1 files changed, 792 insertions, 201 deletions

diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
index 6b264a92064b..bf0906e1e2b9 100644
--- a/kernel/bpf/trampoline.c
+++ b/kernel/bpf/trampoline.c
@@ -5,6 +5,15 @@
 #include <linux/filter.h>
 #include <linux/ftrace.h>
 #include <linux/rbtree_latch.h>
+#include <linux/perf_event.h>
+#include <linux/btf.h>
+#include <linux/rcupdate_trace.h>
+#include <linux/rcupdate_wait.h>
+#include <linux/module.h>
+#include <linux/static_call.h>
+#include <linux/bpf_verifier.h>
+#include <linux/bpf_lsm.h>
+#include <linux/delay.h>
 
 /* dummy _ops. The verifier will operate on target program's ops. */
 const struct bpf_verifier_ops bpf_extension_verifier_ops = {
@@ -17,90 +26,116 @@ const struct bpf_prog_ops bpf_extension_prog_ops = {
 #define TRAMPOLINE_TABLE_SIZE (1 << TRAMPOLINE_HASH_BITS)
 
 static struct hlist_head trampoline_table[TRAMPOLINE_TABLE_SIZE];
-static struct latch_tree_root image_tree __cacheline_aligned;
 
-/* serializes access to trampoline_table and image_tree */
+/* serializes access to trampoline_table */
 static DEFINE_MUTEX(trampoline_mutex);
 
-static void *bpf_jit_alloc_exec_page(void)
+#ifdef CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
+static int bpf_trampoline_update(struct bpf_trampoline *tr, bool lock_direct_mutex);
+
+static int bpf_tramp_ftrace_ops_func(struct ftrace_ops *ops, enum ftrace_ops_cmd cmd)
 {
-	void *image;
+	struct bpf_trampoline *tr = ops->private;
+	int ret = 0;
 
-	image = bpf_jit_alloc_exec(PAGE_SIZE);
-	if (!image)
-		return NULL;
+	if (cmd == FTRACE_OPS_CMD_ENABLE_SHARE_IPMODIFY_SELF) {
+		/* This is called inside register_ftrace_direct_multi(), so
+		 * tr->mutex is already locked.
+		 */
+		lockdep_assert_held_once(&tr->mutex);
 
-	set_vm_flush_reset_perms(image);
-	/* Keep image as writeable. The alternative is to keep flipping ro/rw
-	 * everytime new program is attached or detached.
-	 */
-	set_memory_x((long)image, 1);
-	return image;
-}
+		/* Instead of updating the trampoline here, we propagate
+		 * -EAGAIN to register_ftrace_direct_multi(). Then we can
+		 * retry register_ftrace_direct_multi() after updating the
+		 * trampoline.
+		 */
+		if ((tr->flags & BPF_TRAMP_F_CALL_ORIG) &&
+		    !(tr->flags & BPF_TRAMP_F_ORIG_STACK)) {
+			if (WARN_ON_ONCE(tr->flags & BPF_TRAMP_F_SHARE_IPMODIFY))
+				return -EBUSY;
 
-static __always_inline bool image_tree_less(struct latch_tree_node *a,
-				      struct latch_tree_node *b)
-{
-	struct bpf_image *ia = container_of(a, struct bpf_image, tnode);
-	struct bpf_image *ib = container_of(b, struct bpf_image, tnode);
+			tr->flags |= BPF_TRAMP_F_SHARE_IPMODIFY;
+			return -EAGAIN;
+		}
 
-	return ia < ib;
-}
+		return 0;
+	}
 
-static __always_inline int image_tree_comp(void *addr, struct latch_tree_node *n)
-{
-	void *image = container_of(n, struct bpf_image, tnode);
+	/* The normal locking order is
+	 *    tr->mutex => direct_mutex (ftrace.c) => ftrace_lock (ftrace.c)
+	 *
+	 * The following two commands are called from
+	 *
+	 *   prepare_direct_functions_for_ipmodify
+	 *   cleanup_direct_functions_after_ipmodify
+	 *
+	 * In both cases, direct_mutex is already locked. Use
+	 * mutex_trylock(&tr->mutex) to avoid deadlock in race condition
+	 * (something else is making changes to this same trampoline).
+	 */
+	if (!mutex_trylock(&tr->mutex)) {
+		/* sleep 1 ms to make sure whatever holding tr->mutex makes
+		 * some progress.
+		 */
+		msleep(1);
+		return -EAGAIN;
+	}
 
-	if (addr < image)
-		return -1;
-	if (addr >= image + PAGE_SIZE)
-		return 1;
+	switch (cmd) {
+	case FTRACE_OPS_CMD_ENABLE_SHARE_IPMODIFY_PEER:
+		tr->flags |= BPF_TRAMP_F_SHARE_IPMODIFY;
 
-	return 0;
-}
+		if ((tr->flags & BPF_TRAMP_F_CALL_ORIG) &&
+		    !(tr->flags & BPF_TRAMP_F_ORIG_STACK))
+			ret = bpf_trampoline_update(tr, false /* lock_direct_mutex */);
+		break;
+	case FTRACE_OPS_CMD_DISABLE_SHARE_IPMODIFY_PEER:
+		tr->flags &= ~BPF_TRAMP_F_SHARE_IPMODIFY;
 
-static const struct latch_tree_ops image_tree_ops = {
-	.less	= image_tree_less,
-	.comp	= image_tree_comp,
-};
+		if (tr->flags & BPF_TRAMP_F_ORIG_STACK)
+			ret = bpf_trampoline_update(tr, false /* lock_direct_mutex */);
+		break;
+	default:
+		ret = -EINVAL;
+		break;
+	}
 
-static void *__bpf_image_alloc(bool lock)
-{
-	struct bpf_image *image;
+	mutex_unlock(&tr->mutex);
+	return ret;
+}
+#endif
 
-	image = bpf_jit_alloc_exec_page();
-	if (!image)
-		return NULL;
+bool bpf_prog_has_trampoline(const struct bpf_prog *prog)
+{
+	enum bpf_attach_type eatype = prog->expected_attach_type;
+	enum bpf_prog_type ptype = prog->type;
 
-	if (lock)
-		mutex_lock(&trampoline_mutex);
-	latch_tree_insert(&image->tnode, &image_tree, &image_tree_ops);
-	if (lock)
-		mutex_unlock(&trampoline_mutex);
-	return image->data;
+	return (ptype == BPF_PROG_TYPE_TRACING &&
+		(eatype == BPF_TRACE_FENTRY || eatype == BPF_TRACE_FEXIT ||
+		 eatype == BPF_MODIFY_RETURN)) ||
+		(ptype == BPF_PROG_TYPE_LSM && eatype == BPF_LSM_MAC);
 }
 
-void *bpf_image_alloc(void)
+void bpf_image_ksym_add(void *data, struct bpf_ksym *ksym)
 {
-	return __bpf_image_alloc(true);
+	ksym->start = (unsigned long) data;
+	ksym->end = ksym->start + PAGE_SIZE;
+	bpf_ksym_add(ksym);
+	perf_event_ksymbol(PERF_RECORD_KSYMBOL_TYPE_BPF, ksym->start,
+			   PAGE_SIZE, false, ksym->name);
 }
 
-bool is_bpf_image_address(unsigned long addr)
+void bpf_image_ksym_del(struct bpf_ksym *ksym)
 {
-	bool ret;
-
-	rcu_read_lock();
-	ret = latch_tree_find((void *) addr, &image_tree, &image_tree_ops) != NULL;
-	rcu_read_unlock();
-
-	return ret;
+	bpf_ksym_del(ksym);
+	perf_event_ksymbol(PERF_RECORD_KSYMBOL_TYPE_BPF, ksym->start,
+			   PAGE_SIZE, true, ksym->name);
 }
 
-struct bpf_trampoline *bpf_trampoline_lookup(u64 key)
+static struct bpf_trampoline *bpf_trampoline_lookup(u64 key)
 {
 	struct bpf_trampoline *tr;
 	struct hlist_head *head;
-	void *image;
 	int i;
 
 	mutex_lock(&trampoline_mutex);
@@ -114,14 +149,16 @@ struct bpf_trampoline *bpf_trampoline_lookup(u64 key)
 	tr = kzalloc(sizeof(*tr), GFP_KERNEL);
 	if (!tr)
 		goto out;
-
-	/* is_root was checked earlier. No need for bpf_jit_charge_modmem() */
-	image = __bpf_image_alloc(false);
-	if (!image) {
+#ifdef CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
+	tr->fops = kzalloc(sizeof(struct ftrace_ops), GFP_KERNEL);
+	if (!tr->fops) {
 		kfree(tr);
 		tr = NULL;
 		goto out;
 	}
+	tr->fops->private = tr;
+	tr->fops->ops_func = bpf_tramp_ftrace_ops_func;
+#endif
 
 	tr->key = key;
 	INIT_HLIST_NODE(&tr->hlist);
@@ -130,22 +167,29 @@ struct bpf_trampoline *bpf_trampoline_lookup(u64 key)
 	mutex_init(&tr->mutex);
 	for (i = 0; i < BPF_TRAMP_MAX; i++)
 		INIT_HLIST_HEAD(&tr->progs_hlist[i]);
-	tr->image = image;
 out:
 	mutex_unlock(&trampoline_mutex);
 	return tr;
 }
 
-static int is_ftrace_location(void *ip)
+static int bpf_trampoline_module_get(struct bpf_trampoline *tr)
 {
-	long addr;
+	struct module *mod;
+	int err = 0;
 
-	addr = ftrace_location((long)ip);
-	if (!addr)
-		return 0;
-	if (WARN_ON_ONCE(addr != (long)ip))
-		return -EFAULT;
-	return 1;
+	preempt_disable();
+	mod = __module_text_address((unsigned long) tr->func.addr);
+	if (mod && !try_module_get(mod))
+		err = -ENOENT;
+	preempt_enable();
+	tr->mod = mod;
+	return err;
+}
+
+static void bpf_trampoline_module_put(struct bpf_trampoline *tr)
+{
+	module_put(tr->mod);
+	tr->mod = NULL;
 }
 
 static int unregister_fentry(struct bpf_trampoline *tr, void *old_addr)
@@ -154,21 +198,29 @@ static int unregister_fentry(struct bpf_trampoline *tr, void *old_addr)
 	int ret;
 
 	if (tr->func.ftrace_managed)
-		ret = unregister_ftrace_direct((long)ip, (long)old_addr);
+		ret = unregister_ftrace_direct_multi(tr->fops, (long)old_addr);
 	else
 		ret = bpf_arch_text_poke(ip, BPF_MOD_CALL, old_addr, NULL);
+
+	if (!ret)
+		bpf_trampoline_module_put(tr);
 	return ret;
 }
 
-static int modify_fentry(struct bpf_trampoline *tr, void *old_addr, void *new_addr)
+static int modify_fentry(struct bpf_trampoline *tr, void *old_addr, void *new_addr,
+			 bool lock_direct_mutex)
 {
 	void *ip = tr->func.addr;
 	int ret;
 
-	if (tr->func.ftrace_managed)
-		ret = modify_ftrace_direct((long)ip, (long)old_addr, (long)new_addr);
-	else
+	if (tr->func.ftrace_managed) {
+		if (lock_direct_mutex)
+			ret = modify_ftrace_direct_multi(tr->fops, (long)new_addr);
+		else
+			ret = modify_ftrace_direct_multi_nolock(tr->fops, (long)new_addr);
+	} else {
 		ret = bpf_arch_text_poke(ip, BPF_MOD_CALL, old_addr, new_addr);
+	}
 	return ret;
 }
 
@@ -176,175 +228,588 @@ static int modify_fentry(struct bpf_trampoline *tr, void *old_addr, void *new_ad
 static int register_fentry(struct bpf_trampoline *tr, void *new_addr)
 {
 	void *ip = tr->func.addr;
+	unsigned long faddr;
 	int ret;
 
-	ret = is_ftrace_location(ip);
-	if (ret < 0)
-		return ret;
-	tr->func.ftrace_managed = ret;
+	faddr = ftrace_location((unsigned long)ip);
+	if (faddr) {
+		if (!tr->fops)
+			return -ENOTSUPP;
+		tr->func.ftrace_managed = true;
+	}
 
-	if (tr->func.ftrace_managed)
-		ret = register_ftrace_direct((long)ip, (long)new_addr);
-	else
+	if (bpf_trampoline_module_get(tr))
+		return -ENOENT;
+
+	if (tr->func.ftrace_managed) {
+		ftrace_set_filter_ip(tr->fops, (unsigned long)ip, 0, 1);
+		ret = register_ftrace_direct_multi(tr->fops, (long)new_addr);
+	} else {
 		ret = bpf_arch_text_poke(ip, BPF_MOD_CALL, NULL, new_addr);
+	}
+
+	if (ret)
+		bpf_trampoline_module_put(tr);
 	return ret;
 }
 
-/* Each call __bpf_prog_enter + call bpf_func + call __bpf_prog_exit is ~50
- * bytes on x86.  Pick a number to fit into BPF_IMAGE_SIZE / 2
- */
-#define BPF_MAX_TRAMP_PROGS 40
-
-static int bpf_trampoline_update(struct bpf_trampoline *tr)
-{
-	void *old_image = tr->image + ((tr->selector + 1) & 1) * BPF_IMAGE_SIZE/2;
-	void *new_image = tr->image + (tr->selector & 1) * BPF_IMAGE_SIZE/2;
-	struct bpf_prog *progs_to_run[BPF_MAX_TRAMP_PROGS];
-	int fentry_cnt = tr->progs_cnt[BPF_TRAMP_FENTRY];
-	int fexit_cnt = tr->progs_cnt[BPF_TRAMP_FEXIT];
-	struct bpf_prog **progs, **fentry, **fexit;
-	u32 flags = BPF_TRAMP_F_RESTORE_REGS;
-	struct bpf_prog_aux *aux;
-	int err;
+static struct bpf_tramp_links *
+bpf_trampoline_get_progs(const struct bpf_trampoline *tr, int *total, bool *ip_arg)
+{
+	struct bpf_tramp_link *link;
+	struct bpf_tramp_links *tlinks;
+	struct bpf_tramp_link **links;
+	int kind;
+
+	*total = 0;
+	tlinks = kcalloc(BPF_TRAMP_MAX, sizeof(*tlinks), GFP_KERNEL);
+	if (!tlinks)
+		return ERR_PTR(-ENOMEM);
+
+	for (kind = 0; kind < BPF_TRAMP_MAX; kind++) {
+		tlinks[kind].nr_links = tr->progs_cnt[kind];
+		*total += tr->progs_cnt[kind];
+		links = tlinks[kind].links;
+
+		hlist_for_each_entry(link, &tr->progs_hlist[kind], tramp_hlist) {
+			*ip_arg |= link->link.prog->call_get_func_ip;
+			*links++ = link;
+		}
+	}
+	return tlinks;
+}
+
+static void __bpf_tramp_image_put_deferred(struct work_struct *work)
+{
+	struct bpf_tramp_image *im;
+
+	im = container_of(work, struct bpf_tramp_image, work);
+	bpf_image_ksym_del(&im->ksym);
+	bpf_jit_free_exec(im->image);
+	bpf_jit_uncharge_modmem(PAGE_SIZE);
+	percpu_ref_exit(&im->pcref);
+	kfree_rcu(im, rcu);
+}
+
+/* callback, fexit step 3 or fentry step 2 */
+static void __bpf_tramp_image_put_rcu(struct rcu_head *rcu)
+{
+	struct bpf_tramp_image *im;
+
+	im = container_of(rcu, struct bpf_tramp_image, rcu);
+	INIT_WORK(&im->work, __bpf_tramp_image_put_deferred);
+	schedule_work(&im->work);
+}
+
+/* callback, fexit step 2. Called after percpu_ref_kill confirms. */
+static void __bpf_tramp_image_release(struct percpu_ref *pcref)
+{
+	struct bpf_tramp_image *im;
+
+	im = container_of(pcref, struct bpf_tramp_image, pcref);
+	call_rcu_tasks(&im->rcu, __bpf_tramp_image_put_rcu);
+}
+
+/* callback, fexit or fentry step 1 */
+static void __bpf_tramp_image_put_rcu_tasks(struct rcu_head *rcu)
+{
+	struct bpf_tramp_image *im;
+
+	im = container_of(rcu, struct bpf_tramp_image, rcu);
+	if (im->ip_after_call)
+		/* the case of fmod_ret/fexit trampoline and CONFIG_PREEMPTION=y */
+		percpu_ref_kill(&im->pcref);
+	else
+		/* the case of fentry trampoline */
+		call_rcu_tasks(&im->rcu, __bpf_tramp_image_put_rcu);
+}
+
+static void bpf_tramp_image_put(struct bpf_tramp_image *im)
+{
+	/* The trampoline image that calls original function is using:
+	 * rcu_read_lock_trace to protect sleepable bpf progs
+	 * rcu_read_lock to protect normal bpf progs
+	 * percpu_ref to protect trampoline itself
+	 * rcu tasks to protect trampoline asm not covered by percpu_ref
+	 * (which are few asm insns before __bpf_tramp_enter and
+	 *  after __bpf_tramp_exit)
+	 *
+	 * The trampoline is unreachable before bpf_tramp_image_put().
+	 *
+	 * First, patch the trampoline to avoid calling into fexit progs.
+	 * The progs will be freed even if the original function is still
+	 * executing or sleeping.
+	 * In case of CONFIG_PREEMPT=y use call_rcu_tasks() to wait on
+	 * first few asm instructions to execute and call into
+	 * __bpf_tramp_enter->percpu_ref_get.
+	 * Then use percpu_ref_kill to wait for the trampoline and the original
+	 * function to finish.
+	 * Then use call_rcu_tasks() to make sure few asm insns in
+	 * the trampoline epilogue are done as well.
+	 *
+	 * In !PREEMPT case the task that got interrupted in the first asm
+	 * insns won't go through an RCU quiescent state which the
+	 * percpu_ref_kill will be waiting for. Hence the first
+	 * call_rcu_tasks() is not necessary.
+	 */
+	if (im->ip_after_call) {
+		int err = bpf_arch_text_poke(im->ip_after_call, BPF_MOD_JUMP,
+					     NULL, im->ip_epilogue);
+		WARN_ON(err);
+		if (IS_ENABLED(CONFIG_PREEMPTION))
+			call_rcu_tasks(&im->rcu, __bpf_tramp_image_put_rcu_tasks);
+		else
+			percpu_ref_kill(&im->pcref);
+		return;
+	}
+
+	/* The trampoline without fexit and fmod_ret progs doesn't call original
+	 * function and doesn't use percpu_ref.
+	 * Use call_rcu_tasks_trace() to wait for sleepable progs to finish.
+	 * Then use call_rcu_tasks() to wait for the rest of trampoline asm
+	 * and normal progs.
+	 */
+	call_rcu_tasks_trace(&im->rcu, __bpf_tramp_image_put_rcu_tasks);
+}
+
+static struct bpf_tramp_image *bpf_tramp_image_alloc(u64 key, u32 idx)
+{
+	struct bpf_tramp_image *im;
+	struct bpf_ksym *ksym;
+	void *image;
+	int err = -ENOMEM;
+
+	im = kzalloc(sizeof(*im), GFP_KERNEL);
+	if (!im)
+		goto out;
+
+	err = bpf_jit_charge_modmem(PAGE_SIZE);
+	if (err)
+		goto out_free_im;
+
+	err = -ENOMEM;
+	im->image = image = bpf_jit_alloc_exec(PAGE_SIZE);
+	if (!image)
+		goto out_uncharge;
+	set_vm_flush_reset_perms(image);
 
-	if (fentry_cnt + fexit_cnt == 0) {
-		err = unregister_fentry(tr, old_image);
+	err = percpu_ref_init(&im->pcref, __bpf_tramp_image_release, 0, GFP_KERNEL);
+	if (err)
+		goto out_free_image;
+
+	ksym = &im->ksym;
+	INIT_LIST_HEAD_RCU(&ksym->lnode);
+	snprintf(ksym->name, KSYM_NAME_LEN, "bpf_trampoline_%llu_%u", key, idx);
+	bpf_image_ksym_add(image, ksym);
+	return im;
+
+out_free_image:
+	bpf_jit_free_exec(im->image);
+out_uncharge:
+	bpf_jit_uncharge_modmem(PAGE_SIZE);
+out_free_im:
+	kfree(im);
+out:
+	return ERR_PTR(err);
+}
+
+static int bpf_trampoline_update(struct bpf_trampoline *tr, bool lock_direct_mutex)
+{
+	struct bpf_tramp_image *im;
+	struct bpf_tramp_links *tlinks;
+	u32 orig_flags = tr->flags;
+	bool ip_arg = false;
+	int err, total;
+
+	tlinks = bpf_trampoline_get_progs(tr, &total, &ip_arg);
+	if (IS_ERR(tlinks))
+		return PTR_ERR(tlinks);
+
+	if (total == 0) {
+		err = unregister_fentry(tr, tr->cur_image->image);
+		bpf_tramp_image_put(tr->cur_image);
+		tr->cur_image = NULL;
 		tr->selector = 0;
 		goto out;
 	}
 
-	/* populate fentry progs */
-	fentry = progs = progs_to_run;
-	hlist_for_each_entry(aux, &tr->progs_hlist[BPF_TRAMP_FENTRY], tramp_hlist)
-		*progs++ = aux->prog;
+	im = bpf_tramp_image_alloc(tr->key, tr->selector);
+	if (IS_ERR(im)) {
+		err = PTR_ERR(im);
+		goto out;
+	}
 
-	/* populate fexit progs */
-	fexit = progs;
-	hlist_for_each_entry(aux, &tr->progs_hlist[BPF_TRAMP_FEXIT], tramp_hlist)
-		*progs++ = aux->prog;
+	/* clear all bits except SHARE_IPMODIFY */
+	tr->flags &= BPF_TRAMP_F_SHARE_IPMODIFY;
 
-	if (fexit_cnt)
-		flags = BPF_TRAMP_F_CALL_ORIG | BPF_TRAMP_F_SKIP_FRAME;
+	if (tlinks[BPF_TRAMP_FEXIT].nr_links ||
+	    tlinks[BPF_TRAMP_MODIFY_RETURN].nr_links) {
+		/* NOTE: BPF_TRAMP_F_RESTORE_REGS and BPF_TRAMP_F_SKIP_FRAME
+		 * should not be set together.
+		 */
+		tr->flags |= BPF_TRAMP_F_CALL_ORIG | BPF_TRAMP_F_SKIP_FRAME;
+	} else {
+		tr->flags |= BPF_TRAMP_F_RESTORE_REGS;
+	}
 
-	/* Though the second half of trampoline page is unused a task could be
-	 * preempted in the middle of the first half of trampoline and two
-	 * updates to trampoline would change the code from underneath the
-	 * preempted task. Hence wait for tasks to voluntarily schedule or go
-	 * to userspace.
-	 */
-	synchronize_rcu_tasks();
+	if (ip_arg)
+		tr->flags |= BPF_TRAMP_F_IP_ARG;
 
-	err = arch_prepare_bpf_trampoline(new_image, new_image + BPF_IMAGE_SIZE / 2,
-					  &tr->func.model, flags,
-					  fentry, fentry_cnt,
-					  fexit, fexit_cnt,
+#ifdef CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
+again:
+	if ((tr->flags & BPF_TRAMP_F_SHARE_IPMODIFY) &&
+	    (tr->flags & BPF_TRAMP_F_CALL_ORIG))
+		tr->flags |= BPF_TRAMP_F_ORIG_STACK;
+#endif
+
+	err = arch_prepare_bpf_trampoline(im, im->image, im->image + PAGE_SIZE,
+					  &tr->func.model, tr->flags, tlinks,
 					  tr->func.addr);
 	if (err < 0)
 		goto out;
 
-	if (tr->selector)
+	set_memory_ro((long)im->image, 1);
+	set_memory_x((long)im->image, 1);
+
+	WARN_ON(tr->cur_image && tr->selector == 0);
+	WARN_ON(!tr->cur_image && tr->selector);
+	if (tr->cur_image)
 		/* progs already running at this address */
-		err = modify_fentry(tr, old_image, new_image);
+		err = modify_fentry(tr, tr->cur_image->image, im->image, lock_direct_mutex);
 	else
 		/* first time registering */
-		err = register_fentry(tr, new_image);
+		err = register_fentry(tr, im->image);
+
+#ifdef CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
+	if (err == -EAGAIN) {
+		/* -EAGAIN from bpf_tramp_ftrace_ops_func. Now
+		 * BPF_TRAMP_F_SHARE_IPMODIFY is set, we can generate the
+		 * trampoline again, and retry register.
+		 */
+		/* reset fops->func and fops->trampoline for re-register */
+		tr->fops->func = NULL;
+		tr->fops->trampoline = 0;
+		goto again;
+	}
+#endif
 	if (err)
 		goto out;
+
+	if (tr->cur_image)
+		bpf_tramp_image_put(tr->cur_image);
+	tr->cur_image = im;
 	tr->selector++;
 out:
+	/* If any error happens, restore previous flags */
+	if (err)
+		tr->flags = orig_flags;
+	kfree(tlinks);
 	return err;
 }
 
-static enum bpf_tramp_prog_type bpf_attach_type_to_tramp(enum bpf_attach_type t)
+static enum bpf_tramp_prog_type bpf_attach_type_to_tramp(struct bpf_prog *prog)
 {
-	switch (t) {
+	switch (prog->expected_attach_type) {
 	case BPF_TRACE_FENTRY:
 		return BPF_TRAMP_FENTRY;
+	case BPF_MODIFY_RETURN:
+		return BPF_TRAMP_MODIFY_RETURN;
 	case BPF_TRACE_FEXIT:
 		return BPF_TRAMP_FEXIT;
+	case BPF_LSM_MAC:
+		if (!prog->aux->attach_func_proto->type)
+			/* The function returns void, we cannot modify its
+			 * return value.
+			 */
+			return BPF_TRAMP_FEXIT;
+		else
+			return BPF_TRAMP_MODIFY_RETURN;
 	default:
 		return BPF_TRAMP_REPLACE;
 	}
 }
 
-int bpf_trampoline_link_prog(struct bpf_prog *prog)
+static int __bpf_trampoline_link_prog(struct bpf_tramp_link *link, struct bpf_trampoline *tr)
 {
 	enum bpf_tramp_prog_type kind;
-	struct bpf_trampoline *tr;
+	struct bpf_tramp_link *link_exiting;
 	int err = 0;
-	int cnt;
+	int cnt = 0, i;
 
-	tr = prog->aux->trampoline;
-	kind = bpf_attach_type_to_tramp(prog->expected_attach_type);
-	mutex_lock(&tr->mutex);
-	if (tr->extension_prog) {
+	kind = bpf_attach_type_to_tramp(link->link.prog);
+	if (tr->extension_prog)
 		/* cannot attach fentry/fexit if extension prog is attached.
 		 * cannot overwrite extension prog either.
 		 */
-		err = -EBUSY;
-		goto out;
-	}
-	cnt = tr->progs_cnt[BPF_TRAMP_FENTRY] + tr->progs_cnt[BPF_TRAMP_FEXIT];
+		return -EBUSY;
+
+	for (i = 0; i < BPF_TRAMP_MAX; i++)
+		cnt += tr->progs_cnt[i];
+
 	if (kind == BPF_TRAMP_REPLACE) {
 		/* Cannot attach extension if fentry/fexit are in use. */
-		if (cnt) {
-			err = -EBUSY;
-			goto out;
-		}
-		tr->extension_prog = prog;
-		err = bpf_arch_text_poke(tr->func.addr, BPF_MOD_JUMP, NULL,
-					 prog->bpf_func);
-		goto out;
-	}
-	if (cnt >= BPF_MAX_TRAMP_PROGS) {
-		err = -E2BIG;
-		goto out;
+		if (cnt)
+			return -EBUSY;
+		tr->extension_prog = link->link.prog;
+		return bpf_arch_text_poke(tr->func.addr, BPF_MOD_JUMP, NULL,
+					  link->link.prog->bpf_func);
 	}
-	if (!hlist_unhashed(&prog->aux->tramp_hlist)) {
+	if (cnt >= BPF_MAX_TRAMP_LINKS)
+		return -E2BIG;
+	if (!hlist_unhashed(&link->tramp_hlist))
 		/* prog already linked */
-		err = -EBUSY;
-		goto out;
+		return -EBUSY;
+	hlist_for_each_entry(link_exiting, &tr->progs_hlist[kind], tramp_hlist) {
+		if (link_exiting->link.prog != link->link.prog)
+			continue;
+		/* prog already linked */
+		return -EBUSY;
 	}
-	hlist_add_head(&prog->aux->tramp_hlist, &tr->progs_hlist[kind]);
+
+	hlist_add_head(&link->tramp_hlist, &tr->progs_hlist[kind]);
 	tr->progs_cnt[kind]++;
-	err = bpf_trampoline_update(prog->aux->trampoline);
+	err = bpf_trampoline_update(tr, true /* lock_direct_mutex */);
 	if (err) {
-		hlist_del(&prog->aux->tramp_hlist);
+		hlist_del_init(&link->tramp_hlist);
 		tr->progs_cnt[kind]--;
 	}
-out:
+	return err;
+}
+
+int bpf_trampoline_link_prog(struct bpf_tramp_link *link, struct bpf_trampoline *tr)
+{
+	int err;
+
+	mutex_lock(&tr->mutex);
+	err = __bpf_trampoline_link_prog(link, tr);
 	mutex_unlock(&tr->mutex);
 	return err;
 }
 
-/* bpf_trampoline_unlink_prog() should never fail. */
-int bpf_trampoline_unlink_prog(struct bpf_prog *prog)
+static int __bpf_trampoline_unlink_prog(struct bpf_tramp_link *link, struct bpf_trampoline *tr)
 {
 	enum bpf_tramp_prog_type kind;
-	struct bpf_trampoline *tr;
 	int err;
 
-	tr = prog->aux->trampoline;
-	kind = bpf_attach_type_to_tramp(prog->expected_attach_type);
-	mutex_lock(&tr->mutex);
+	kind = bpf_attach_type_to_tramp(link->link.prog);
 	if (kind == BPF_TRAMP_REPLACE) {
 		WARN_ON_ONCE(!tr->extension_prog);
 		err = bpf_arch_text_poke(tr->func.addr, BPF_MOD_JUMP,
 					 tr->extension_prog->bpf_func, NULL);
 		tr->extension_prog = NULL;
-		goto out;
+		return err;
 	}
-	hlist_del(&prog->aux->tramp_hlist);
+	hlist_del_init(&link->tramp_hlist);
 	tr->progs_cnt[kind]--;
-	err = bpf_trampoline_update(prog->aux->trampoline);
-out:
+	return bpf_trampoline_update(tr, true /* lock_direct_mutex */);
+}
+
+/* bpf_trampoline_unlink_prog() should never fail. */
+int bpf_trampoline_unlink_prog(struct bpf_tramp_link *link, struct bpf_trampoline *tr)
+{
+	int err;
+
+	mutex_lock(&tr->mutex);
+	err = __bpf_trampoline_unlink_prog(link, tr);
 	mutex_unlock(&tr->mutex);
 	return err;
 }
 
+#if defined(CONFIG_CGROUP_BPF) && defined(CONFIG_BPF_LSM)
+static void bpf_shim_tramp_link_release(struct bpf_link *link)
+{
+	struct bpf_shim_tramp_link *shim_link =
+		container_of(link, struct bpf_shim_tramp_link, link.link);
+
+	/* paired with 'shim_link->trampoline = tr' in bpf_trampoline_link_cgroup_shim */
+	if (!shim_link->trampoline)
+		return;
+
+	WARN_ON_ONCE(bpf_trampoline_unlink_prog(&shim_link->link, shim_link->trampoline));
+	bpf_trampoline_put(shim_link->trampoline);
+}
+
+static void bpf_shim_tramp_link_dealloc(struct bpf_link *link)
+{
+	struct bpf_shim_tramp_link *shim_link =
+		container_of(link, struct bpf_shim_tramp_link, link.link);
+
+	kfree(shim_link);
+}
+
+static const struct bpf_link_ops bpf_shim_tramp_link_lops = {
+	.release = bpf_shim_tramp_link_release,
+	.dealloc = bpf_shim_tramp_link_dealloc,
+};
+
+static struct bpf_shim_tramp_link *cgroup_shim_alloc(const struct bpf_prog *prog,
+						     bpf_func_t bpf_func,
+						     int cgroup_atype)
+{
+	struct bpf_shim_tramp_link *shim_link = NULL;
+	struct bpf_prog *p;
+
+	shim_link = kzalloc(sizeof(*shim_link), GFP_USER);
+	if (!shim_link)
+		return NULL;
+
+	p = bpf_prog_alloc(1, 0);
+	if (!p) {
+		kfree(shim_link);
+		return NULL;
+	}
+
+	p->jited = false;
+	p->bpf_func = bpf_func;
+
+	p->aux->cgroup_atype = cgroup_atype;
+	p->aux->attach_func_proto = prog->aux->attach_func_proto;
+	p->aux->attach_btf_id = prog->aux->attach_btf_id;
+	p->aux->attach_btf = prog->aux->attach_btf;
+	btf_get(p->aux->attach_btf);
+	p->type = BPF_PROG_TYPE_LSM;
+	p->expected_attach_type = BPF_LSM_MAC;
+	bpf_prog_inc(p);
+	bpf_link_init(&shim_link->link.link, BPF_LINK_TYPE_UNSPEC,
+		      &bpf_shim_tramp_link_lops, p);
+	bpf_cgroup_atype_get(p->aux->attach_btf_id, cgroup_atype);
+
+	return shim_link;
+}
+
+static struct bpf_shim_tramp_link *cgroup_shim_find(struct bpf_trampoline *tr,
+						    bpf_func_t bpf_func)
+{
+	struct bpf_tramp_link *link;
+	int kind;
+
+	for (kind = 0; kind < BPF_TRAMP_MAX; kind++) {
+		hlist_for_each_entry(link, &tr->progs_hlist[kind], tramp_hlist) {
+			struct bpf_prog *p = link->link.prog;
+
+			if (p->bpf_func == bpf_func)
+				return container_of(link, struct bpf_shim_tramp_link, link);
+		}
+	}
+
+	return NULL;
+}
+
+int bpf_trampoline_link_cgroup_shim(struct bpf_prog *prog,
+				    int cgroup_atype)
+{
+	struct bpf_shim_tramp_link *shim_link = NULL;
+	struct bpf_attach_target_info tgt_info = {};
+	struct bpf_trampoline *tr;
+	bpf_func_t bpf_func;
+	u64 key;
+	int err;
+
+	err = bpf_check_attach_target(NULL, prog, NULL,
+				      prog->aux->attach_btf_id,
+				      &tgt_info);
+	if (err)
+		return err;
+
+	key = bpf_trampoline_compute_key(NULL, prog->aux->attach_btf,
+					 prog->aux->attach_btf_id);
+
+	bpf_lsm_find_cgroup_shim(prog, &bpf_func);
+	tr = bpf_trampoline_get(key, &tgt_info);
+	if (!tr)
+		return  -ENOMEM;
+
+	mutex_lock(&tr->mutex);
+
+	shim_link = cgroup_shim_find(tr, bpf_func);
+	if (shim_link) {
+		/* Reusing existing shim attached by the other program. */
+		bpf_link_inc(&shim_link->link.link);
+
+		mutex_unlock(&tr->mutex);
+		bpf_trampoline_put(tr); /* bpf_trampoline_get above */
+		return 0;
+	}
+
+	/* Allocate and install new shim. */
+
+	shim_link = cgroup_shim_alloc(prog, bpf_func, cgroup_atype);
+	if (!shim_link) {
+		err = -ENOMEM;
+		goto err;
+	}
+
+	err = __bpf_trampoline_link_prog(&shim_link->link, tr);
+	if (err)
+		goto err;
+
+	shim_link->trampoline = tr;
+	/* note, we're still holding tr refcnt from above */
+
+	mutex_unlock(&tr->mutex);
+
+	return 0;
+err:
+	mutex_unlock(&tr->mutex);
+
+	if (shim_link)
+		bpf_link_put(&shim_link->link.link);
+
+	/* have to release tr while _not_ holding its mutex */
+	bpf_trampoline_put(tr); /* bpf_trampoline_get above */
+
+	return err;
+}
+
+void bpf_trampoline_unlink_cgroup_shim(struct bpf_prog *prog)
+{
+	struct bpf_shim_tramp_link *shim_link = NULL;
+	struct bpf_trampoline *tr;
+	bpf_func_t bpf_func;
+	u64 key;
+
+	key = bpf_trampoline_compute_key(NULL, prog->aux->attach_btf,
+					 prog->aux->attach_btf_id);
+
+	bpf_lsm_find_cgroup_shim(prog, &bpf_func);
+	tr = bpf_trampoline_lookup(key);
+	if (WARN_ON_ONCE(!tr))
+		return;
+
+	mutex_lock(&tr->mutex);
+	shim_link = cgroup_shim_find(tr, bpf_func);
+	mutex_unlock(&tr->mutex);
+
+	if (shim_link)
+		bpf_link_put(&shim_link->link.link);
+
+	bpf_trampoline_put(tr); /* bpf_trampoline_lookup above */
+}
+#endif
+
+struct bpf_trampoline *bpf_trampoline_get(u64 key,
+					  struct bpf_attach_target_info *tgt_info)
+{
+	struct bpf_trampoline *tr;
+
+	tr = bpf_trampoline_lookup(key);
+	if (!tr)
+		return NULL;
+
+	mutex_lock(&tr->mutex);
+	if (tr->func.addr)
+		goto out;
+
+	memcpy(&tr->func.model, &tgt_info->fmodel, sizeof(tgt_info->fmodel));
+	tr->func.addr = (void *)tgt_info->tgt_addr;
+out:
+	mutex_unlock(&tr->mutex);
+	return tr;
+}
+
 void bpf_trampoline_put(struct bpf_trampoline *tr)
 {
-	struct bpf_image *image;
+	int i;
 
 	if (!tr)
 		return;
@@ -352,64 +817,190 @@ void bpf_trampoline_put(struct bpf_trampoline *tr)
 	if (!refcount_dec_and_test(&tr->refcnt))
 		goto out;
 	WARN_ON_ONCE(mutex_is_locked(&tr->mutex));
-	if (WARN_ON_ONCE(!hlist_empty(&tr->progs_hlist[BPF_TRAMP_FENTRY])))
-		goto out;
-	if (WARN_ON_ONCE(!hlist_empty(&tr->progs_hlist[BPF_TRAMP_FEXIT])))
-		goto out;
-	image = container_of(tr->image, struct bpf_image, data);
-	latch_tree_erase(&image->tnode, &image_tree, &image_tree_ops);
-	/* wait for tasks to get out of trampoline before freeing it */
-	synchronize_rcu_tasks();
-	bpf_jit_free_exec(image);
+
+	for (i = 0; i < BPF_TRAMP_MAX; i++)
+		if (WARN_ON_ONCE(!hlist_empty(&tr->progs_hlist[i])))
+			goto out;
+
+	/* This code will be executed even when the last bpf_tramp_image
+	 * is alive. All progs are detached from the trampoline and the
+	 * trampoline image is patched with jmp into epilogue to skip
+	 * fexit progs. The fentry-only trampoline will be freed via
+	 * multiple rcu callbacks.
+	 */
 	hlist_del(&tr->hlist);
+	if (tr->fops) {
+		ftrace_free_filter(tr->fops);
+		kfree(tr->fops);
+	}
 	kfree(tr);
 out:
 	mutex_unlock(&trampoline_mutex);
 }
 
-/* The logic is similar to BPF_PROG_RUN, but with explicit rcu and preempt that
- * are needed for trampoline. The macro is split into
- * call _bpf_prog_enter
+#define NO_START_TIME 1
+static __always_inline u64 notrace bpf_prog_start_time(void)
+{
+	u64 start = NO_START_TIME;
+
+	if (static_branch_unlikely(&bpf_stats_enabled_key)) {
+		start = sched_clock();
+		if (unlikely(!start))
+			start = NO_START_TIME;
+	}
+	return start;
+}
+
+/* The logic is similar to bpf_prog_run(), but with an explicit
+ * rcu_read_lock() and migrate_disable() which are required
+ * for the trampoline. The macro is split into
+ * call __bpf_prog_enter
  * call prog->bpf_func
  * call __bpf_prog_exit
+ *
+ * __bpf_prog_enter returns:
+ * 0 - skip execution of the bpf prog
+ * 1 - execute bpf prog
+ * [2..MAX_U64] - execute bpf prog and record execution time.
+ *     This is start time.
  */
-u64 notrace __bpf_prog_enter(void)
+u64 notrace __bpf_prog_enter(struct bpf_prog *prog, struct bpf_tramp_run_ctx *run_ctx)
+	__acquires(RCU)
 {
-	u64 start = 0;
-
 	rcu_read_lock();
-	preempt_disable();
-	if (static_branch_unlikely(&bpf_stats_enabled_key))
-		start = sched_clock();
-	return start;
+	migrate_disable();
+
+	run_ctx->saved_run_ctx = bpf_set_run_ctx(&run_ctx->run_ctx);
+
+	if (unlikely(this_cpu_inc_return(*(prog->active)) != 1)) {
+		bpf_prog_inc_misses_counter(prog);
+		return 0;
+	}
+	return bpf_prog_start_time();
 }
 
-void notrace __bpf_prog_exit(struct bpf_prog *prog, u64 start)
+static void notrace update_prog_stats(struct bpf_prog *prog,
+				      u64 start)
 {
 	struct bpf_prog_stats *stats;
 
 	if (static_branch_unlikely(&bpf_stats_enabled_key) &&
-	    /* static_key could be enabled in __bpf_prog_enter
-	     * and disabled in __bpf_prog_exit.
+	    /* static_key could be enabled in __bpf_prog_enter*
+	     * and disabled in __bpf_prog_exit*.
 	     * And vice versa.
-	     * Hence check that 'start' is not zero.
+	     * Hence check that 'start' is valid.
 	     */
-	    start) {
-		stats = this_cpu_ptr(prog->aux->stats);
-		u64_stats_update_begin(&stats->syncp);
-		stats->cnt++;
-		stats->nsecs += sched_clock() - start;
-		u64_stats_update_end(&stats->syncp);
+	    start > NO_START_TIME) {
+		unsigned long flags;
+
+		stats = this_cpu_ptr(prog->stats);
+		flags = u64_stats_update_begin_irqsave(&stats->syncp);
+		u64_stats_inc(&stats->cnt);
+		u64_stats_add(&stats->nsecs, sched_clock() - start);
+		u64_stats_update_end_irqrestore(&stats->syncp, flags);
 	}
-	preempt_enable();
+}
+
+void notrace __bpf_prog_exit(struct bpf_prog *prog, u64 start, struct bpf_tramp_run_ctx *run_ctx)
+	__releases(RCU)
+{
+	bpf_reset_run_ctx(run_ctx->saved_run_ctx);
+
+	update_prog_stats(prog, start);
+	this_cpu_dec(*(prog->active));
+	migrate_enable();
 	rcu_read_unlock();
 }
 
+u64 notrace __bpf_prog_enter_lsm_cgroup(struct bpf_prog *prog,
+					struct bpf_tramp_run_ctx *run_ctx)
+	__acquires(RCU)
+{
+	/* Runtime stats are exported via actual BPF_LSM_CGROUP
+	 * programs, not the shims.
+	 */
+	rcu_read_lock();
+	migrate_disable();
+
+	run_ctx->saved_run_ctx = bpf_set_run_ctx(&run_ctx->run_ctx);
+
+	return NO_START_TIME;
+}
+
+void notrace __bpf_prog_exit_lsm_cgroup(struct bpf_prog *prog, u64 start,
+					struct bpf_tramp_run_ctx *run_ctx)
+	__releases(RCU)
+{
+	bpf_reset_run_ctx(run_ctx->saved_run_ctx);
+
+	migrate_enable();
+	rcu_read_unlock();
+}
+
+u64 notrace __bpf_prog_enter_sleepable(struct bpf_prog *prog, struct bpf_tramp_run_ctx *run_ctx)
+{
+	rcu_read_lock_trace();
+	migrate_disable();
+	might_fault();
+
+	if (unlikely(this_cpu_inc_return(*(prog->active)) != 1)) {
+		bpf_prog_inc_misses_counter(prog);
+		return 0;
+	}
+
+	run_ctx->saved_run_ctx = bpf_set_run_ctx(&run_ctx->run_ctx);
+
+	return bpf_prog_start_time();
+}
+
+void notrace __bpf_prog_exit_sleepable(struct bpf_prog *prog, u64 start,
+				       struct bpf_tramp_run_ctx *run_ctx)
+{
+	bpf_reset_run_ctx(run_ctx->saved_run_ctx);
+
+	update_prog_stats(prog, start);
+	this_cpu_dec(*(prog->active));
+	migrate_enable();
+	rcu_read_unlock_trace();
+}
+
+u64 notrace __bpf_prog_enter_struct_ops(struct bpf_prog *prog,
+					struct bpf_tramp_run_ctx *run_ctx)
+	__acquires(RCU)
+{
+	rcu_read_lock();
+	migrate_disable();
+
+	run_ctx->saved_run_ctx = bpf_set_run_ctx(&run_ctx->run_ctx);
+
+	return bpf_prog_start_time();
+}
+
+void notrace __bpf_prog_exit_struct_ops(struct bpf_prog *prog, u64 start,
+					struct bpf_tramp_run_ctx *run_ctx)
+	__releases(RCU)
+{
+	bpf_reset_run_ctx(run_ctx->saved_run_ctx);
+
+	update_prog_stats(prog, start);
+	migrate_enable();
+	rcu_read_unlock();
+}
+
+void notrace __bpf_tramp_enter(struct bpf_tramp_image *tr)
+{
+	percpu_ref_get(&tr->pcref);
+}
+
+void notrace __bpf_tramp_exit(struct bpf_tramp_image *tr)
+{
+	percpu_ref_put(&tr->pcref);
+}
+
 int __weak
-arch_prepare_bpf_trampoline(void *image, void *image_end,
+arch_prepare_bpf_trampoline(struct bpf_tramp_image *tr, void *image, void *image_end,
 			    const struct btf_func_model *m, u32 flags,
-			    struct bpf_prog **fentry_progs, int fentry_cnt,
-			    struct bpf_prog **fexit_progs, int fexit_cnt,
+			    struct bpf_tramp_links *tlinks,
 			    void *orig_call)
 {
 	return -ENOTSUPP;