4ee0-a83c-224b9d60b755@efficios.com/
> 
> Cc: "Paul E. McKenney" <paulmck@kernel.org>
> Suggested-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
> Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>

One quite likely stupid question below.

							Thanx, Paul

> ---
>  kernel/unwind/deferred.c | 27 +++++++++++++++++++++------
>  1 file changed, 21 insertions(+), 6 deletions(-)
> 
> diff --git a/kernel/unwind/deferred.c b/kernel/unwind/deferred.c
> index 2311b725d691..a5ef1c1f915e 100644
> --- a/kernel/unwind/deferred.c
> +++ b/kernel/unwind/deferred.c
> @@ -41,7 +41,7 @@ static inline bool try_assign_cnt(struct unwind_task_info *info, u32 cnt)
>  #define UNWIND_MAX_ENTRIES					\
>  	((SZ_4K - sizeof(struct unwind_cache)) / sizeof(long))
>  
> -/* Guards adding to and reading the list of callbacks */
> +/* Guards adding to or removing from the list of callbacks */
>  static DEFINE_MUTEX(callback_mutex);
>  static LIST_HEAD(callbacks);
>  
> @@ -49,6 +49,7 @@ static LIST_HEAD(callbacks);
>  
>  /* Zero'd bits are available for assigning callback users */
>  static unsigned long unwind_mask = RESERVED_BITS;
> +DEFINE_STATIC_SRCU(unwind_srcu);
>  
>  static inline bool unwind_pending(struct unwind_task_info *info)
>  {
> @@ -174,8 +175,9 @@ static void unwind_deferred_task_work(struct callback_head *head)
>  
>  	cookie = info->id.id;
>  
> -	guard(mutex)(&callback_mutex);
> -	list_for_each_entry(work, &callbacks, list) {
> +	guard(srcu)(&unwind_srcu);
> +	list_for_each_entry_srcu(work, &callbacks, list,
> +				 srcu_read_lock_held(&unwind_srcu)) {
>  		if (test_bit(work->bit, &bits)) {
>  			work->func(work, &trace, cookie);
>  			if (info->cache)
> @@ -213,7 +215,7 @@ int unwind_deferred_request(struct unwind_work *work, u64 *cookie)
>  {
>  	struct unwind_task_info *info = &current->unwind_info;
>  	unsigned long old, bits;
> -	unsigned long bit = BIT(work->bit);
> +	unsigned long bit;
>  	int ret;
>  
>  	*cookie = 0;
> @@ -230,6 +232,14 @@ int unwind_deferred_request(struct unwind_work *work, u64 *cookie)
>  	if (WARN_ON_ONCE(!CAN_USE_IN_NMI && in_nmi()))
>  		return -EINVAL;
>  
> +	/* Do not allow cancelled works to request again */
> +	bit = READ_ONCE(work->bit);
> +	if (WARN_ON_ONCE(bit < 0))
> +		return -EINVAL;
> +
> +	/* Only need the mask now */
> +	bit = BIT(bit);
> +
>  	guard(irqsave)();
>  
>  	*cookie = get_cookie(info);
> @@ -281,10 +291,15 @@ void unwind_deferred_cancel(struct unwind_work *work)
>  		return;
>  
>  	guard(mutex)(&callback_mutex);
> -	list_del(&work->list);

What happens if unwind_deferred_task_work() finds this item right here...

> +	list_del_rcu(&work->list);

...and then unwind_deferred_request() does its WARN_ON_ONCE() check
against -1 right here?

Can't that cause UAF?

This is quite possibly a stupid question because I am not seeing where to
apply this patch, so I don't know what other mechanisms might be in place.

> +	/* Do not allow any more requests and prevent callbacks */
> +	work->bit = -1;
>  
>  	__clear_bit(bit, &unwind_mask);
>  
> +	synchronize_srcu(&unwind_srcu);
> +
>  	guard(rcu)();
>  	/* Clear this bit from all threads */
>  	for_each_process_thread(g, t) {
> @@ -307,7 +322,7 @@ int unwind_deferred_init(struct unwind_work *work, unwind_callback_t func)
>  	work->bit = ffz(unwind_mask);
>  	__set_bit(work->bit, &unwind_mask);
>  
> -	list_add(&work->list, &callbacks);
> +	list_add_rcu(&work->list, &callbacks);
>  	work->func = func;
>  	return 0;
>  }
> -- 
> 2.47.2
> 
> 

Re: [PATCH v16 09/10] unwind deferred: Use SRCU unwind_deferred_task_work()

[Steven Rostedt]

On Thu, 31 Jul 2025 17:29:44 -0700
"Paul E. McKenney" <paulmck@kernel.org> wrote:

> > @@ -281,10 +291,15 @@ void unwind_deferred_cancel(struct unwind_work *work)
> >  		return;
> >  
> >  	guard(mutex)(&callback_mutex);
> > -	list_del(&work->list);  
> 
> What happens if unwind_deferred_task_work() finds this item right here...

Should be fine.

> 
> > +	list_del_rcu(&work->list);  
> 
> ...and then unwind_deferred_request() does its WARN_ON_ONCE() check
> against -1 right here?

If unwind_deferred_request() is called after unwind_deferred_cancel()
then that's a bug. As both functions are called by the tracer. The
cancel() function is for the tracer to tell this infrastructure that
it's done with the deferred tracing. If it calls a request() function
after (or during) the call to cancel(), then it's a bug in the tracer.
The tracer is responsible for making sure it will not do any more
requests before calling the cancel() function.

But what the tracer can't do is to know if there's a pending request
still happening and this has to handle that.

> 
> Can't that cause UAF?
> 
> This is quite possibly a stupid question because I am not seeing where to
> apply this patch, so I don't know what other mechanisms might be in place.

Yeah, you were only Cc'd on this patch because it was the only one that
uses RCU. You can see the entire series here:

  https://lore.kernel.org/all/20250729182304.965835871@kernel.org/

Or in patchwork:

  https://patchwork.kernel.org/project/linux-trace-kernel/list/?series=986813

-- Steve


> 
> > +	/* Do not allow any more requests and prevent callbacks */
> > +	work->bit = -1;
> >  
> >  	__clear_bit(bit, &unwind_mask);
> >  
> > +	synchronize_srcu(&unwind_srcu);
> > +
> >  	guard(rcu)();
> >  	/* Clear this bit from all threads */
> >  	for_each_process_thread(g, t) {

[PATCH v16 10/10] unwind: Finish up unwind when a task exits

[Steven Rostedt]

From: Steven Rostedt <rostedt@goodmis.org>

On do_exit() when a task is exiting, if a unwind is requested and the
deferred user stacktrace is deferred via the task_work, the task_work
callback is called after exit_mm() is called in do_exit(). This means that
the user stack trace will not be retrieved and an empty stack is created.

Instead, add a function unwind_deferred_task_exit() and call it just
before exit_mm() so that the unwinder can call the requested callbacks
with the user space stack.

Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
---
 include/linux/unwind_deferred.h |  3 +++
 kernel/exit.c                   |  2 ++
 kernel/unwind/deferred.c        | 23 ++++++++++++++++++++---
 3 files changed, 25 insertions(+), 3 deletions(-)

diff --git a/include/linux/unwind_deferred.h b/include/linux/unwind_deferred.h
index 2efbda01e959..26122d00708a 100644
--- a/include/linux/unwind_deferred.h
+++ b/include/linux/unwind_deferred.h
@@ -39,6 +39,8 @@ int unwind_deferred_init(struct unwind_work *work, unwind_callback_t func);
 int unwind_deferred_request(struct unwind_work *work, u64 *cookie);
 void unwind_deferred_cancel(struct unwind_work *work);
 
+void unwind_deferred_task_exit(struct task_struct *task);
+
 static __always_inline void unwind_reset_info(void)
 {
 	struct unwind_task_info *info = &current->unwind_info;
@@ -71,6 +73,7 @@ static inline int unwind_deferred_init(struct unwind_work *work, unwind_callback
 static inline int unwind_deferred_request(struct unwind_work *work, u64 *timestamp) { return -ENOSYS; }
 static inline void unwind_deferred_cancel(struct unwind_work *work) {}
 
+static inline void unwind_deferred_task_exit(struct task_struct *task) {}
 static inline void unwind_reset_info(void) {}
 
 #endif /* !CONFIG_UNWIND_USER */
diff --git a/kernel/exit.c b/kernel/exit.c
index bb184a67ac73..1d8c8ac33c4f 100644
--- a/kernel/exit.c
+++ b/kernel/exit.c
@@ -68,6 +68,7 @@
 #include <linux/rethook.h>
 #include <linux/sysfs.h>
 #include <linux/user_events.h>
+#include <linux/unwind_deferred.h>
 #include <linux/uaccess.h>
 #include <linux/pidfs.h>
 
@@ -938,6 +939,7 @@ void __noreturn do_exit(long code)
 
 	tsk->exit_code = code;
 	taskstats_exit(tsk, group_dead);
+	unwind_deferred_task_exit(tsk);
 	trace_sched_process_exit(tsk, group_dead);
 
 	/*
diff --git a/kernel/unwind/deferred.c b/kernel/unwind/deferred.c
index a5ef1c1f915e..dc6040aae3ee 100644
--- a/kernel/unwind/deferred.c
+++ b/kernel/unwind/deferred.c
@@ -114,7 +114,7 @@ int unwind_user_faultable(struct unwind_stacktrace *trace)
 	/* Should always be called from faultable context */
 	might_fault();
 
-	if (current->flags & PF_EXITING)
+	if (!current->mm)
 		return -EINVAL;
 
 	if (!info->cache) {
@@ -147,9 +147,9 @@ int unwind_user_faultable(struct unwind_stacktrace *trace)
 	return 0;
 }
 
-static void unwind_deferred_task_work(struct callback_head *head)
+static void process_unwind_deferred(struct task_struct *task)
 {
-	struct unwind_task_info *info = container_of(head, struct unwind_task_info, work);
+	struct unwind_task_info *info = &task->unwind_info;
 	struct unwind_stacktrace trace;
 	struct unwind_work *work;
 	unsigned long bits;
@@ -186,6 +186,23 @@ static void unwind_deferred_task_work(struct callback_head *head)
 	}
 }
 
+static void unwind_deferred_task_work(struct callback_head *head)
+{
+	process_unwind_deferred(current);
+}
+
+void unwind_deferred_task_exit(struct task_struct *task)
+{
+	struct unwind_task_info *info = &current->unwind_info;
+
+	if (!unwind_pending(info))
+		return;
+
+	process_unwind_deferred(task);
+
+	task_work_cancel(task, &info->work);
+}
+
 /**
  * unwind_deferred_request - Request a user stacktrace on task kernel exit
  * @work: Unwind descriptor requesting the trace
-- 
2.47.2

Re: [PATCH v16 00/10] unwind_user: Deferred unwinding infrastructure

[patchwork-bot+netdevbpf]

Hello:

This series was applied to bpf/bpf-next.git (master)
by Steven Rostedt (Google) <rostedt@goodmis.org>:

On Tue, 29 Jul 2025 14:23:04 -0400 you wrote:
> Jen Remus suggested some updates from v15:
> 
>   https://lore.kernel.org/linux-trace-kernel/20250725185512.673587297@kernel.org/
> 
> Those were:
> 
> - Make fp_frame into a constant
> 
> [...]

Here is the summary with links:
  - [v16,01/10] unwind_user: Add user space unwinding API with frame pointer support
    https://git.kernel.org/bpf/bpf-next/c/71753c6ed2bf
  - [v16,02/10] unwind_user/deferred: Add unwind_user_faultable()
    https://git.kernel.org/bpf/bpf-next/c/5e32d0f15cc5
  - [v16,03/10] unwind_user/deferred: Add unwind cache
    https://git.kernel.org/bpf/bpf-next/c/b9c73524106e
  - [v16,04/10] unwind_user/deferred: Add deferred unwinding interface
    https://git.kernel.org/bpf/bpf-next/c/2dffa355f6c2
  - [v16,05/10] unwind_user/deferred: Make unwind deferral requests NMI-safe
    https://git.kernel.org/bpf/bpf-next/c/055c7060e7ca
  - [v16,06/10] unwind deferred: Use bitmask to determine which callbacks to call
    https://git.kernel.org/bpf/bpf-next/c/be3d526a5b34
  - [v16,07/10] unwind deferred: Add unwind_completed mask to stop spurious callbacks
    https://git.kernel.org/bpf/bpf-next/c/4c75133e745a
  - [v16,08/10] unwind: Add USED bit to only have one conditional on way back to user space
    https://git.kernel.org/bpf/bpf-next/c/858fa8a3b083
  - [v16,09/10] unwind deferred: Use SRCU unwind_deferred_task_work()
    https://git.kernel.org/bpf/bpf-next/c/357eda2d7450
  - [v16,10/10] unwind: Finish up unwind when a task exits
    https://git.kernel.org/bpf/bpf-next/c/b3b9cb11aa03

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html