Skip to contentSkip to conversation

Fix wrong type cast

PATCH In review Nam Cao 2 participants 6 messages Jul 27, 2025 – Jul 28, 2025

Subsystems (2)
lore
Share & tools
lore

Nam Cao

author
#0 lore raw

cover Fix wrong type cast

Hi,

After my previous rv simplification series:
https://lore.kernel.org/linux-trace-kernel/[email protected]/T/#t

Some RV's available/enabled reactors/monitors files do not work anymore:

    root@yellow:~# cat /sys/kernel/tracing/rv/available_monitors
    nop: �������`������� ���������Z������u�����-s�����
    nop:���������������� �������@>H�����

nop:
��������
��������8�������:��������
`���������������

root@yellow:~# cat /sys/kernel/tracing/rv/available_reactors ��������8�������
P�������x������� 8����������������^�����81s������I?�������������8�������

It turned out that there has been a type cast mistake in RV all this time, but it still happened to work. Due to the recent refactor, the mistaken type cast got "promoted" and became a real problem.

This series fixes up this type cast.

Nam Cao (2):
rv: Fix wrong type cast in monitors_show()
rv: Fix wrong type cast in reactors_show() and monitor_reactor_show()

 kernel/trace/rv/rv.c          | 2 +-
 kernel/trace/rv/rv_reactors.c | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)
Signature · 1 line
2.39.5
author
#1 lore raw

PATCH 1/2 Fix wrong type cast in monitors_show()

Argument 'p' of monitors_show() is not a pointer to struct rv_monitor, it is actually a pointer to the list_head inside struct rv_monitor. Therefore it is wrong to cast 'p' to struct rv_monitor *.

This wrong type cast has been there since the beginning. But it still worked because the list_head was the first field in struct rv_monitor_def. This is no longer true since commit 24cbfe18d55a ("rv: Merge struct rv_monitor_def into struct rv_monitor") moved the list_head, and this wrong type cast became a functional problem.

Properly use container_of() instead.

Fixes 24cbfe18d55a ("rv: Merge struct rv_monitor_def into struct rv_monitor")
---
 kernel/trace/rv/rv.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/trace/rv/rv.c +1 -1
index 2ac0e8bf4e60..1482e91c39f4 100644
@@ -458,7 +458,7 @@ static int create_monitor_dir(struct rv_monitor *mon, struct rv_monitor *parent)
  */
 static int monitors_show(struct seq_file *m, void *p)
 {
-	struct rv_monitor *mon = p;
+	struct rv_monitor *mon = container_of(p, struct rv_monitor, list);
 
 	if (mon->parent)
 		seq_printf(m, "%s:%s\n", mon->parent->name, mon->name);
Signature · 1 line
2.39.5
author
#5 lore raw

PATCH 2/2 Fix wrong type cast in reactors_show() and monitor_reactor_show()

In reply to Nam Cao

Argument 'p' of reactors_show() and monitor_reactor_show() is not a pointer to struct rv_reactor, it is actually a pointer to the list_head inside struct rv_reactor. Therefore it's wrong to cast 'p' to struct rv_reactor *.

This wrong type cast has been there since the beginning. But it still worked because the list_head was the first field in struct rv_reactor_def. This is no longer true since commit 3d3c376118b5 ("rv: Merge struct rv_reactor_def into struct rv_reactor") moved the list_head, and this wrong type cast became a functional problem.

Properly use container_of() instead.

Fixes 3d3c376118b5 ("rv: Merge struct rv_reactor_def into struct rv_reactor")
---
 kernel/trace/rv/rv_reactors.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
kernel/trace/rv/rv_reactors.c +2 -2
index 106f2c4740f2..d32859fec238 100644
@@ -86,7 +86,7 @@ static struct rv_reactor *get_reactor_rdef_by_name(char *name)
  */
 static int reactors_show(struct seq_file *m, void *p)
 {
-	struct rv_reactor *reactor = p;
+	struct rv_reactor *reactor = container_of(p, struct rv_reactor, list);
 
 	seq_printf(m, "%s\n", reactor->name);
 	return 0;
@@ -139,7 +139,7 @@ static const struct file_operations available_reactors_ops = {
 static int monitor_reactor_show(struct seq_file *m, void *p)
 {
 	struct rv_monitor *mon = m->private;
-	struct rv_reactor *reactor = p;
+	struct rv_reactor *reactor = container_of(p, struct rv_reactor, list);
 
 	if (mon->reactor == reactor)
 		seq_printf(m, "[%s]\n", reactor->name);
Signature · 1 line
2.39.5
maintainer
#2 lore raw

Re: [PATCH 1/2] rv: Fix wrong type cast in monitors_show()

In reply to Nam Cao
Quoted 26 lines · Nam Cao

Argument 'p' of monitors_show() is not a pointer to struct rv_monitor, it is actually a pointer to the list_head inside struct rv_monitor. Therefore it is wrong to cast 'p' to struct rv_monitor *.

This wrong type cast has been there since the beginning. But it still worked because the list_head was the first field in struct rv_monitor_def. This is no longer true since commit 24cbfe18d55a ("rv: Merge struct rv_monitor_def into struct rv_monitor") moved the list_head, and this wrong type cast became a functional problem.

Properly use container_of() instead.

Fixes 24cbfe18d55a ("rv: Merge struct rv_monitor_def into struct

rv_monitor")

---
 kernel/trace/rv/rv.c | 2 +-

kernel/trace/rv/rv.c +1 -1
@@ -458,7 +458,7 @@ static int create_monitor_dir(struct rv_monitor
 *mon, struct rv_monitor *parent)
   */
  static int monitors_show(struct seq_file *m, void *p)
  {
-	struct rv_monitor *mon = p;
+	struct rv_monitor *mon = container_of(p, struct rv_monitor,
 list);
Quoted 1 line

Good catch, thanks! The container_of is the way to go.
Do you have valid reasons not to move the list_head to the top? It's not a big deal but it would save computing and summing the offset. It doesn't seem name (the current first element) really needs to stay there.

Thanks,
Gabriele

Quoted 3 lines

  if (mon->parent)
  seq_printf(m, "%s:%s\n", mon->parent->name, mon-

Quoted 1 line

name);

author
#3 lore raw

Re: [PATCH 1/2] rv: Fix wrong type cast in monitors_show()

Quoted 5 lines · Gabriele Monaco

Good catch, thanks! The container_of is the way to go.
Do you have valid reasons not to move the list_head to the top? It's not a big deal but it would save computing and summing the offset. It doesn't seem name (the current first element) really needs to stay there.

I checked x86_64 and riscv64, the generated assembly of this function before & after moving the list_head on top is almost the same except for some instructions' intermediate values. Both architectures have instructions which load data at (pointer + offset), so this offset computing does not require any extra instruction.

Best regards,
Nam

maintainer Reviewed-by
#4 lore raw

Re: [PATCH 1/2] rv: Fix wrong type cast in monitors_show()

Quoted 16 lines · Nam Cao
Quoted 7 lines · Gabriele Monaco

Good catch, thanks! The container_of is the way to go. Do you have valid reasons not to move the list_head to the top? It's
not a big deal but it would save computing and summing the offset. It
doesn't seem name (the current first element) really needs to stay there.

I checked x86_64 and riscv64, the generated assembly of this function before & after moving the list_head on top is almost the same except for
some instructions' intermediate values. Both architectures have instructions which load data at (pointer + offset), so this offset computing does not require any extra instruction.

Quoted 2 lines

Best regards,
Nam

Alright then, thanks for checking!