Merge branch 'implement-mechanism-to-signal-other-threads'

Puranjay Mohan says:

====================
Implement mechanism to signal other threads

This set implements a kfunc called bpf_send_signal_task() that is similar
to sigqueue() as it can send a signal along with a cookie to a thread or
thread group.

The send_signal selftest has been updated to also test this new kfunc under
all contexts.

Changes in v5:
v4: https://lore.kernel.org/all/20241008114940.44305-1-puranjay@kernel.org/
- Call copy_siginfo() only if work->has_siginfo is true in
  bpf_send_signal_common()
- Add Acked-by: Andrii Nakryiko <andrii@kernel.org>

Changes in v4:
v3: https://lore.kernel.org/all/20241007103426.128923-1-puranjay@kernel.org/
- Fix the selftest to make it work for big-endian archs.
- Fix a build warning on 32-bit archs.
- Some style changes and code refactors suggested by Andrii

Changes in v3:
v2: https://lore.kernel.org/all/20240926115328.105634-1-puranjay@kernel.org/
- make the cookie u64 instead of int.
- re use code from bpf_send_signal_common

Changes in v2:
v1: https://lore.kernel.org/bpf/20240724113944.75977-1-puranjay@kernel.org/
- Convert to a kfunc
- Add mechanism to send a cookie with the signal.
====================

Link: https://lore.kernel.org/r/20241016084136.10305-1-puranjay@kernel.org
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>

kernel/bpf/helpers.c

@@ -3055,6 +3055,7 @@ BTF_ID_FLAGS(func, bpf_task_get_cgroup1, KF_ACQUIRE | KF_RCU | KF_RET_NULL)
 BTF_ID_FLAGS(func, bpf_task_from_pid, KF_ACQUIRE | KF_RET_NULL)
 BTF_ID_FLAGS(func, bpf_task_from_vpid, KF_ACQUIRE | KF_RET_NULL)
 BTF_ID_FLAGS(func, bpf_throw)
+BTF_ID_FLAGS(func, bpf_send_signal_task, KF_TRUSTED_ARGS)
 BTF_KFUNCS_END(generic_btf_ids)
 
 static const struct btf_kfunc_id_set generic_kfunc_set = {

kernel/trace/bpf_trace.c

@@ -802,6 +802,8 @@ struct send_signal_irq_work {
 	struct task_struct *task;
 	u32 sig;
 	enum pid_type type;
+	bool has_siginfo;
+	struct kernel_siginfo info;
 };
 
 static DEFINE_PER_CPU(struct send_signal_irq_work, send_signal_work);
@@ -809,27 +811,46 @@ static DEFINE_PER_CPU(struct send_signal_irq_work, send_signal_work);
 static void do_bpf_send_signal(struct irq_work *entry)
 {
 	struct send_signal_irq_work *work;
+	struct kernel_siginfo *siginfo;
 
 	work = container_of(entry, struct send_signal_irq_work, irq_work);
-	group_send_sig_info(work->sig, SEND_SIG_PRIV, work->task, work->type);
+	siginfo = work->has_siginfo ? &work->info : SEND_SIG_PRIV;
+
+	group_send_sig_info(work->sig, siginfo, work->task, work->type);
 	put_task_struct(work->task);
 }
 
-static int bpf_send_signal_common(u32 sig, enum pid_type type)
+static int bpf_send_signal_common(u32 sig, enum pid_type type, struct task_struct *task, u64 value)
 {
 	struct send_signal_irq_work *work = NULL;
+	struct kernel_siginfo info;
+	struct kernel_siginfo *siginfo;
+
+	if (!task) {
+		task = current;
+		siginfo = SEND_SIG_PRIV;
+	} else {
+		clear_siginfo(&info);
+		info.si_signo = sig;
+		info.si_errno = 0;
+		info.si_code = SI_KERNEL;
+		info.si_pid = 0;
+		info.si_uid = 0;
+		info.si_value.sival_ptr = (void *)(unsigned long)value;
+		siginfo = &info;
+	}
 
 	/* Similar to bpf_probe_write_user, task needs to be
 	 * in a sound condition and kernel memory access be
 	 * permitted in order to send signal to the current
 	 * task.
 	 */
-	if (unlikely(current->flags & (PF_KTHREAD | PF_EXITING)))
+	if (unlikely(task->flags & (PF_KTHREAD | PF_EXITING)))
 		return -EPERM;
 	if (unlikely(!nmi_uaccess_okay()))
 		return -EPERM;
 	/* Task should not be pid=1 to avoid kernel panic. */
-	if (unlikely(is_global_init(current)))
+	if (unlikely(is_global_init(task)))
 		return -EPERM;
 
 	if (irqs_disabled()) {
@@ -847,19 +868,22 @@ static int bpf_send_signal_common(u32 sig, enum pid_type type)
 		 * to the irq_work. The current task may change when queued
 		 * irq works get executed.
 		 */
-		work->task = get_task_struct(current);
+		work->task = get_task_struct(task);
+		work->has_siginfo = siginfo == &info;
+		if (work->has_siginfo)
+			copy_siginfo(&work->info, &info);
 		work->sig = sig;
 		work->type = type;
 		irq_work_queue(&work->irq_work);
 		return 0;
 	}
 
-	return group_send_sig_info(sig, SEND_SIG_PRIV, current, type);
+	return group_send_sig_info(sig, siginfo, task, type);
 }
 
 BPF_CALL_1(bpf_send_signal, u32, sig)
 {
-	return bpf_send_signal_common(sig, PIDTYPE_TGID);
+	return bpf_send_signal_common(sig, PIDTYPE_TGID, NULL, 0);
 }
 
 static const struct bpf_func_proto bpf_send_signal_proto = {
@@ -871,7 +895,7 @@ static const struct bpf_func_proto bpf_send_signal_proto = {
 
 BPF_CALL_1(bpf_send_signal_thread, u32, sig)
 {
-	return bpf_send_signal_common(sig, PIDTYPE_PID);
+	return bpf_send_signal_common(sig, PIDTYPE_PID, NULL, 0);
 }
 
 static const struct bpf_func_proto bpf_send_signal_thread_proto = {
@@ -3484,3 +3508,16 @@ static int __init bpf_kprobe_multi_kfuncs_init(void)
 }
 
 late_initcall(bpf_kprobe_multi_kfuncs_init);
+
+__bpf_kfunc_start_defs();
+
+__bpf_kfunc int bpf_send_signal_task(struct task_struct *task, int sig, enum pid_type type,
+				     u64 value)
+{
+	if (type != PIDTYPE_PID && type != PIDTYPE_TGID)
+		return -EINVAL;
+
+	return bpf_send_signal_common(sig, type, task, value);
+}
+
+__bpf_kfunc_end_defs();

tools/testing/selftests/bpf/prog_tests/send_signal.c

@@ -8,17 +8,25 @@ static int sigusr1_received;
 
 static void sigusr1_handler(int signum)
 {
-	sigusr1_received = 1;
+	sigusr1_received = 8;
+}
+
+static void sigusr1_siginfo_handler(int s, siginfo_t *i, void *v)
+{
+	sigusr1_received = (int)(long long)i->si_value.sival_ptr;
 }
 
 static void test_send_signal_common(struct perf_event_attr *attr,
-				    bool signal_thread)
+				    bool signal_thread, bool remote)
 {
 	struct test_send_signal_kern *skel;
+	struct sigaction sa;
 	int pipe_c2p[2], pipe_p2c[2];
 	int err = -1, pmu_fd = -1;
+	volatile int j = 0;
 	char buf[256];
 	pid_t pid;
+	int old_prio;
 
 	if (!ASSERT_OK(pipe(pipe_c2p), "pipe_c2p"))
 		return;
@@ -39,11 +47,14 @@ static void test_send_signal_common(struct perf_event_attr *attr,
 	}
 
 	if (pid == 0) {
-		int old_prio;
-		volatile int j = 0;
-
 		/* install signal handler and notify parent */
-		ASSERT_NEQ(signal(SIGUSR1, sigusr1_handler), SIG_ERR, "signal");
+		if (remote) {
+			sa.sa_sigaction = sigusr1_siginfo_handler;
+			sa.sa_flags = SA_RESTART | SA_SIGINFO;
+			ASSERT_NEQ(sigaction(SIGUSR1, &sa, NULL), -1, "sigaction");
+		} else {
+			ASSERT_NEQ(signal(SIGUSR1, sigusr1_handler), SIG_ERR, "signal");
+		}
 
 		close(pipe_c2p[0]); /* close read */
 		close(pipe_p2c[1]); /* close write */
@@ -52,10 +63,12 @@ static void test_send_signal_common(struct perf_event_attr *attr,
 		 * that if an interrupt happens, the underlying task
 		 * is this process.
 		 */
-		errno = 0;
-		old_prio = getpriority(PRIO_PROCESS, 0);
-		ASSERT_OK(errno, "getpriority");
-		ASSERT_OK(setpriority(PRIO_PROCESS, 0, -20), "setpriority");
+		if (!remote) {
+			errno = 0;
+			old_prio = getpriority(PRIO_PROCESS, 0);
+			ASSERT_OK(errno, "getpriority");
+			ASSERT_OK(setpriority(PRIO_PROCESS, 0, -20), "setpriority");
+		}
 
 		/* notify parent signal handler is installed */
 		ASSERT_EQ(write(pipe_c2p[1], buf, 1), 1, "pipe_write");
@@ -66,20 +79,25 @@ static void test_send_signal_common(struct perf_event_attr *attr,
 		/* wait a little for signal handler */
 		for (int i = 0; i < 1000000000 && !sigusr1_received; i++) {
 			j /= i + j + 1;
-			if (!attr)
-				/* trigger the nanosleep tracepoint program. */
-				usleep(1);
+			if (remote)
+				sleep(1);
+			else
+				if (!attr)
+					/* trigger the nanosleep tracepoint program. */
+					usleep(1);
 		}
 
-		buf[0] = sigusr1_received ? '2' : '0';
-		ASSERT_EQ(sigusr1_received, 1, "sigusr1_received");
+		buf[0] = sigusr1_received;
+
+		ASSERT_EQ(sigusr1_received, 8, "sigusr1_received");
 		ASSERT_EQ(write(pipe_c2p[1], buf, 1), 1, "pipe_write");
 
 		/* wait for parent notification and exit */
 		ASSERT_EQ(read(pipe_p2c[0], buf, 1), 1, "pipe_read");
 
 		/* restore the old priority */
-		ASSERT_OK(setpriority(PRIO_PROCESS, 0, old_prio), "setpriority");
+		if (!remote)
+			ASSERT_OK(setpriority(PRIO_PROCESS, 0, old_prio), "setpriority");
 
 		close(pipe_c2p[1]);
 		close(pipe_p2c[0]);
@@ -93,6 +111,17 @@ static void test_send_signal_common(struct perf_event_attr *attr,
 	if (!ASSERT_OK_PTR(skel, "skel_open_and_load"))
 		goto skel_open_load_failure;
 
+	/* boost with a high priority so we got a higher chance
+	 * that if an interrupt happens, the underlying task
+	 * is this process.
+	 */
+	if (remote) {
+		errno = 0;
+		old_prio = getpriority(PRIO_PROCESS, 0);
+		ASSERT_OK(errno, "getpriority");
+		ASSERT_OK(setpriority(PRIO_PROCESS, 0, -20), "setpriority");
+	}
+
 	if (!attr) {
 		err = test_send_signal_kern__attach(skel);
 		if (!ASSERT_OK(err, "skel_attach")) {
@@ -100,8 +129,12 @@ static void test_send_signal_common(struct perf_event_attr *attr,
 			goto destroy_skel;
 		}
 	} else {
-		pmu_fd = syscall(__NR_perf_event_open, attr, pid, -1 /* cpu */,
-				 -1 /* group id */, 0 /* flags */);
+		if (!remote)
+			pmu_fd = syscall(__NR_perf_event_open, attr, pid, -1 /* cpu */,
+					 -1 /* group id */, 0 /* flags */);
+		else
+			pmu_fd = syscall(__NR_perf_event_open, attr, getpid(), -1 /* cpu */,
+					 -1 /* group id */, 0 /* flags */);
 		if (!ASSERT_GE(pmu_fd, 0, "perf_event_open")) {
 			err = -1;
 			goto destroy_skel;
@@ -119,11 +152,30 @@ static void test_send_signal_common(struct perf_event_attr *attr,
 	/* trigger the bpf send_signal */
 	skel->bss->signal_thread = signal_thread;
 	skel->bss->sig = SIGUSR1;
-	skel->bss->pid = pid;
+	if (!remote) {
+		skel->bss->target_pid = 0;
+		skel->bss->pid = pid;
+	} else {
+		skel->bss->target_pid = pid;
+		skel->bss->pid = getpid();
+	}
 
 	/* notify child that bpf program can send_signal now */
 	ASSERT_EQ(write(pipe_p2c[1], buf, 1), 1, "pipe_write");
 
+	/* For the remote test, the BPF program is triggered from this
+	 * process but the other process/thread is signaled.
+	 */
+	if (remote) {
+		if (!attr) {
+			for (int i = 0; i < 10; i++)
+				usleep(1);
+		} else {
+			for (int i = 0; i < 100000000; i++)
+				j /= i + 1;
+		}
+	}
+
 	/* wait for result */
 	err = read(pipe_c2p[0], buf, 1);
 	if (!ASSERT_GE(err, 0, "reading pipe"))
@@ -133,7 +185,7 @@ static void test_send_signal_common(struct perf_event_attr *attr,
 		goto disable_pmu;
 	}
 
-	ASSERT_EQ(buf[0], '2', "incorrect result");
+	ASSERT_EQ(buf[0], 8, "incorrect result");
 
 	/* notify child safe to exit */
 	ASSERT_EQ(write(pipe_p2c[1], buf, 1), 1, "pipe_write");
@@ -142,18 +194,21 @@ disable_pmu:
 	close(pmu_fd);
 destroy_skel:
 	test_send_signal_kern__destroy(skel);
+	/* restore the old priority */
+	if (remote)
+		ASSERT_OK(setpriority(PRIO_PROCESS, 0, old_prio), "setpriority");
 skel_open_load_failure:
 	close(pipe_c2p[0]);
 	close(pipe_p2c[1]);
 	wait(NULL);
 }
 
-static void test_send_signal_tracepoint(bool signal_thread)
+static void test_send_signal_tracepoint(bool signal_thread, bool remote)
 {
-	test_send_signal_common(NULL, signal_thread);
+	test_send_signal_common(NULL, signal_thread, remote);
 }
 
-static void test_send_signal_perf(bool signal_thread)
+static void test_send_signal_perf(bool signal_thread, bool remote)
 {
 	struct perf_event_attr attr = {
 		.freq = 1,
@@ -162,10 +217,10 @@ static void test_send_signal_perf(bool signal_thread)
 		.config = PERF_COUNT_SW_CPU_CLOCK,
 	};
 
-	test_send_signal_common(&attr, signal_thread);
+	test_send_signal_common(&attr, signal_thread, remote);
 }
 
-static void test_send_signal_nmi(bool signal_thread)
+static void test_send_signal_nmi(bool signal_thread, bool remote)
 {
 	struct perf_event_attr attr = {
 		.sample_period = 1,
@@ -191,21 +246,35 @@ static void test_send_signal_nmi(bool signal_thread)
 		close(pmu_fd);
 	}
 
-	test_send_signal_common(&attr, signal_thread);
+	test_send_signal_common(&attr, signal_thread, remote);
 }
 
 void test_send_signal(void)
 {
 	if (test__start_subtest("send_signal_tracepoint"))
-		test_send_signal_tracepoint(false);
+		test_send_signal_tracepoint(false, false);
 	if (test__start_subtest("send_signal_perf"))
-		test_send_signal_perf(false);
+		test_send_signal_perf(false, false);
 	if (test__start_subtest("send_signal_nmi"))
-		test_send_signal_nmi(false);
+		test_send_signal_nmi(false, false);
 	if (test__start_subtest("send_signal_tracepoint_thread"))
-		test_send_signal_tracepoint(true);
+		test_send_signal_tracepoint(true, false);
 	if (test__start_subtest("send_signal_perf_thread"))
-		test_send_signal_perf(true);
+		test_send_signal_perf(true, false);
 	if (test__start_subtest("send_signal_nmi_thread"))
-		test_send_signal_nmi(true);
+		test_send_signal_nmi(true, false);
+
+	/* Signal remote thread and thread group */
+	if (test__start_subtest("send_signal_tracepoint_remote"))
+		test_send_signal_tracepoint(false, true);
+	if (test__start_subtest("send_signal_perf_remote"))
+		test_send_signal_perf(false, true);
+	if (test__start_subtest("send_signal_nmi_remote"))
+		test_send_signal_nmi(false, true);
+	if (test__start_subtest("send_signal_tracepoint_thread_remote"))
+		test_send_signal_tracepoint(true, true);
+	if (test__start_subtest("send_signal_perf_thread_remote"))
+		test_send_signal_perf(true, true);
+	if (test__start_subtest("send_signal_nmi_thread_remote"))
+		test_send_signal_nmi(true, true);
 }

tools/testing/selftests/bpf/progs/test_send_signal_kern.c

@@ -1,27 +1,50 @@
 // SPDX-License-Identifier: GPL-2.0
 // Copyright (c) 2019 Facebook
-#include <linux/bpf.h>
+#include <vmlinux.h>
 #include <linux/version.h>
 #include <bpf/bpf_helpers.h>
 
-__u32 sig = 0, pid = 0, status = 0, signal_thread = 0;
+struct task_struct *bpf_task_from_pid(int pid) __ksym;
+void bpf_task_release(struct task_struct *p) __ksym;
+int bpf_send_signal_task(struct task_struct *task, int sig, enum pid_type type, u64 value) __ksym;
+
+__u32 sig = 0, pid = 0, status = 0, signal_thread = 0, target_pid = 0;
 
 static __always_inline int bpf_send_signal_test(void *ctx)
 {
+	struct task_struct *target_task = NULL;
 	int ret;
+	u64 value;
 
 	if (status != 0 || pid == 0)
 		return 0;
 
 	if ((bpf_get_current_pid_tgid() >> 32) == pid) {
-		if (signal_thread)
-			ret = bpf_send_signal_thread(sig);
-		else
-			ret = bpf_send_signal(sig);
+		if (target_pid) {
+			target_task = bpf_task_from_pid(target_pid);
+			if (!target_task)
+				return 0;
+			value = 8;
+		}
+
+		if (signal_thread) {
+			if (target_pid)
+				ret = bpf_send_signal_task(target_task, sig, PIDTYPE_PID, value);
+			else
+				ret = bpf_send_signal_thread(sig);
+		} else {
+			if (target_pid)
+				ret = bpf_send_signal_task(target_task, sig, PIDTYPE_TGID, value);
+			else
+				ret = bpf_send_signal(sig);
+		}
 		if (ret == 0)
 			status = 1;
 	}
 
+	if (target_task)
+		bpf_task_release(target_task);
+
 	return 0;
 }