The current get_wchan() implementations do their best to avoid problems
when walking a stack given a process in an unknown state, but this is
fragile and unnecessary. It's only useful to report wchan if a process
is actually blocked, so use the new sched_task_get_wchan() instead.
Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
Cc: Christian Brauner <christian.brauner@xxxxxxxxxx>
Cc: Jann Horn <jannh@xxxxxxxxxx>
Cc: Michal Hocko <mhocko@xxxxxxxx>
Cc: Helge Deller <deller@xxxxxx>
Cc: linux-fsdevel@xxxxxxxxxxxxxxx
Signed-off-by: Kees Cook <keescook@xxxxxxxxxxxx>
---
fs/proc/base.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/fs/proc/base.c b/fs/proc/base.c
index 1f394095eb88..7853592778b2 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -392,7 +392,8 @@ static int proc_pid_wchan(struct seq_file *m, struct pid_namespace *ns,
if (!ptrace_may_access(task, PTRACE_MODE_READ_FSCREDS))
goto print0;
- wchan = get_wchan(task);
+ wchan = sched_task_get_wchan(task);
+ /* Must only report symbolized addresses and never raw pointers. */
if (wchan && !lookup_symbol_name(wchan, symname)) {
seq_puts(m, symname);
return 0;
--
2.30.2
