linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Steven Rostedt <rostedt@kernel.org>
To: linux-kernel@vger.kernel.org, linux-trace-kernel@vger.kernel.org,
	bpf@vger.kernel.org, x86@kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>,
	Mathieu Desnoyers <mathieu.desnoyers@efficios.com>,
	Josh Poimboeuf <jpoimboe@kernel.org>,
	Peter Zijlstra <peterz@infradead.org>,
	Ingo Molnar <mingo@kernel.org>, Jiri Olsa <jolsa@kernel.org>,
	Namhyung Kim <namhyung@kernel.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	Andrii Nakryiko <andrii@kernel.org>,
	Indu Bhagat <indu.bhagat@oracle.com>,
	"Jose E. Marchesi" <jemarch@gnu.org>,
	Beau Belgrave <beaub@linux.microsoft.com>,
	Jens Remus <jremus@linux.ibm.com>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Andrew Morton <akpm@linux-foundation.org>,
	Jens Axboe <axboe@kernel.dk>, Florian Weimer <fweimer@redhat.com>,
	Sam James <sam@gentoo.org>
Subject: [PATCH v13 07/14] unwind_user/deferred: Make unwind deferral requests NMI-safe
Date: Mon, 07 Jul 2025 21:22:46 -0400	[thread overview]
Message-ID: <20250708012358.831631671@kernel.org> (raw)
In-Reply-To: 20250708012239.268642741@kernel.org

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.
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>
---
Changes since v12: https://lore.kernel.org/20250701005451.737614486@goodmis.org

- Now that the timestamp has been replaced by a cookie that uses only a 32
  bit cmpxchg(), this code just checks if the architecture has a safe
  cmpxchg that can be used in NMI and doesn't do the 64 bit check.
  Only the pending value is converted to local_t.

 include/linux/unwind_deferred_types.h |  4 +-
 kernel/unwind/deferred.c              | 56 ++++++++++++++++++++++-----
 2 files changed, 49 insertions(+), 11 deletions(-)

diff --git a/include/linux/unwind_deferred_types.h b/include/linux/unwind_deferred_types.h
index 79b4f8cece53..cd95ed1c8610 100644
--- a/include/linux/unwind_deferred_types.h
+++ b/include/linux/unwind_deferred_types.h
@@ -2,6 +2,8 @@
 #ifndef _LINUX_UNWIND_USER_DEFERRED_TYPES_H
 #define _LINUX_UNWIND_USER_DEFERRED_TYPES_H
 
+#include <asm/local.h>
+
 struct unwind_cache {
 	unsigned int		nr_entries;
 	unsigned long		entries[];
@@ -20,7 +22,7 @@ struct unwind_task_info {
 	struct unwind_cache	*cache;
 	struct callback_head	work;
 	union unwind_task_id	id;
-	int			pending;
+	local_t			pending;
 };
 
 #endif /* _LINUX_UNWIND_USER_DEFERRED_TYPES_H */
diff --git a/kernel/unwind/deferred.c b/kernel/unwind/deferred.c
index b1faaa55e5d5..2417e4ebbc82 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))
@@ -43,7 +68,6 @@ static u64 get_cookie(struct unwind_task_info *info)
 {
 	u32 cpu_cnt;
 	u32 cnt;
-	u32 old = 0;
 
 	if (info->id.cpu)
 		return info->id.id;
@@ -52,7 +76,7 @@ static u64 get_cookie(struct unwind_task_info *info)
 	cpu_cnt += 2;
 	cnt = cpu_cnt | 1; /* Always make non zero */
 
-	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, cpu_cnt);
 	}
@@ -119,11 +143,11 @@ static void unwind_deferred_task_work(struct callback_head *head)
 	struct unwind_work *work;
 	u64 cookie;
 
-	if (WARN_ON_ONCE(!info->pending))
+	if (WARN_ON_ONCE(!local_read(&info->pending)))
 		return;
 
 	/* Allow work to come in again */
-	WRITE_ONCE(info->pending, 0);
+	local_set(&info->pending, 0);
 
 	/*
 	 * From here on out, the callback must always be called, even if it's
@@ -170,31 +194,43 @@ 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 */
+	if (!CAN_USE_IN_NMI && in_nmi())
+		return -EINVAL;
+
 	guard(irqsave)();
 
 	*cookie = get_cookie(info);
 
 	/* callback already pending? */
-	if (info->pending)
+	pending = local_read(&info->pending);
+	if (pending)
 		return 1;
 
+	if (CAN_USE_IN_NMI) {
+		/* Claim the work unless an NMI just now swooped in to do so. */
+		if (!local_try_cmpxchg(&info->pending, &pending, 1))
+			return 1;
+	} else {
+		local_set(&info->pending, 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)) {
+		local_set(&info->pending, 0);
 		return ret;
+	}
 
-	info->pending = 1;
 	return 0;
 }
 
-- 
2.47.2



  parent reply	other threads:[~2025-07-08  1:23 UTC|newest]

Thread overview: 45+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-07-08  1:22 [PATCH v13 00/14] unwind_user: x86: Deferred unwinding infrastructure Steven Rostedt
2025-07-08  1:22 ` [PATCH v13 01/14] unwind_user: Add user space unwinding API Steven Rostedt
2025-07-08  1:22 ` [PATCH v13 02/14] unwind_user: Add frame pointer support Steven Rostedt
2025-07-09 10:01   ` Jens Remus
2025-07-10 12:28     ` Jens Remus
2025-07-10 15:21     ` Steven Rostedt
2025-07-10 15:41       ` Jens Remus
2025-07-10 17:08         ` Steven Rostedt
2025-07-14 12:52           ` Jens Remus
2025-07-08  1:22 ` [PATCH v13 03/14] unwind_user: Add compat mode " Steven Rostedt
2025-07-08  1:22 ` [PATCH v13 04/14] unwind_user/deferred: Add unwind_user_faultable() Steven Rostedt
2025-07-08  1:22 ` [PATCH v13 05/14] unwind_user/deferred: Add unwind cache Steven Rostedt
2025-07-08  1:22 ` [PATCH v13 06/14] unwind_user/deferred: Add deferred unwinding interface Steven Rostedt
2025-07-08  1:22 ` Steven Rostedt [this message]
2025-07-14 13:29   ` [PATCH v13 07/14] unwind_user/deferred: Make unwind deferral requests NMI-safe Peter Zijlstra
2025-07-14 14:19     ` Steven Rostedt
2025-07-14 15:05       ` Peter Zijlstra
2025-07-14 15:11         ` Steven Rostedt
2025-07-15  9:09           ` Peter Zijlstra
2025-07-15 12:35             ` Steven Rostedt
2025-07-08  1:22 ` [PATCH v13 08/14] unwind deferred: Use bitmask to determine which callbacks to call Steven Rostedt
2025-07-08  1:22 ` [PATCH v13 09/14] unwind deferred: Use SRCU unwind_deferred_task_work() Steven Rostedt
2025-07-14 13:56   ` Peter Zijlstra
2025-07-14 14:21     ` Steven Rostedt
2025-07-14 15:03       ` Peter Zijlstra
2025-07-08  1:22 ` [PATCH v13 10/14] unwind: Clear unwind_mask on exit back to user space Steven Rostedt
2025-07-15 10:29   ` Peter Zijlstra
2025-07-15 12:49     ` Steven Rostedt
2025-07-15 18:06       ` Steven Rostedt
2025-07-15 18:10         ` Steven Rostedt
2025-07-15 18:26         ` Steven Rostedt
2025-07-15 19:04         ` Peter Zijlstra
2025-07-15 19:01       ` Peter Zijlstra
2025-07-15 17:20     ` Steven Rostedt
2025-07-15 19:07       ` Peter Zijlstra
2025-07-15 22:01     ` Steven Rostedt
2025-07-16 18:26     ` Steven Rostedt
2025-07-16 18:33       ` Steven Rostedt
2025-07-16 19:25         ` Steven Rostedt
2025-07-08  1:22 ` [PATCH v13 11/14] unwind: Add USED bit to only have one conditional on way " Steven Rostedt
2025-07-08  1:22 ` [PATCH v13 12/14] unwind: Finish up unwind when a task exits Steven Rostedt
2025-07-08  1:22 ` [PATCH v13 13/14] unwind_user/x86: Enable frame pointer unwinding on x86 Steven Rostedt
2025-07-11  8:43   ` David Laight
2025-07-11 16:11     ` Steven Rostedt
2025-07-08  1:22 ` [PATCH v13 14/14] unwind_user/x86: Enable compat mode " Steven Rostedt

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20250708012358.831631671@kernel.org \
    --to=rostedt@kernel.org \
    --cc=akpm@linux-foundation.org \
    --cc=andrii@kernel.org \
    --cc=axboe@kernel.dk \
    --cc=beaub@linux.microsoft.com \
    --cc=bpf@vger.kernel.org \
    --cc=fweimer@redhat.com \
    --cc=indu.bhagat@oracle.com \
    --cc=jemarch@gnu.org \
    --cc=jolsa@kernel.org \
    --cc=jpoimboe@kernel.org \
    --cc=jremus@linux.ibm.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-trace-kernel@vger.kernel.org \
    --cc=mathieu.desnoyers@efficios.com \
    --cc=mhiramat@kernel.org \
    --cc=mingo@kernel.org \
    --cc=namhyung@kernel.org \
    --cc=peterz@infradead.org \
    --cc=sam@gentoo.org \
    --cc=tglx@linutronix.de \
    --cc=torvalds@linux-foundation.org \
    --cc=x86@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).