On Thu, 29 Feb 2024 16:13:20 +0900 Masami Hiramatsu (Google) mhiramat@kernel.org wrote:
On Thu, 29 Feb 2024 15:38:55 +0900 Masami Hiramatsu (Google) mhiramat@kernel.org wrote:
Hmm, this seems arch_rethook_trampoline caused the issue.
And curiously, it depends on the number of stored data.
OK: /sys/kernel/tracing # echo 'f vfs_read%return $arg1 $arg2 $arg3' >> dynamic_events /sys/kernel/tracing # echo 1 > events/fprobes/enable
NG: /sys/kernel/tracing # echo 'f vfs_read%return $arg1 $arg2 $arg3 $arg4' >> dynamic_events /sys/kernel/tracing # echo 1 > events/fprobes/enable
I also confirmed that on 'vfs_write' caused the same result. 3 arguments(24 bytes) is OK, but 4 arguments (32bytes) is NG.
And this may be the fprobe bug. kretprobe events doesn't show this issue.
OK: /sys/kernel/tracing # echo 'r vfs_read $arg*' >> kprobe_events /sys/kernel/tracing # echo 1 > events/kprobes/enable
But this is strange because both uses same rethook...
Lol, I haven't allocate the entry data size when initialize rethook. That's a bug. Please try below.
diff --git a/kernel/trace/fprobe.c b/kernel/trace/fprobe.c index 6cd2a4e3afb8..9ff018245840 100644 --- a/kernel/trace/fprobe.c +++ b/kernel/trace/fprobe.c @@ -189,9 +189,6 @@ static int fprobe_init_rethook(struct fprobe *fp, int num) { int size;
- if (num <= 0) - return -EINVAL; - if (!fp->exit_handler) { fp->rethook = NULL; return 0; @@ -199,15 +196,16 @@ static int fprobe_init_rethook(struct fprobe *fp, int num)
/* Initialize rethook if needed */ if (fp->nr_maxactive) - size = fp->nr_maxactive; + num = fp->nr_maxactive; else - size = num * num_possible_cpus() * 2; - if (size <= 0) + num *= num_possible_cpus() * 2; + if (num <= 0) return -EINVAL;
+ size = sizeof(struct fprobe_rethook_node) + fp->entry_data_size; + /* Initialize rethook */ - fp->rethook = rethook_alloc((void *)fp, fprobe_exit_handler, - sizeof(struct fprobe_rethook_node), size); + fp->rethook = rethook_alloc((void *)fp, fprobe_exit_handler, size, num); if (IS_ERR(fp->rethook)) return PTR_ERR(fp->rethook);