offcputime result is empty when I try to attach it to a specific process

hi，when I run offcputime with specifying pid, its output is empty. But it works without specifying pid.
This problem is very similar to https://github.com/iovisor/bcc/issues/1566. But our environment is different and the solution in [issues-1566](https://github.com/iovisor/bcc/issues/1566#issuecomment-362123531) is useless here. Thanks.

my environment:
$uname -r
4.19.95-27
$cat /etc/redhat-release
CentOS Linux release 7.9.2009 (Core)

PS, the output with --ebpf is :

#include
#include

#define MINBLOCK_US 1ULL
#define MAXBLOCK_US 18446744073709551615ULL

struct key_t {
u32 pid;
u32 tgid;
int user_stack_id;
int kernel_stack_id;
char name[TASK_COMM_LEN];
};
BPF_HASH(counts, struct key_t);
BPF_HASH(start, u32);
BPF_STACK_TRACE(stack_traces, 1024);

struct rq;

int oncpu(struct pt_regs *ctx, struct rq *rq, struct task_struct *prev) {
u32 pid = prev->pid;
u32 tgid = prev->tgid;
u64 ts, *tsp;

// record previous thread sleep time
if ((tgid == 28) && (1)) {
ts = bpf_ktime_get_ns();
start.update(&pid, &ts);
}

// get the current thread's start time
pid = bpf_get_current_pid_tgid();
tgid = bpf_get_current_pid_tgid() >> 32;
tsp = start.lookup(&pid);
if (tsp == 0) {
return 0; // missed start or filtered
}

// calculate current thread's delta time
u64 delta = bpf_ktime_get_ns() - *tsp;
start.delete(&pid);
delta = delta / 1000;
if ((delta < MINBLOCK_US) || (delta > MAXBLOCK_US)) {
return 0;
}

// create map key
struct key_t key = {};

key.pid = pid;
key.tgid = tgid;
key.user_stack_id = stack_traces.get_stackid(ctx, BPF_F_USER_STACK);
key.kernel_stack_id = stack_traces.get_stackid(ctx, 0);
bpf_get_current_comm(&key.name, sizeof(key.name));

counts.increment(key, delta);
return 0;
}