Skip to contentSkip to conversation

tracing: Read user data from futex system call trace event

PATCH Superseded Steven Rostedt 3 participants 10 messages Mar 10 – Mar 12

Subsystems (3)
Memory management Tracing Futex

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Steven Rostedt

author
#0 lore raw

v2 cover tracing: Read user data from futex system call trace event

We are looking at the performance of futexes and require a bit more information when tracing them.

The two patches here extend the system call reading of user space to create specific handling of the futex system call. It now reads the user space relevant data (the addr, utime and addr2), as well as parses the flags. This adds a little smarts to the trace event as it only shows the parameters that are relevant, as well as parses utime as either a timespec or as val2 depending on the futex_op.

Here's an example of the new output:

sys_futex(uaddr: 0x56196292e830 (0), FUTEX_WAKE|FUTEX_PRIVATE_FLAG)
sys_futex(uaddr: 0x56196292e834 (0x4a7) tid: 1191, FUTEX_UNLOCK_PI|FUTEX_PRIVATE_FLAG)
sys_futex(uaddr: 0x56196292e834 (0) tid: 0, FUTEX_LOCK_PI|FUTEX_PRIVATE_FLAG)
sys_futex(uaddr: 0x56196292e830 (0), FUTEX_WAIT|FUTEX_PRIVATE_FLAG)
sys_futex(uaddr: 0x56196292e838 (0), FUTEX_WAIT_REQUEUE_PI|FUTEX_PRIVATE_FLAG, timespec: 0x7ffc1b91a9f0 (163.048528790), uaddr2: 0x56196292e834 (4aa), val3: 0) sys_futex(uaddr: 0x56196292e834 (0x4aa) tid: 1194, FUTEX_LOCK_PI|FUTEX_PRIVATE_FLAG)
sys_futex(uaddr: 0x56196292e838 (0), FUTEX_WAIT_REQUEUE_PI|FUTEX_PRIVATE_FLAG, timespec: 0x7ffc1b91a9f0 (163.048528790), uaddr2: 0x56196292e834 (800004aa), val3: 0) sys_futex(uaddr: 0x7f7ed6b29990 (0x4ab), FUTEX_WAIT_BITSET|FUTEX_CLOCK_REALTIME)
sys_futex(uaddr: 0x56196292e834 (0x800004aa) tid: 1194 (WAITERS), FUTEX_LOCK_PI|FUTEX_PRIVATE_FLAG)
sys_futex(uaddr: 0x56196292e838 (0), FUTEX_WAIT_REQUEUE_PI|FUTEX_PRIVATE_FLAG, timespec: 0x7ffc1b91a9f0 (163.048528790), uaddr2: 0x56196292e834 (800004aa), val3: 0) sys_futex(uaddr: 0x56196292e834 (0x800004aa) tid: 1194 (WAITERS), FUTEX_LOCK_PI|FUTEX_PRIVATE_FLAG)

Changes since v1: https://lore.kernel.org/all/[email protected]/

- Moved the printing of the futex parameters to kernel/futex/syscall.c to keep it closer to the actual system call.

- Moved the always_inline functions: futex_cmd_has_timeout() and futex_cmd_has_addr2() into include/linux/futex.h so that they can be both used by the futex and tracing syscall code.

Steven Rostedt (3):
tracing: Have futex syscall trace event show specific user data
tracing: Update futex syscall trace event to show more commands
tracing: Show TID and flags for PI futex system call trace event

----

 include/linux/futex.h         |  38 +++++-
 kernel/futex/syscalls.c       | 129 ++++++++++++++++++---
 kernel/trace/trace_syscalls.c | 261 +++++++++++++++++++++++++++++++++++++++++-
 3 files changed, 411 insertions(+), 17 deletions(-)
author
#1 lore raw

PATCH v2 1/3 tracing: Have futex syscall trace event show specific user data

From: Steven Rostedt <[email protected]>

Add specific reporting of the futex system call. This allows for debugging the futex code a bit easier. Instead of just showing the values passed into the futex system call, read the value of the user space memory pointed to by the addr parameter.

Also make the op parameter more readable by parsing the values to show what the command is:

 futex_requeue_p-3251    [002] .....  2101.068479: sys_futex(uaddr: 0x55e79a4da834 (0x80000cb1), FUTEX_LOCK_PI|FUTEX_PRIVATE_FLAG, val: 0)
 futex_requeue_p-3248    [001] .....  2101.068970: sys_futex(uaddr: 0x7f859072f990 (0xcb2), FUTEX_WAIT_BITSET|FUTEX_CLOCK_REALTIME, val: 3250)
 futex_requeue_p-3252    [005] .....  2101.069108: sys_futex(uaddr: 0x55e79a4da838 (0), FUTEX_WAIT_REQUEUE_PI|FUTEX_PRIVATE_FLAG, val: 0, timespec: 0x7ffe61076aa0, uaddr2: 0x55e79a4da834, uaddr2: 94453214586932, val3: 0)
 futex_requeue_p-3252    [005] .....  2101.069410: sys_futex(uaddr: 0x55e79a4da834 (0x80000cb1), FUTEX_LOCK_PI|FUTEX_PRIVATE_FLAG, val: 0)
Signed-off-by Steven Rostedt (Google) <[email protected]>

---
Changes since v1: https://lore.kernel.org/all/[email protected]/

- Moved the printing of the futex parameters to kernel/futex/syscall.c to keep it closer to the actual system call.

 include/linux/futex.h         |   7 +-
 kernel/futex/syscalls.c       |  89 ++++++++++++++++++
 kernel/trace/trace_syscalls.c | 168 +++++++++++++++++++++++++++++++++-
 3 files changed, 260 insertions(+), 4 deletions(-)
include/linux/futex.h +6 -1
index 9e9750f04980..9fc47aa01a8b 100644
@@ -82,6 +82,10 @@ long do_futex(u32 __user *uaddr, int op, u32 val, ktime_t *timeout,
 	      u32 __user *uaddr2, u32 val2, u32 val3);
 int futex_hash_prctl(unsigned long arg2, unsigned long arg3, unsigned long arg4);
 
+#ifdef CONFIG_FTRACE_SYSCALLS
+void futex_print_syscall(struct seq_buf *s, int nr_args, unsigned long *args, u32 *ptr);
+#endif
+
 #ifdef CONFIG_FUTEX_PRIVATE_HASH
 int futex_hash_allocate_default(void);
 void futex_hash_free(struct mm_struct *mm);

@@ -114,7 +118,8 @@ static inline int futex_hash_allocate_default(void)
 }
 static inline int futex_hash_free(struct mm_struct *mm) { return 0; }
 static inline int futex_mm_init(struct mm_struct *mm) { return 0; }
-
+static inline void futex_print_syscall(struct seq_buf *s, int nr_args,
+				       unsigned long *args, u32 *ptr) { }
 #endif
 
 #endif
kernel/futex/syscalls.c +89 -0

101 lines not on this page · raw message

kernel/trace/trace_syscalls.c +165 -3

258 lines not on this page · raw message

Signature · 1 line
2.51.0
build bot
#2 lore raw

Re: [PATCH v2 1/3] tracing: Have futex syscall trace event show specific user data

kernel test robot noticed the following build warnings:

Hi Steven,

kernel test robot noticed the following build warnings:

[auto build test WARNING on tip/locking/core]
[also build test WARNING on trace/for-next akpm-mm/mm-everything linus/master v7.0-rc3 next-20260310] [If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Steven-Rostedt/tracing-Have-futex-syscall-trace-event-show-specific-user-data/20260311-041422
base:   tip/locking/core
patch link:    https://lore.kernel.org/r/20260310201036.542627924%40kernel.org
patch subject: [PATCH v2 1/3] tracing: Have futex syscall trace event show specific user data
config: x86_64-rhel-9.4 (https://download.01.org/0day-ci/archive/20260311/[email protected]/config)
compiler: gcc-14 (Debian 14.2.0-19) 14.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20260311/[email protected]/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/

All warnings (new ones prefixed by >>):

kernel/trace/trace_syscalls.c: In function 'syscall_get_futex':

>> kernel/trace/trace_syscalls.c:807:15: warning: variable 'buf' set but not used [-Wunused-but-set-variable]
     807 |         char *buf;
         |               ^~~

vim +/buf +807 kernel/trace/trace_syscalls.c

801
802 static int
803 syscall_get_futex(unsigned long *args, char **buffer, int *size, int buf_size)
804 {
805 struct syscall_user_buffer *sbuf;
806 const char __user *ptr;

 > 807		char *buf;

808
809 /* buf_size of zero means user doesn't want user space read */
810 if (!buf_size)
811 return -1;
812
813 /* If the syscall_buffer is NULL, tracing is being shutdown */
814 sbuf = READ_ONCE(syscall_buffer);
815 if (!sbuf)
816 return -1;
817
818 ptr = (char __user *)args[0];
819
820 *buffer = trace_user_fault_read(&sbuf->buf, ptr, 4, NULL, NULL);
821 if (!*buffer)
822 return -1;
823
824 /* Add room for the value */
825 *size += 4;
826
827 buf = *buffer;
828
829 return 0;
830 }
831

Signature · 2 lines
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki
maintainer
#3 lore raw

Re: [PATCH v2 1/3] tracing: Have futex syscall trace event show specific user data

In reply to Steven Rostedt
Quoted 10 lines · On Tue, 10 Mar 2026 16:09:55 -0400 Steven Rostedt <[email protected]> wrote:
kernel/futex/syscalls.c +10 -0
+#ifdef CONFIG_FTRACE_SYSCALLS
+static const char * __futex_cmds[] =
+{
+	"FUTEX_WAIT", "FUTEX_WAKE", "FUTEX_FD", "FUTEX_REQUEUE",
+	"FUTEX_CMP_REQUEUE", "FUTEX_WAKE_OP", "FUTEX_LOCK_PI",
+	"FUTEX_UNLOCK_PI", "FUTEX_TRYLOCK_PI", "FUTEX_WAIT_BITSET",
+	"FUTEX_WAKE_BITSET", "FUTEX_WAIT_REQUEUE_PI", "FUTEX_CMP_REQUEUE_PI",
+	"FUTEX_LOCK_PI2",
+};
+

[...]

Quoted 41 lines
kernel/trace/trace_syscalls.c +37 -0
@@ -437,6 +464,69 @@ sys_enter_openat_print_fmt(struct syscall_metadata *entry, char *buf, int len)
 	return pos;
 }
 
+static int __init
+sys_enter_futex_print_fmt(struct syscall_metadata *entry, char *buf, int len)
+{
+	int pos = 0;
+
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"\"uaddr: 0x%%lx (0x%%lx) cmd=%%s%%s%%s");
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"  val: 0x%%x timeout/val2: 0x%%llx");
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			" uaddr2: 0x%%lx val3: 0x%%x\", ");
+
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			" REC->uaddr,");
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			" REC->__value,");
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"  __print_symbolic(REC->op & 0x%x, ", FUTEX_CMD_MASK);
+
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_WAIT\"}, ", FUTEX_WAIT);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_WAKE\"}, ", FUTEX_WAKE);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_FD\"}, ", FUTEX_FD);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_REQUEUE\"}, ", FUTEX_REQUEUE);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_CMP_REQUEUE\"}, ", FUTEX_CMP_REQUEUE);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_WAKE_OP\"}, ", FUTEX_WAKE_OP);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_LOCK_PI\"}, ", FUTEX_LOCK_PI);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_UNLOCK_PI\"}, ", FUTEX_UNLOCK_PI);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_TRYLOCK_PI\"}, ", FUTEX_TRYLOCK_PI);

kernel/trace/trace_syscalls.c sys_enter_openat_print_fmt(struct syscall_metadata *entry, char *buf, int len) · line 513

Quoted 10 lines
kernel/trace/trace_syscalls.c +10 -0
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_WAIT_BITSET\"}, ", FUTEX_WAIT_BITSET);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_WAKE_BITSET\"}, ", FUTEX_WAKE_BITSET);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_WAIT_REQUEUE_PI\"}, ", FUTEX_WAIT_REQUEUE_PI);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_CMP_REQUEUE_PI\"}, ", FUTEX_CMP_REQUEUE_PI);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_LOCK_PI2\"}),", FUTEX_LOCK_PI2);

Hmm can we share __futex_cmds[] with kernel/futex/syscalls.c? Then these could be

for (i = 0; i <= FUTEX_LOCK_PI2; i++)

	pos += snprintf(buf + pos, LEN_OR_ZERO,
			"{%d, \"%s\"}%s", i, __futex_cmds[i],
			i == FUTEX_LOCK_PI2 ? ")," : ", ");
Quoted 19 lines
kernel/trace/trace_syscalls.c +16 -0
+
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			" (REC->op & %d) ? \"|FUTEX_PRIVATE_FLAG\" : \"\",",
+			FUTEX_PRIVATE_FLAG);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			" (REC->op & %d) ? \"|FUTEX_CLOCK_REALTIME\" : \"\",",
+			FUTEX_CLOCK_REALTIME);
+
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			" REC->val, REC->utime,");
+
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			" REC->uaddr, REC->val3");
+	return pos;
+}
+
 static int __init
 __set_enter_print_fmt(struct syscall_metadata *entry, char *buf, int len)
 {

[...]

Quoted 20 lines
kernel/trace/trace_syscalls.c +16 -0
@@ -689,6 +799,48 @@ static int syscall_copy_user_array(char *buf, const char __user *ptr,
 	return 0;
 }
 
+static int
+syscall_get_futex(unsigned long *args, char **buffer, int *size, int buf_size)
+{
+	struct syscall_user_buffer *sbuf;
+	const char __user *ptr;
+	char *buf;
+
+	/* buf_size of zero means user doesn't want user space read */
+	if (!buf_size)
+		return -1;
+
+	/* If the syscall_buffer is NULL, tracing is being shutdown */
+	sbuf = READ_ONCE(syscall_buffer);
+	if (!sbuf)
+		return -1;
+

kernel/trace/trace_syscalls.c static int syscall_copy_user_array(char *buf, const char __user *ptr, · line 827

Quoted 10 lines
kernel/trace/trace_syscalls.c +10 -0
+	ptr = (char __user *)args[0];
+
+	*buffer = trace_user_fault_read(&sbuf->buf, ptr, 4, NULL, NULL);
+	if (!*buffer)
+		return -1;
+
+	/* Add room for the value */
+	*size += 4;
+
+	buf = *buffer;

As kernel test bot says, this does nothing. (*buffer is already assigned)

kernel/trace/trace_syscalls.c static int syscall_copy_user_array(char *buf, const char __user *ptr, · line 830

Quoted 3 lines
kernel/trace/trace_syscalls.c +3 -0
+
+	return 0;
+}

Thanks,

Signature · 1 line
Masami Hiramatsu (Google) <[email protected]>
author
#4 lore raw

Re: [PATCH v2 1/3] tracing: Have futex syscall trace event show specific user data

Quoted 35 lines · On Wed, 11 Mar 2026 18:03:25 +0900 Masami Hiramatsu (Google)…
Quoted 21 lines
kernel/trace/trace_syscalls.c +21 -0
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"  __print_symbolic(REC->op & 0x%x, ", FUTEX_CMD_MASK);
+
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_WAIT\"}, ", FUTEX_WAIT);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_WAKE\"}, ", FUTEX_WAKE);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_FD\"}, ", FUTEX_FD);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_REQUEUE\"}, ", FUTEX_REQUEUE);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_CMP_REQUEUE\"}, ", FUTEX_CMP_REQUEUE);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_WAKE_OP\"}, ", FUTEX_WAKE_OP);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_LOCK_PI\"}, ", FUTEX_LOCK_PI);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_UNLOCK_PI\"}, ", FUTEX_UNLOCK_PI);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_TRYLOCK_PI\"}, ", FUTEX_TRYLOCK_PI);

kernel/trace/trace_syscalls.c sys_enter_openat_print_fmt(struct syscall_metadata *entry, char *buf, int len) · line 513

Quoted 10 lines
kernel/trace/trace_syscalls.c +10 -0
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_WAIT_BITSET\"}, ", FUTEX_WAIT_BITSET);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_WAKE_BITSET\"}, ", FUTEX_WAKE_BITSET);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_WAIT_REQUEUE_PI\"}, ", FUTEX_WAIT_REQUEUE_PI);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_CMP_REQUEUE_PI\"}, ", FUTEX_CMP_REQUEUE_PI);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			"{%d, \"FUTEX_LOCK_PI2\"}),", FUTEX_LOCK_PI2);

Hmm can we share __futex_cmds[] with kernel/futex/syscalls.c? Then these could be

Quoted 4 lines

for (i = 0; i <= FUTEX_LOCK_PI2; i++)

	pos += snprintf(buf + pos, LEN_OR_ZERO,
			"{%d, \"%s\"}%s", i, __futex_cmds[i],
			i == FUTEX_LOCK_PI2 ? ")," : ", ");

Hmm, I created the above *before* creating the __futex_cmds. But yes, that makes sense. Thanks for the suggestion.

Quoted 46 lines
Quoted 21 lines
kernel/trace/trace_syscalls.c +16 -0
+
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			" (REC->op & %d) ? \"|FUTEX_PRIVATE_FLAG\" : \"\",",
+			FUTEX_PRIVATE_FLAG);
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			" (REC->op & %d) ? \"|FUTEX_CLOCK_REALTIME\" : \"\",",
+			FUTEX_CLOCK_REALTIME);
+
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			" REC->val, REC->utime,");
+
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			" REC->uaddr, REC->val3");
+	return pos;
+}
+
 static int __init
 __set_enter_print_fmt(struct syscall_metadata *entry, char *buf, int len)
 {

[...]

Quoted 24 lines
kernel/trace/trace_syscalls.c +20 -0
@@ -689,6 +799,48 @@ static int syscall_copy_user_array(char *buf, const char __user *ptr,
 	return 0;
 }
 
+static int
+syscall_get_futex(unsigned long *args, char **buffer, int *size, int buf_size)
+{
+	struct syscall_user_buffer *sbuf;
+	const char __user *ptr;
+	char *buf;
+
+	/* buf_size of zero means user doesn't want user space read */
+	if (!buf_size)
+		return -1;
+
+	/* If the syscall_buffer is NULL, tracing is being shutdown */
+	sbuf = READ_ONCE(syscall_buffer);
+	if (!sbuf)
+		return -1;
+
+	ptr = (char __user *)args[0];
+
+	*buffer = trace_user_fault_read(&sbuf->buf, ptr, 4, NULL, NULL);
+	if (!*buffer)
Quoted 9 lines

kernel/trace/trace_syscalls.c static int syscall_copy_user_array(char *buf, const char __user *ptr, · line 827

Quoted 6 lines
kernel/trace/trace_syscalls.c +6 -0
+		return -1;
+
+	/* Add room for the value */
+	*size += 4;
+
+	buf = *buffer;

As kernel test bot says, this does nothing. (*buffer is already assigned)

Oops, I think this was a cut-and-paste error :-p

-- Steve

build bot
#5 lore raw

Re: [PATCH v2 1/3] tracing: Have futex syscall trace event show specific user data

In reply to Steven Rostedt
kernel test robot noticed the following build warnings:

Hi Steven,

kernel test robot noticed the following build warnings:

[auto build test WARNING on tip/locking/core]
[also build test WARNING on trace/for-next akpm-mm/mm-everything linus/master v7.0-rc3 next-20260310] [If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Steven-Rostedt/tracing-Have-futex-syscall-trace-event-show-specific-user-data/20260311-041422
base:   tip/locking/core
patch link:    https://lore.kernel.org/r/20260310201036.542627924%40kernel.org
patch subject: [PATCH v2 1/3] tracing: Have futex syscall trace event show specific user data
config: arc-randconfig-001-20260311 (https://download.01.org/0day-ci/archive/20260311/[email protected]/config)
compiler: arc-linux-gcc (GCC) 14.3.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20260311/[email protected]/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/

All warnings (new ones prefixed by >>):

kernel/trace/trace_syscalls.c: In function 'syscall_get_futex':

>> kernel/trace/trace_syscalls.c:807:15: warning: variable 'buf' set but not used [-Wunused-but-set-variable]
     807 |         char *buf;
         |               ^~~

vim +/buf +807 kernel/trace/trace_syscalls.c

801
802 static int
803 syscall_get_futex(unsigned long *args, char **buffer, int *size, int buf_size)
804 {
805 struct syscall_user_buffer *sbuf;
806 const char __user *ptr;

 > 807		char *buf;

808
809 /* buf_size of zero means user doesn't want user space read */
810 if (!buf_size)
811 return -1;
812
813 /* If the syscall_buffer is NULL, tracing is being shutdown */
814 sbuf = READ_ONCE(syscall_buffer);
815 if (!sbuf)
816 return -1;
817
818 ptr = (char __user *)args[0];
819
820 *buffer = trace_user_fault_read(&sbuf->buf, ptr, 4, NULL, NULL);
821 if (!*buffer)
822 return -1;
823
824 /* Add room for the value */
825 *size += 4;
826
827 buf = *buffer;
828
829 return 0;
830 }
831

Signature · 2 lines
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki
author
#6 lore raw

PATCH v2 2/3 tracing: Update futex syscall trace event to show more commands

In reply to Steven Rostedt

From: Steven Rostedt <[email protected]>

Make the futex syscall trace event a little more smart. Have it read the futex_op instruction to determine what else it can save and print. For the appropriate options, it will read the utime (timespec) parameter and show its output as well as the uaddr2.

 futex_requeue_p-1154    [004] .....   144.568339: sys_futex(uaddr: 0x5652b178d834 (0x482), FUTEX_UNLOCK_PI|FUTEX_PRIVATE_FLAG, val: 0)
 futex_requeue_p-1162    [002] .....   144.568696: sys_futex(uaddr: 0x7f763b7fece0 (2), FUTEX_WAIT|FUTEX_PRIVATE_FLAG, val: 2)
 futex_requeue_p-1151    [000] .....   144.568700: sys_futex(uaddr: 0x7f763b7fece0 (0), FUTEX_WAKE|FUTEX_PRIVATE_FLAG, val: 1)
 futex_requeue_p-1162    [002] .....   144.568705: sys_futex(uaddr: 0x7f763b7fece0 (0), FUTEX_WAKE|FUTEX_PRIVATE_FLAG, val: 1)
 futex_requeue_p-1151    [000] .....   144.568715: sys_futex(uaddr: 0x7f764369e990 (0x483), FUTEX_WAIT_BITSET|FUTEX_CLOCK_REALTIME, val: 1155)
 futex_requeue_p-1155    [005] .....   144.569420: sys_futex(uaddr: 0x5652b178d838 (0), FUTEX_WAIT_REQUEUE_PI|FUTEX_PRIVATE_FLAG, val: 0, timespec: 0x7ffdacfba500 (143.890024054), uaddr2: 0x5652b178d834 (0), val3: 0)
 futex_requeue_p-1155    [005] .....   144.569454: sys_futex(uaddr: 0x5652b178d834 (0), FUTEX_LOCK_PI|FUTEX_PRIVATE_FLAG, val: 0)
Signed-off-by Steven Rostedt (Google) <[email protected]>

---
Changes since v1: https://lore.kernel.org/all/[email protected]/

- Updated to have the print processing in kernel/futex/syscall.c

 include/linux/futex.h         |  35 ++++++++-
 kernel/futex/syscalls.c       |  48 ++++++-------
 kernel/trace/trace_syscalls.c | 129 +++++++++++++++++++++++++++++-----
 3 files changed, 164 insertions(+), 48 deletions(-)
include/linux/futex.h +33 -2
index 9fc47aa01a8b..976fa257ab5c 100644
@@ -82,8 +82,35 @@ long do_futex(u32 __user *uaddr, int op, u32 val, ktime_t *timeout,
 	      u32 __user *uaddr2, u32 val2, u32 val3);
 int futex_hash_prctl(unsigned long arg2, unsigned long arg3, unsigned long arg4);
 
+static __always_inline bool futex_cmd_has_timeout(u32 cmd)
+{
+	switch (cmd) {
+	case FUTEX_WAIT:
+	case FUTEX_LOCK_PI:
+	case FUTEX_LOCK_PI2:
+	case FUTEX_WAIT_BITSET:
+	case FUTEX_WAIT_REQUEUE_PI:
+		return true;
+	}
+	return false;
+}
+
+static __always_inline bool futex_cmd_has_addr2(u32 cmd)
+{
+	switch (cmd) {
+	case FUTEX_REQUEUE:
+	case FUTEX_CMP_REQUEUE:
+	case FUTEX_WAKE_OP:
+	case FUTEX_WAIT_REQUEUE_PI:
+		return true;
+	}
+	return false;
+}
+
 #ifdef CONFIG_FTRACE_SYSCALLS
-void futex_print_syscall(struct seq_buf *s, int nr_args, unsigned long *args, u32 *ptr);
+void futex_print_syscall(struct seq_buf *s, int nr_args, unsigned long *args,
+			 u32 *ptr1, u32 *ptr2, unsigned long *ts1,
+			 unsigned long *ts2);
 #endif
 
 #ifdef CONFIG_FUTEX_PRIVATE_HASH

@@ -119,7 +146,11 @@ static inline int futex_hash_allocate_default(void)
 static inline int futex_hash_free(struct mm_struct *mm) { return 0; }
 static inline int futex_mm_init(struct mm_struct *mm) { return 0; }
 static inline void futex_print_syscall(struct seq_buf *s, int nr_args,
-				       unsigned long *args, u32 *ptr) { }
+				       unsigned long *args, u32 *ptr1,
+				       u32 *ptr2, unsigned long *ts1,
+				       unsigned long *ts2) { }
+static __always_inline bool futex_cmd_has_timeout(u32 cmd) { return false; }
+static __always_inline bool futex_cmd_has_addr2(u32 cmd) { return false; }
 #endif
 
 #endif
kernel/futex/syscalls.c +20 -28

78 lines not on this page · raw message

kernel/trace/trace_syscalls.c +111 -18

195 lines not on this page · raw message

Signature · 1 line
2.51.0
maintainer Reviewed-by on v2 2/3
#7 lore raw

Re: [PATCH v2 2/3] tracing: Update futex syscall trace event to show more commands

Quoted 15 lines · On Tue, 10 Mar 2026 16:09:56 -0400 Steven Rostedt <[email protected]> wrote:

From: Steven Rostedt <[email protected]>

Make the futex syscall trace event a little more smart. Have it read the futex_op instruction to determine what else it can save and print. For the appropriate options, it will read the utime (timespec) parameter and show its output as well as the uaddr2.

 futex_requeue_p-1154    [004] .....   144.568339: sys_futex(uaddr: 0x5652b178d834 (0x482), FUTEX_UNLOCK_PI|FUTEX_PRIVATE_FLAG, val: 0)
 futex_requeue_p-1162    [002] .....   144.568696: sys_futex(uaddr: 0x7f763b7fece0 (2), FUTEX_WAIT|FUTEX_PRIVATE_FLAG, val: 2)
 futex_requeue_p-1151    [000] .....   144.568700: sys_futex(uaddr: 0x7f763b7fece0 (0), FUTEX_WAKE|FUTEX_PRIVATE_FLAG, val: 1)
 futex_requeue_p-1162    [002] .....   144.568705: sys_futex(uaddr: 0x7f763b7fece0 (0), FUTEX_WAKE|FUTEX_PRIVATE_FLAG, val: 1)
 futex_requeue_p-1151    [000] .....   144.568715: sys_futex(uaddr: 0x7f764369e990 (0x483), FUTEX_WAIT_BITSET|FUTEX_CLOCK_REALTIME, val: 1155)
 futex_requeue_p-1155    [005] .....   144.569420: sys_futex(uaddr: 0x5652b178d838 (0), FUTEX_WAIT_REQUEUE_PI|FUTEX_PRIVATE_FLAG, val: 0, timespec: 0x7ffdacfba500 (143.890024054), uaddr2: 0x5652b178d834 (0), val3: 0)
 futex_requeue_p-1155    [005] .....   144.569454: sys_futex(uaddr: 0x5652b178d834 (0), FUTEX_LOCK_PI|FUTEX_PRIVATE_FLAG, val: 0)
Quoted 1 line
Signed-off-by Steven Rostedt (Google) <[email protected]>

This looks good to me.

Reviewed-by Masami Hiramatsu (Google) <[email protected]>

Thanks,

Quoted 361 lines

---
Changes since v1: https://lore.kernel.org/all/[email protected]/

- Updated to have the print processing in kernel/futex/syscall.c

 include/linux/futex.h         |  35 ++++++++-
 kernel/futex/syscalls.c       |  48 ++++++-------
 kernel/trace/trace_syscalls.c | 129 +++++++++++++++++++++++++++++-----
 3 files changed, 164 insertions(+), 48 deletions(-)
include/linux/futex.h +33 -2
index 9fc47aa01a8b..976fa257ab5c 100644
@@ -82,8 +82,35 @@ long do_futex(u32 __user *uaddr, int op, u32 val, ktime_t *timeout,
 	      u32 __user *uaddr2, u32 val2, u32 val3);
 int futex_hash_prctl(unsigned long arg2, unsigned long arg3, unsigned long arg4);
 
+static __always_inline bool futex_cmd_has_timeout(u32 cmd)
+{
+	switch (cmd) {
+	case FUTEX_WAIT:
+	case FUTEX_LOCK_PI:
+	case FUTEX_LOCK_PI2:
+	case FUTEX_WAIT_BITSET:
+	case FUTEX_WAIT_REQUEUE_PI:
+		return true;
+	}
+	return false;
+}
+
+static __always_inline bool futex_cmd_has_addr2(u32 cmd)
+{
+	switch (cmd) {
+	case FUTEX_REQUEUE:
+	case FUTEX_CMP_REQUEUE:
+	case FUTEX_WAKE_OP:
+	case FUTEX_WAIT_REQUEUE_PI:
+		return true;
+	}
+	return false;
+}
+
 #ifdef CONFIG_FTRACE_SYSCALLS
-void futex_print_syscall(struct seq_buf *s, int nr_args, unsigned long *args, u32 *ptr);
+void futex_print_syscall(struct seq_buf *s, int nr_args, unsigned long *args,
+			 u32 *ptr1, u32 *ptr2, unsigned long *ts1,
+			 unsigned long *ts2);
 #endif
 
 #ifdef CONFIG_FUTEX_PRIVATE_HASH

@@ -119,7 +146,11 @@ static inline int futex_hash_allocate_default(void)
 static inline int futex_hash_free(struct mm_struct *mm) { return 0; }
 static inline int futex_mm_init(struct mm_struct *mm) { return 0; }
 static inline void futex_print_syscall(struct seq_buf *s, int nr_args,
-				       unsigned long *args, u32 *ptr) { }
+				       unsigned long *args, u32 *ptr1,
+				       u32 *ptr2, unsigned long *ts1,
+				       unsigned long *ts2) { }
+static __always_inline bool futex_cmd_has_timeout(u32 cmd) { return false; }
+static __always_inline bool futex_cmd_has_addr2(u32 cmd) { return false; }
 #endif
 
 #endif
kernel/futex/syscalls.c +20 -28
index a1cd512aa502..a46706d6bc6c 100644
@@ -158,31 +158,6 @@ long do_futex(u32 __user *uaddr, int op, u32 val, ktime_t *timeout,
 	return -ENOSYS;
 }
 
-static __always_inline bool futex_cmd_has_timeout(u32 cmd)
-{
-	switch (cmd) {
-	case FUTEX_WAIT:
-	case FUTEX_LOCK_PI:
-	case FUTEX_LOCK_PI2:
-	case FUTEX_WAIT_BITSET:
-	case FUTEX_WAIT_REQUEUE_PI:
-		return true;
-	}
-	return false;
-}
-
-static __always_inline bool futex_cmd_has_addr2(u32 cmd)
-{
-	switch (cmd) {
-	case FUTEX_REQUEUE:
-	case FUTEX_CMP_REQUEUE:
-	case FUTEX_WAKE_OP:
-	case FUTEX_WAIT_REQUEUE_PI:
-		return true;
-	}
-	return false;
-}
-
 static __always_inline int
 futex_init_timeout(u32 cmd, u32 op, struct timespec64 *ts, ktime_t *t)
 {

@@ -229,7 +204,9 @@ static const char * __futex_cmds[] =
 	"FUTEX_LOCK_PI2",
 };
 
-void futex_print_syscall(struct seq_buf *s, int nr_args, unsigned long *args, u32 *ptr)
+void futex_print_syscall(struct seq_buf *s, int nr_args, unsigned long *args,
+			 u32 *ptr1, u32 *ptr2, unsigned long *ts1,
+			 unsigned long *ts2)
 {
 	unsigned int op, cmd;
 	bool done = false;

@@ -242,8 +219,8 @@ void futex_print_syscall(struct seq_buf *s, int nr_args, unsigned long *args, u3
 		switch (i) {
 		case 0:
 			seq_buf_printf(s, "uaddr: 0x%lx", args[i]);
-			if (ptr) {
-				u32 val = *ptr;
+			if (ptr1) {
+				u32 val = *ptr1;
 				if (val < 10)
 					seq_buf_printf(s, " (%u)", val);
 				else

@@ -279,6 +256,15 @@ void futex_print_syscall(struct seq_buf *s, int nr_args, unsigned long *args, u3
 				continue;
 
 			seq_buf_printf(s, ", timespec: 0x%lx", args[i]);
+
+			if (!ts1 || !ts2)
+				continue;
+
+			if (!*ts1 && !*ts2) {
+				seq_buf_puts(s, " (0)");
+				continue;
+			}
+			seq_buf_printf(s, " (%lu.%09lu)", *ts1, *ts2);
 			continue;
 		case 4:
 			if (!futex_cmd_has_addr2(cmd)) {

@@ -286,6 +272,12 @@ void futex_print_syscall(struct seq_buf *s, int nr_args, unsigned long *args, u3
 				continue;
 			}
 			seq_buf_printf(s, ", uaddr2: 0x%lx", args[i]);
+
+			if (!ptr2)
+				continue;
+
+			seq_buf_printf(s, " (%x)", *ptr2);
+
 			continue;
 		case 5:
 			seq_buf_printf(s, ", val3: %lu", args[i]);
kernel/trace/trace_syscalls.c +111 -18
index 8cb3af569157..de1fa97547a3 100644
@@ -239,21 +239,35 @@ sys_enter_openat_print(struct syscall_trace_enter *trace, struct syscall_metadat
 	return trace_handle_return(s);
 }
 
+struct futex_data {
+	u32		val1;
+	u32		val2;
+	unsigned long	ts1;
+	unsigned long	ts2;
+};
+
 static enum print_line_t
 sys_enter_futex_print(struct syscall_trace_enter *trace, struct syscall_metadata *entry,
 		      struct trace_seq *s, struct trace_event *event, int ent_size)
 {
+	struct futex_data *data;
 	void *end = (void *)trace + ent_size;
-	void *ptr;
+	unsigned long *ts1 = NULL, *ts2 = NULL;
+	u32 *ptr1 = NULL, *ptr2 = NULL;
 
 	/* Set ptr to the user space copied area */
-	ptr = (void *)trace->args + sizeof(unsigned long) * entry->nb_args;
-	if (ptr + 4 > end)
-		ptr = NULL;
+	data = (void *)trace->args + sizeof(unsigned long) * entry->nb_args;
+	if ((void *)data + sizeof(*data) <= end) {
+		ptr1 = &data->val1;
+		ptr2 = &data->val2;
+		ts1 = &data->ts1;
+		ts2 = &data->ts2;
+	}
 
 	trace_seq_printf(s, "%s(", entry->name);
 
-	futex_print_syscall(&s->seq, entry->nb_args, trace->args, ptr);
+	futex_print_syscall(&s->seq, entry->nb_args, trace->args, ptr1, ptr2,
+			    ts1, ts2);
 
 	trace_seq_puts(s, ")\n");
 

@@ -472,9 +486,9 @@ sys_enter_futex_print_fmt(struct syscall_metadata *entry, char *buf, int len)
 	pos += snprintf(buf + pos, LEN_OR_ZERO,
 			"\"uaddr: 0x%%lx (0x%%lx) cmd=%%s%%s%%s");
 	pos += snprintf(buf + pos, LEN_OR_ZERO,
-			"  val: 0x%%x timeout/val2: 0x%%llx");
+			"  val: 0x%%x timeout/val2: 0x%%llx (%%lu.%%lu)");
 	pos += snprintf(buf + pos, LEN_OR_ZERO,
-			" uaddr2: 0x%%lx val3: 0x%%x\", ");
+			" uaddr2: 0x%%lx (0x%%lx) val3: 0x%%x\", ");
 
 	pos += snprintf(buf + pos, LEN_OR_ZERO,
 			" REC->uaddr,");

@@ -520,10 +534,12 @@ sys_enter_futex_print_fmt(struct syscall_metadata *entry, char *buf, int len)
 			FUTEX_CLOCK_REALTIME);
 
 	pos += snprintf(buf + pos, LEN_OR_ZERO,
-			" REC->val, REC->utime,");
+			" REC->val, REC->utime, REC->__ts1, REC->__ts2,");
 
 	pos += snprintf(buf + pos, LEN_OR_ZERO,
-			" REC->uaddr, REC->val3");
+			" REC->uaddr,");
+	pos += snprintf(buf + pos, LEN_OR_ZERO,
+			" REC->__value2, REC->val3");
 	return pos;
 }
 

@@ -626,7 +642,39 @@ static int __init futex_fields(struct trace_event_call *call, int offset)
 	ret = trace_define_field(call, "u32", arg, offset, sizeof(int), 0,
 				 FILTER_OTHER);
 	if (ret)
-		kfree(arg);
+		goto free;
+	offset += sizeof(int);
+
+	arg = kstrdup("__value2", GFP_KERNEL);
+	if (WARN_ON_ONCE(!arg))
+		return -ENOMEM;
+	ret = trace_define_field(call, "u32", arg, offset, sizeof(int), 0,
+				 FILTER_OTHER);
+	if (ret)
+		goto free;
+	offset += sizeof(int);
+
+	arg = kstrdup("__ts1", GFP_KERNEL);
+	if (WARN_ON_ONCE(!arg))
+		return -ENOMEM;
+	ret = trace_define_field(call, "unsigned long", arg, offset,
+				 sizeof(unsigned long), 0, FILTER_OTHER);
+	if (ret)
+		goto free;
+	offset += sizeof(long);
+
+	arg = kstrdup("__ts2", GFP_KERNEL);
+	if (WARN_ON_ONCE(!arg))
+		return -ENOMEM;
+	ret = trace_define_field(call, "unsigned long", arg, offset,
+				 sizeof(unsigned long), 0, FILTER_OTHER);
+	if (ret)
+		goto free;
+
+	return 0;
+
+free:
+	kfree(arg);
 	return ret;
 }
 

@@ -799,11 +847,51 @@ static int syscall_copy_user_array(char *buf, const char __user *ptr,
 	return 0;
 }
 
+struct tp_futex_data {
+	u32			cmd;
+	const u32		__user *val1;
+	const u32 		__user *val2;
+	void			__user *timeout;
+};
+
+static int syscall_copy_futex(char *buf, const char __user *ptr,
+			      size_t size, void *data)
+{
+	struct tp_futex_data *tp_data = data;
+	struct futex_data *fdata = (void *)buf;
+	int cmd = tp_data->cmd & FUTEX_CMD_MASK;
+	int ret;
+
+	memset(fdata, 0, sizeof(*fdata));
+
+	if (tp_data->val1) {
+		ret = __copy_from_user(&fdata->val1, tp_data->val1, 4);
+		if (ret)
+			return -1;
+	}
+
+	if (tp_data->val2 && futex_cmd_has_addr2(cmd)) {
+		ret = __copy_from_user(&fdata->val2, tp_data->val2, 4);
+		if (ret)
+			return -1;
+	}
+
+	if (tp_data->timeout && futex_cmd_has_timeout(cmd)) {
+		/* Copies both ts1 and ts2 */
+		ret = __copy_from_user(&fdata->ts1, tp_data->timeout,
+				       sizeof(long) * 2);
+		if (ret)
+			return -1;
+	}
+
+	return 0;
+}
+
 static int
 syscall_get_futex(unsigned long *args, char **buffer, int *size, int buf_size)
 {
 	struct syscall_user_buffer *sbuf;
-	const char __user *ptr;
+	struct tp_futex_data tp_data;
 	char *buf;
 
 	/* buf_size of zero means user doesn't want user space read */

@@ -815,14 +903,18 @@ syscall_get_futex(unsigned long *args, char **buffer, int *size, int buf_size)
 	if (!sbuf)
 		return -1;
 
-	ptr = (char __user *)args[0];
+	tp_data.cmd = args[1];
+	tp_data.val1 = (u32 __user *)args[0];
+	tp_data.val2 = (u32 __user *)args[4];
+	tp_data.timeout = (u64 __user *)args[3];
 
-	*buffer = trace_user_fault_read(&sbuf->buf, ptr, 4, NULL, NULL);
+	*buffer = trace_user_fault_read(&sbuf->buf, NULL, 0,
+					syscall_copy_futex, &tp_data);
 	if (!*buffer)
 		return -1;
 
-	/* Add room for the value */
-	*size += 4;
+	/* Add room for values */
+	*size += sizeof(struct futex_data);
 
 	buf = *buffer;
 

@@ -833,12 +925,13 @@ static void syscall_put_futex(struct syscall_metadata *sys_data,
 			      struct syscall_trace_enter *entry,
 			      char *buffer)
 {
-	u32 *ptr;
+	struct futex_data *fdata = (void *)buffer;
+	struct futex_data *data;
 
 	/* Place the futex key into the storage */
-	ptr = (void *)entry->args + sizeof(unsigned long) * sys_data->nb_args;
+	data = (void *)entry->args + sizeof(unsigned long) * sys_data->nb_args;
 
-	*ptr = *(u32 *)buffer;
+	*data = *fdata;
 }
 
 static char *sys_fault_user(unsigned int buf_size,
Signature · 1 line
2.51.0
Signature · 1 line
Masami Hiramatsu (Google) <[email protected]>
author
#8 lore raw

PATCH v2 3/3 tracing: Show TID and flags for PI futex system call trace event

In reply to Steven Rostedt

From: Steven Rostedt <[email protected]>

For the futex system call trace event for FUTEX_LOCK_PI and FUTEX_UNLOCK_PI commands, show the TID from the value (which is usually in hex) as well as translate the flags (DIED and WAITERS).

 pi_mutex_hammer-1098    [000] .....   121.876928: sys_futex(uaddr: 0x560f40cc8180 (0x450) tid: 1104, FUTEX_LOCK_PI|FUTEX_PRIVATE_FLAG, val: 0, timespec: 0x7f2f9d4b1e50 (0.000100000))
 pi_mutex_hammer-1128    [000] .....   121.877120: sys_futex(uaddr: 0x560f40cc8180 (0x8000042a) tid: 1066 (WAITERS), FUTEX_LOCK_PI|FUTEX_PRIVATE_FLAG, val: 0, timespec: 0x7f2f8e493e50 (0.000100000))
 pi_mutex_hammer-1106    [000] .....   121.877242: sys_futex(uaddr: 0x560f40cc8180 (0x80000452) tid: 1106 (WAITERS), FUTEX_UNLOCK_PI|FUTEX_PRIVATE_FLAG, val: 0)

This makes it easier to see the hand off of a mutex and who the owner was.

Signed-off-by Steven Rostedt (Google) <[email protected]>

---
Changes since v1: https://lore.kernel.org/all/[email protected]/

- Updated to have the print processing in kernel/futex/syscall.c

 kernel/futex/syscalls.c | 26 ++++++++++++++++++++++++--
 1 file changed, 24 insertions(+), 2 deletions(-)
kernel/futex/syscalls.c +24 -2
index a46706d6bc6c..24d912d9b20d 100644
@@ -211,6 +211,9 @@ void futex_print_syscall(struct seq_buf *s, int nr_args, unsigned long *args,
 	unsigned int op, cmd;
 	bool done = false;
 
+	op = args[1];
+	cmd = op & FUTEX_CMD_MASK;
+
 	for (int i = 0; !done && i < nr_args; i++) {
 
 		if (seq_buf_has_overflowed(s))

@@ -225,11 +228,30 @@ void futex_print_syscall(struct seq_buf *s, int nr_args, unsigned long *args,
 					seq_buf_printf(s, " (%u)", val);
 				else
 					seq_buf_printf(s, " (0x%x)", val);
+
+				switch(cmd) {
+				case FUTEX_LOCK_PI:
+				case FUTEX_UNLOCK_PI:
+					seq_buf_printf(s, " tid: %d",
+						       val & FUTEX_TID_MASK);
+
+					if (!(val & (FUTEX_OWNER_DIED|FUTEX_WAITERS)))
+						break;
+
+					seq_buf_puts(s, " (");
+					if (val & FUTEX_WAITERS)
+						seq_buf_puts(s, "WAITERS");
+					if (val & FUTEX_OWNER_DIED) {
+						if (op & FUTEX_WAITERS)
+							seq_buf_putc(s, '|');
+						seq_buf_puts(s, "DIED");
+					}
+					seq_buf_putc(s, ')');
+					break;
+				}
 			}
 			continue;
 		case 1:
-			op = args[i];
-			cmd = op & FUTEX_CMD_MASK;
 			if (cmd <= FUTEX_LOCK_PI2)
 				seq_buf_printf(s, ", %s", __futex_cmds[cmd]);
 			else
Signature · 1 line
2.51.0
maintainer Reviewed-by on v2 3/3
#9 lore raw

Re: [PATCH v2 3/3] tracing: Show TID and flags for PI futex system call trace event

Quoted 6 lines · On Tue, 10 Mar 2026 16:09:57 -0400 Steven Rostedt <[email protected]> wrote:

From: Steven Rostedt <[email protected]>

For the futex system call trace event for FUTEX_LOCK_PI and FUTEX_UNLOCK_PI commands, show the TID from the value (which is usually in hex) as well as translate the flags (DIED and WAITERS).

Quoted 7 lines
 pi_mutex_hammer-1098    [000] .....   121.876928: sys_futex(uaddr: 0x560f40cc8180 (0x450) tid: 1104, FUTEX_LOCK_PI|FUTEX_PRIVATE_FLAG, val: 0, timespec: 0x7f2f9d4b1e50 (0.000100000))
 pi_mutex_hammer-1128    [000] .....   121.877120: sys_futex(uaddr: 0x560f40cc8180 (0x8000042a) tid: 1066 (WAITERS), FUTEX_LOCK_PI|FUTEX_PRIVATE_FLAG, val: 0, timespec: 0x7f2f8e493e50 (0.000100000))
 pi_mutex_hammer-1106    [000] .....   121.877242: sys_futex(uaddr: 0x560f40cc8180 (0x80000452) tid: 1106 (WAITERS), FUTEX_UNLOCK_PI|FUTEX_PRIVATE_FLAG, val: 0)

This makes it easier to see the hand off of a mutex and who the owner was.

Signed-off-by Steven Rostedt (Google) <[email protected]>

Looks good to me.

Reviewed-by Masami Hiramatsu (Google) <[email protected]>

Thanks!

Quoted 59 lines

---
Changes since v1: https://lore.kernel.org/all/[email protected]/

- Updated to have the print processing in kernel/futex/syscall.c

 kernel/futex/syscalls.c | 26 ++++++++++++++++++++++++--
 1 file changed, 24 insertions(+), 2 deletions(-)
kernel/futex/syscalls.c +24 -2
index a46706d6bc6c..24d912d9b20d 100644
@@ -211,6 +211,9 @@ void futex_print_syscall(struct seq_buf *s, int nr_args, unsigned long *args,
 	unsigned int op, cmd;
 	bool done = false;
 
+	op = args[1];
+	cmd = op & FUTEX_CMD_MASK;
+
 	for (int i = 0; !done && i < nr_args; i++) {
 
 		if (seq_buf_has_overflowed(s))

@@ -225,11 +228,30 @@ void futex_print_syscall(struct seq_buf *s, int nr_args, unsigned long *args,
 					seq_buf_printf(s, " (%u)", val);
 				else
 					seq_buf_printf(s, " (0x%x)", val);
+
+				switch(cmd) {
+				case FUTEX_LOCK_PI:
+				case FUTEX_UNLOCK_PI:
+					seq_buf_printf(s, " tid: %d",
+						       val & FUTEX_TID_MASK);
+
+					if (!(val & (FUTEX_OWNER_DIED|FUTEX_WAITERS)))
+						break;
+
+					seq_buf_puts(s, " (");
+					if (val & FUTEX_WAITERS)
+						seq_buf_puts(s, "WAITERS");
+					if (val & FUTEX_OWNER_DIED) {
+						if (op & FUTEX_WAITERS)
+							seq_buf_putc(s, '|');
+						seq_buf_puts(s, "DIED");
+					}
+					seq_buf_putc(s, ')');
+					break;
+				}
 			}
 			continue;
 		case 1:
-			op = args[i];
-			cmd = op & FUTEX_CMD_MASK;
 			if (cmd <= FUTEX_LOCK_PI2)
 				seq_buf_printf(s, ", %s", __futex_cmds[cmd]);
 			else
Signature · 1 line
2.51.0
Signature · 1 line
Masami Hiramatsu (Google) <[email protected]>