[PATCH bpf-next v2 0/8] Implement task_local_storage

[PATCH bpf-next v2 0/8] Implement task_local_storage

[KP Singh]

From: KP Singh <kpsingh@google.com>

# v1 -> v2

- Updated the refcounting for task_struct and simplified conversion
  of fd -> struct pid.
- Some fixes suggested by Martin and Andrii, notably:
   * long return type for the bpf_task_storage_delete helper (update
     for bpf_inode_storage_delete will be sent separately).
   * Remove extra nullness check to task_storage_ptr in map syscall
     ops.
   * Changed the argument signature of the BPF helpers to use
     task_struct pointer in uapi headers.
   * Remove unnecessary verifier logic for the bpf_get_current_task_btf
     helper.
   * Split the changes for bpftool and libbpf.
- Exercised syscall operations for local storage (kept a simpler verison
  in test_local_storage.c, the eventual goal will be to update
  sk_storage_map.c for all local storage types).
- Formatting fixes + Rebase.

We already have socket and inode local storage since [1]

This patch series:

* Implements bpf_local_storage for task_struct.
* Implements the bpf_get_current_task_btf helper which returns a BTF
  pointer to the current task. Not only is this generally cleaner
  (reading from the task_struct currently requires BPF_CORE_READ), it
  also allows the BTF pointer to be used in task_local_storage helpers.
* In order to implement this helper, a RET_PTR_TO_BTF_ID is introduced
  which works similar to RET_PTR_TO_BTF_ID_OR_NULL but does not require
  a nullness check.
* Implements a detection in selftests which uses the
  task local storage to deny a running executable from unlinking itself.

[1]: https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next.git/commit/?id=f836a56e84ffc9f1a1cd73f77e10404ca46a4616

KP Singh (8):
  bpf: Implement task local storage
  libbpf: Add support for task local storage
  bpftool: Add support for task local storage
  bpf: Implement get_current_task_btf and RET_PTR_TO_BTF_ID
  bpf: Fix tests for local_storage
  bpf: Update selftests for local_storage to use vmlinux.h
  bpf: Add tests for task_local_storage
  bpf: Exercise syscall operations for inode and sk storage

 include/linux/bpf.h                           |   1 +
 include/linux/bpf_lsm.h                       |  23 ++
 include/linux/bpf_types.h                     |   1 +
 include/uapi/linux/bpf.h                      |  48 +++
 kernel/bpf/Makefile                           |   1 +
 kernel/bpf/bpf_lsm.c                          |   4 +
 kernel/bpf/bpf_task_storage.c                 | 313 ++++++++++++++++++
 kernel/bpf/syscall.c                          |   3 +-
 kernel/bpf/verifier.c                         |  17 +-
 kernel/trace/bpf_trace.c                      |  16 +
 security/bpf/hooks.c                          |   2 +
 .../bpf/bpftool/Documentation/bpftool-map.rst |   3 +-
 tools/bpf/bpftool/bash-completion/bpftool     |   2 +-
 tools/bpf/bpftool/map.c                       |   4 +-
 tools/include/uapi/linux/bpf.h                |  48 +++
 tools/lib/bpf/libbpf_probes.c                 |   2 +
 .../bpf/prog_tests/test_local_storage.c       | 181 +++++++++-
 .../selftests/bpf/progs/local_storage.c       |  87 +++--
 18 files changed, 705 insertions(+), 51 deletions(-)
 create mode 100644 kernel/bpf/bpf_task_storage.c

-- 
2.29.1.341.ge80a0c044ae-goog

[PATCH bpf-next v2 1/8] bpf: Implement task local storage

[KP Singh]

From: KP Singh <kpsingh@google.com>

Similar to bpf_local_storage for sockets and inodes add local storage
for task_struct.

The life-cycle of storage is managed with the life-cycle of the
task_struct.  i.e. the storage is destroyed along with the owning task
with a callback to the bpf_task_storage_free from the task_free LSM
hook.

The BPF LSM allocates an __rcu pointer to the bpf_local_storage in
the security blob which are now stackable and can co-exist with other
LSMs.

The userspace map operations can be done by using a pid fd as a key
passed to the lookup, update and delete operations.

Signed-off-by: KP Singh <kpsingh@google.com>
---
 include/linux/bpf_lsm.h        |  23 +++
 include/linux/bpf_types.h      |   1 +
 include/uapi/linux/bpf.h       |  39 ++++
 kernel/bpf/Makefile            |   1 +
 kernel/bpf/bpf_lsm.c           |   4 +
 kernel/bpf/bpf_task_storage.c  | 313 +++++++++++++++++++++++++++++++++
 kernel/bpf/syscall.c           |   3 +-
 kernel/bpf/verifier.c          |  10 ++
 security/bpf/hooks.c           |   2 +
 tools/include/uapi/linux/bpf.h |  39 ++++
 10 files changed, 434 insertions(+), 1 deletion(-)
 create mode 100644 kernel/bpf/bpf_task_storage.c

diff --git a/include/linux/bpf_lsm.h b/include/linux/bpf_lsm.h
index aaacb6aafc87..326cb68a3632 100644
--- a/include/linux/bpf_lsm.h
+++ b/include/linux/bpf_lsm.h
@@ -7,6 +7,7 @@
 #ifndef _LINUX_BPF_LSM_H
 #define _LINUX_BPF_LSM_H
 
+#include "linux/sched.h"
 #include <linux/bpf.h>
 #include <linux/lsm_hooks.h>
 
@@ -35,9 +36,21 @@ static inline struct bpf_storage_blob *bpf_inode(
 	return inode->i_security + bpf_lsm_blob_sizes.lbs_inode;
 }
 
+static inline struct bpf_storage_blob *bpf_task(
+	const struct task_struct *task)
+{
+	if (unlikely(!task->security))
+		return NULL;
+
+	return task->security + bpf_lsm_blob_sizes.lbs_task;
+}
+
 extern const struct bpf_func_proto bpf_inode_storage_get_proto;
 extern const struct bpf_func_proto bpf_inode_storage_delete_proto;
+extern const struct bpf_func_proto bpf_task_storage_get_proto;
+extern const struct bpf_func_proto bpf_task_storage_delete_proto;
 void bpf_inode_storage_free(struct inode *inode);
+void bpf_task_storage_free(struct task_struct *task);
 
 #else /* !CONFIG_BPF_LSM */
 
@@ -53,10 +66,20 @@ static inline struct bpf_storage_blob *bpf_inode(
 	return NULL;
 }
 
+static inline struct bpf_storage_blob *bpf_task(
+	const struct task_struct *task)
+{
+	return NULL;
+}
+
 static inline void bpf_inode_storage_free(struct inode *inode)
 {
 }
 
+static inline void bpf_task_storage_free(struct task_struct *task)
+{
+}
+
 #endif /* CONFIG_BPF_LSM */
 
 #endif /* _LINUX_BPF_LSM_H */
diff --git a/include/linux/bpf_types.h b/include/linux/bpf_types.h
index 2e6f568377f1..99f7fd657d87 100644
--- a/include/linux/bpf_types.h
+++ b/include/linux/bpf_types.h
@@ -109,6 +109,7 @@ BPF_MAP_TYPE(BPF_MAP_TYPE_SOCKHASH, sock_hash_ops)
 #endif
 #ifdef CONFIG_BPF_LSM
 BPF_MAP_TYPE(BPF_MAP_TYPE_INODE_STORAGE, inode_storage_map_ops)
+BPF_MAP_TYPE(BPF_MAP_TYPE_TASK_STORAGE, task_storage_map_ops)
 #endif
 BPF_MAP_TYPE(BPF_MAP_TYPE_CPUMAP, cpu_map_ops)
 #if defined(CONFIG_XDP_SOCKETS)
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index e6ceac3f7d62..f4037b2161a6 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -157,6 +157,7 @@ enum bpf_map_type {
 	BPF_MAP_TYPE_STRUCT_OPS,
 	BPF_MAP_TYPE_RINGBUF,
 	BPF_MAP_TYPE_INODE_STORAGE,
+	BPF_MAP_TYPE_TASK_STORAGE,
 };
 
 /* Note that tracing related programs such as
@@ -3742,6 +3743,42 @@ union bpf_attr {
  * 	Return
  * 		The helper returns **TC_ACT_REDIRECT** on success or
  * 		**TC_ACT_SHOT** on error.
+ *
+ * void *bpf_task_storage_get(struct bpf_map *map, struct task_struct *task, void *value, u64 flags)
+ *	Description
+ *		Get a bpf_local_storage from the *task*.
+ *
+ *		Logically, it could be thought of as getting the value from
+ *		a *map* with *task* as the **key**.  From this
+ *		perspective,  the usage is not much different from
+ *		**bpf_map_lookup_elem**\ (*map*, **&**\ *task*) except this
+ *		helper enforces the key must be an task_struct and the map must also
+ *		be a **BPF_MAP_TYPE_TASK_STORAGE**.
+ *
+ *		Underneath, the value is stored locally at *task* instead of
+ *		the *map*.  The *map* is used as the bpf-local-storage
+ *		"type". The bpf-local-storage "type" (i.e. the *map*) is
+ *		searched against all bpf_local_storage residing at *task*.
+ *
+ *		An optional *flags* (**BPF_LOCAL_STORAGE_GET_F_CREATE**) can be
+ *		used such that a new bpf_local_storage will be
+ *		created if one does not exist.  *value* can be used
+ *		together with **BPF_LOCAL_STORAGE_GET_F_CREATE** to specify
+ *		the initial value of a bpf_local_storage.  If *value* is
+ *		**NULL**, the new bpf_local_storage will be zero initialized.
+ *	Return
+ *		A bpf_local_storage pointer is returned on success.
+ *
+ *		**NULL** if not found or there was an error in adding
+ *		a new bpf_local_storage.
+ *
+ * long bpf_task_storage_delete(struct bpf_map *map, struct task_struct *task)
+ *	Description
+ *		Delete a bpf_local_storage from a *task*.
+ *	Return
+ *		0 on success.
+ *
+ *		**-ENOENT** if the bpf_local_storage cannot be found.
  */
 #define __BPF_FUNC_MAPPER(FN)		\
 	FN(unspec),			\
@@ -3900,6 +3937,8 @@ union bpf_attr {
 	FN(bpf_per_cpu_ptr),            \
 	FN(bpf_this_cpu_ptr),		\
 	FN(redirect_peer),		\
+	FN(task_storage_get),		\
+	FN(task_storage_delete),	\
 	/* */
 
 /* integer value in 'imm' field of BPF_CALL instruction selects which helper
diff --git a/kernel/bpf/Makefile b/kernel/bpf/Makefile
index bdc8cd1b6767..f0b93ced5a7f 100644
--- a/kernel/bpf/Makefile
+++ b/kernel/bpf/Makefile
@@ -6,6 +6,7 @@ obj-$(CONFIG_BPF_SYSCALL) += syscall.o verifier.o inode.o helpers.o tnum.o bpf_i
 obj-$(CONFIG_BPF_SYSCALL) += hashtab.o arraymap.o percpu_freelist.o bpf_lru_list.o lpm_trie.o map_in_map.o
 obj-$(CONFIG_BPF_SYSCALL) += local_storage.o queue_stack_maps.o ringbuf.o
 obj-${CONFIG_BPF_LSM}	  += bpf_inode_storage.o
+obj-${CONFIG_BPF_LSM}	  += bpf_task_storage.o
 obj-$(CONFIG_BPF_SYSCALL) += disasm.o
 obj-$(CONFIG_BPF_JIT) += trampoline.o
 obj-$(CONFIG_BPF_SYSCALL) += btf.o
diff --git a/kernel/bpf/bpf_lsm.c b/kernel/bpf/bpf_lsm.c
index 78ea8a7bd27f..61f8cc52fd5b 100644
--- a/kernel/bpf/bpf_lsm.c
+++ b/kernel/bpf/bpf_lsm.c
@@ -59,6 +59,10 @@ bpf_lsm_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 		return &bpf_sk_storage_get_proto;
 	case BPF_FUNC_sk_storage_delete:
 		return &bpf_sk_storage_delete_proto;
+	case BPF_FUNC_task_storage_get:
+		return &bpf_task_storage_get_proto;
+	case BPF_FUNC_task_storage_delete:
+		return &bpf_task_storage_delete_proto;
 	default:
 		return tracing_prog_func_proto(func_id, prog);
 	}
diff --git a/kernel/bpf/bpf_task_storage.c b/kernel/bpf/bpf_task_storage.c
new file mode 100644
index 000000000000..f5ed5eedc532
--- /dev/null
+++ b/kernel/bpf/bpf_task_storage.c
@@ -0,0 +1,313 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2019 Facebook
+ * Copyright 2020 Google LLC.
+ */
+
+#include "linux/pid.h"
+#include "linux/sched.h"
+#include <linux/rculist.h>
+#include <linux/list.h>
+#include <linux/hash.h>
+#include <linux/types.h>
+#include <linux/spinlock.h>
+#include <linux/bpf.h>
+#include <linux/bpf_local_storage.h>
+#include <linux/filter.h>
+#include <uapi/linux/btf.h>
+#include <linux/bpf_lsm.h>
+#include <linux/btf_ids.h>
+#include <linux/fdtable.h>
+
+DEFINE_BPF_STORAGE_CACHE(task_cache);
+
+static struct bpf_local_storage __rcu **task_storage_ptr(void *owner)
+{
+	struct task_struct *task = owner;
+	struct bpf_storage_blob *bsb;
+
+	bsb = bpf_task(task);
+	if (!bsb)
+		return NULL;
+	return &bsb->storage;
+}
+
+static struct bpf_local_storage_data *
+task_storage_lookup(struct task_struct *task, struct bpf_map *map,
+		    bool cacheit_lockit)
+{
+	struct bpf_local_storage *task_storage;
+	struct bpf_local_storage_map *smap;
+	struct bpf_storage_blob *bsb;
+
+	bsb = bpf_task(task);
+	if (!bsb)
+		return NULL;
+
+	task_storage = rcu_dereference(bsb->storage);
+	if (!task_storage)
+		return NULL;
+
+	smap = (struct bpf_local_storage_map *)map;
+	return bpf_local_storage_lookup(task_storage, smap, cacheit_lockit);
+}
+
+void bpf_task_storage_free(struct task_struct *task)
+{
+	struct bpf_local_storage_elem *selem;
+	struct bpf_local_storage *local_storage;
+	bool free_task_storage = false;
+	struct bpf_storage_blob *bsb;
+	struct hlist_node *n;
+
+	bsb = bpf_task(task);
+	if (!bsb)
+		return;
+
+	rcu_read_lock();
+
+	local_storage = rcu_dereference(bsb->storage);
+	if (!local_storage) {
+		rcu_read_unlock();
+		return;
+	}
+
+	/* Neither the bpf_prog nor the bpf-map's syscall
+	 * could be modifying the local_storage->list now.
+	 * Thus, no elem can be added-to or deleted-from the
+	 * local_storage->list by the bpf_prog or by the bpf-map's syscall.
+	 *
+	 * It is racing with bpf_local_storage_map_free() alone
+	 * when unlinking elem from the local_storage->list and
+	 * the map's bucket->list.
+	 */
+	raw_spin_lock_bh(&local_storage->lock);
+	hlist_for_each_entry_safe(selem, n, &local_storage->list, snode) {
+		/* Always unlink from map before unlinking from
+		 * local_storage.
+		 */
+		bpf_selem_unlink_map(selem);
+		free_task_storage = bpf_selem_unlink_storage_nolock(
+			local_storage, selem, false);
+	}
+	raw_spin_unlock_bh(&local_storage->lock);
+	rcu_read_unlock();
+
+	/* free_task_storage should always be true as long as
+	 * local_storage->list was non-empty.
+	 */
+	if (free_task_storage)
+		kfree_rcu(local_storage, rcu);
+}
+
+static void *bpf_pid_task_storage_lookup_elem(struct bpf_map *map, void *key)
+{
+	struct bpf_local_storage_data *sdata;
+	struct task_struct *task;
+	unsigned int f_flags;
+	struct pid *pid;
+	int fd, err;
+
+	fd = *(int *)key;
+	pid = pidfd_get_pid(fd, &f_flags);
+	if (IS_ERR(pid))
+		return ERR_CAST(pid);
+
+	/* We should be in an RCU read side critical section, it should be safe
+	 * to call pid_task.
+	 */
+	WARN_ON_ONCE(!rcu_read_lock_held());
+	task = pid_task(pid, PIDTYPE_PID);
+	if (!task) {
+		err = -ENOENT;
+		goto out;
+	}
+
+	sdata = task_storage_lookup(task, map, true);
+	put_pid(pid);
+	return sdata ? sdata->data : NULL;
+out:
+	put_pid(pid);
+	return ERR_PTR(err);
+}
+
+static int bpf_pid_task_storage_update_elem(struct bpf_map *map, void *key,
+					    void *value, u64 map_flags)
+{
+	struct bpf_local_storage_data *sdata;
+	struct task_struct *task;
+	unsigned int f_flags;
+	struct pid *pid;
+	int fd, err;
+
+	fd = *(int *)key;
+	pid = pidfd_get_pid(fd, &f_flags);
+	if (IS_ERR(pid))
+		return PTR_ERR(pid);
+
+	/* We should be in an RCU read side critical section, it should be safe
+	 * to call pid_task.
+	 */
+	WARN_ON_ONCE(!rcu_read_lock_held());
+	task = pid_task(pid, PIDTYPE_PID);
+	if (!task) {
+		err = -ENOENT;
+		goto out;
+	}
+
+	sdata = bpf_local_storage_update(
+		task, (struct bpf_local_storage_map *)map, value, map_flags);
+
+	err = PTR_ERR_OR_ZERO(sdata);
+out:
+	put_pid(pid);
+	return err;
+}
+
+static int task_storage_delete(struct task_struct *task, struct bpf_map *map)
+{
+	struct bpf_local_storage_data *sdata;
+
+	sdata = task_storage_lookup(task, map, false);
+	if (!sdata)
+		return -ENOENT;
+
+	bpf_selem_unlink(SELEM(sdata));
+
+	return 0;
+}
+
+static int bpf_pid_task_storage_delete_elem(struct bpf_map *map, void *key)
+{
+	struct task_struct *task;
+	unsigned int f_flags;
+	struct pid *pid;
+	int fd, err;
+
+	fd = *(int *)key;
+	pid = pidfd_get_pid(fd, &f_flags);
+	if (IS_ERR(pid))
+		return PTR_ERR(pid);
+
+	/* We should be in an RCU read side critical section, it should be safe
+	 * to call pid_task.
+	 */
+	WARN_ON_ONCE(!rcu_read_lock_held());
+	task = pid_task(pid, PIDTYPE_PID);
+	if (!task) {
+		err = -ENOENT;
+		goto out;
+	}
+
+	err = task_storage_delete(task, map);
+out:
+	put_pid(pid);
+	return err;
+}
+
+BPF_CALL_4(bpf_task_storage_get, struct bpf_map *, map, struct task_struct *,
+	   task, void *, value, u64, flags)
+{
+	struct bpf_local_storage_data *sdata;
+
+	if (flags & ~(BPF_LOCAL_STORAGE_GET_F_CREATE))
+		return (unsigned long)NULL;
+
+	/* explicitly check that the task_storage_ptr is not
+	 * NULL as task_storage_lookup returns NULL in this case and
+	 * bpf_local_storage_update expects the owner to have a
+	 * valid storage pointer.
+	 */
+	if (!task_storage_ptr(task))
+		return (unsigned long)NULL;
+
+	sdata = task_storage_lookup(task, map, true);
+	if (sdata)
+		return (unsigned long)sdata->data;
+
+	/* This helper must only called from where the task is guaranteed
+	 * to have a refcount and cannot be freed.
+	 */
+	if (flags & BPF_LOCAL_STORAGE_GET_F_CREATE) {
+		sdata = bpf_local_storage_update(
+			task, (struct bpf_local_storage_map *)map, value,
+			BPF_NOEXIST);
+		return IS_ERR(sdata) ? (unsigned long)NULL :
+					     (unsigned long)sdata->data;
+	}
+
+	return (unsigned long)NULL;
+}
+
+BPF_CALL_2(bpf_task_storage_delete, struct bpf_map *, map, struct task_struct *,
+	   task)
+{
+	/* This helper must only called from where the task is guaranteed
+	 * to have a refcount and cannot be freed.
+	 */
+	return task_storage_delete(task, map);
+}
+
+static int notsupp_get_next_key(struct bpf_map *map, void *key, void *next_key)
+{
+	return -ENOTSUPP;
+}
+
+static struct bpf_map *task_storage_map_alloc(union bpf_attr *attr)
+{
+	struct bpf_local_storage_map *smap;
+
+	smap = bpf_local_storage_map_alloc(attr);
+	if (IS_ERR(smap))
+		return ERR_CAST(smap);
+
+	smap->cache_idx = bpf_local_storage_cache_idx_get(&task_cache);
+	return &smap->map;
+}
+
+static void task_storage_map_free(struct bpf_map *map)
+{
+	struct bpf_local_storage_map *smap;
+
+	smap = (struct bpf_local_storage_map *)map;
+	bpf_local_storage_cache_idx_free(&task_cache, smap->cache_idx);
+	bpf_local_storage_map_free(smap);
+}
+
+static int task_storage_map_btf_id;
+const struct bpf_map_ops task_storage_map_ops = {
+	.map_meta_equal = bpf_map_meta_equal,
+	.map_alloc_check = bpf_local_storage_map_alloc_check,
+	.map_alloc = task_storage_map_alloc,
+	.map_free = task_storage_map_free,
+	.map_get_next_key = notsupp_get_next_key,
+	.map_lookup_elem = bpf_pid_task_storage_lookup_elem,
+	.map_update_elem = bpf_pid_task_storage_update_elem,
+	.map_delete_elem = bpf_pid_task_storage_delete_elem,
+	.map_check_btf = bpf_local_storage_map_check_btf,
+	.map_btf_name = "bpf_local_storage_map",
+	.map_btf_id = &task_storage_map_btf_id,
+	.map_owner_storage_ptr = task_storage_ptr,
+};
+
+BTF_ID_LIST_SINGLE(bpf_task_storage_btf_ids, struct, task_struct)
+
+const struct bpf_func_proto bpf_task_storage_get_proto = {
+	.func = bpf_task_storage_get,
+	.gpl_only = false,
+	.ret_type = RET_PTR_TO_MAP_VALUE_OR_NULL,
+	.arg1_type = ARG_CONST_MAP_PTR,
+	.arg2_type = ARG_PTR_TO_BTF_ID,
+	.arg2_btf_id = &bpf_task_storage_btf_ids[0],
+	.arg3_type = ARG_PTR_TO_MAP_VALUE_OR_NULL,
+	.arg4_type = ARG_ANYTHING,
+};
+
+const struct bpf_func_proto bpf_task_storage_delete_proto = {
+	.func = bpf_task_storage_delete,
+	.gpl_only = false,
+	.ret_type = RET_INTEGER,
+	.arg1_type = ARG_CONST_MAP_PTR,
+	.arg2_type = ARG_PTR_TO_BTF_ID,
+	.arg2_btf_id = &bpf_task_storage_btf_ids[0],
+};
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index 8f50c9c19f1b..f3fe9f53f93c 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -773,7 +773,8 @@ static int map_check_btf(struct bpf_map *map, const struct btf *btf,
 		    map->map_type != BPF_MAP_TYPE_ARRAY &&
 		    map->map_type != BPF_MAP_TYPE_CGROUP_STORAGE &&
 		    map->map_type != BPF_MAP_TYPE_SK_STORAGE &&
-		    map->map_type != BPF_MAP_TYPE_INODE_STORAGE)
+		    map->map_type != BPF_MAP_TYPE_INODE_STORAGE &&
+		    map->map_type != BPF_MAP_TYPE_TASK_STORAGE)
 			return -ENOTSUPP;
 		if (map->spin_lock_off + sizeof(struct bpf_spin_lock) >
 		    map->value_size) {
diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 6200519582a6..b0790876694f 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -4469,6 +4469,11 @@ static int check_map_func_compatibility(struct bpf_verifier_env *env,
 		    func_id != BPF_FUNC_inode_storage_delete)
 			goto error;
 		break;
+	case BPF_MAP_TYPE_TASK_STORAGE:
+		if (func_id != BPF_FUNC_task_storage_get &&
+		    func_id != BPF_FUNC_task_storage_delete)
+			goto error;
+		break;
 	default:
 		break;
 	}
@@ -4547,6 +4552,11 @@ static int check_map_func_compatibility(struct bpf_verifier_env *env,
 		if (map->map_type != BPF_MAP_TYPE_INODE_STORAGE)
 			goto error;
 		break;
+	case BPF_FUNC_task_storage_get:
+	case BPF_FUNC_task_storage_delete:
+		if (map->map_type != BPF_MAP_TYPE_TASK_STORAGE)
+			goto error;
+		break;
 	default:
 		break;
 	}
diff --git a/security/bpf/hooks.c b/security/bpf/hooks.c
index 788667d582ae..e5971fa74fd7 100644
--- a/security/bpf/hooks.c
+++ b/security/bpf/hooks.c
@@ -12,6 +12,7 @@ static struct security_hook_list bpf_lsm_hooks[] __lsm_ro_after_init = {
 	#include <linux/lsm_hook_defs.h>
 	#undef LSM_HOOK
 	LSM_HOOK_INIT(inode_free_security, bpf_inode_storage_free),
+	LSM_HOOK_INIT(task_free, bpf_task_storage_free),
 };
 
 static int __init bpf_lsm_init(void)
@@ -23,6 +24,7 @@ static int __init bpf_lsm_init(void)
 
 struct lsm_blob_sizes bpf_lsm_blob_sizes __lsm_ro_after_init = {
 	.lbs_inode = sizeof(struct bpf_storage_blob),
+	.lbs_task = sizeof(struct bpf_storage_blob),
 };
 
 DEFINE_LSM(bpf) = {
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index e6ceac3f7d62..f4037b2161a6 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -157,6 +157,7 @@ enum bpf_map_type {
 	BPF_MAP_TYPE_STRUCT_OPS,
 	BPF_MAP_TYPE_RINGBUF,
 	BPF_MAP_TYPE_INODE_STORAGE,
+	BPF_MAP_TYPE_TASK_STORAGE,
 };
 
 /* Note that tracing related programs such as
@@ -3742,6 +3743,42 @@ union bpf_attr {
  * 	Return
  * 		The helper returns **TC_ACT_REDIRECT** on success or
  * 		**TC_ACT_SHOT** on error.
+ *
+ * void *bpf_task_storage_get(struct bpf_map *map, struct task_struct *task, void *value, u64 flags)
+ *	Description
+ *		Get a bpf_local_storage from the *task*.
+ *
+ *		Logically, it could be thought of as getting the value from
+ *		a *map* with *task* as the **key**.  From this
+ *		perspective,  the usage is not much different from
+ *		**bpf_map_lookup_elem**\ (*map*, **&**\ *task*) except this
+ *		helper enforces the key must be an task_struct and the map must also
+ *		be a **BPF_MAP_TYPE_TASK_STORAGE**.
+ *
+ *		Underneath, the value is stored locally at *task* instead of
+ *		the *map*.  The *map* is used as the bpf-local-storage
+ *		"type". The bpf-local-storage "type" (i.e. the *map*) is
+ *		searched against all bpf_local_storage residing at *task*.
+ *
+ *		An optional *flags* (**BPF_LOCAL_STORAGE_GET_F_CREATE**) can be
+ *		used such that a new bpf_local_storage will be
+ *		created if one does not exist.  *value* can be used
+ *		together with **BPF_LOCAL_STORAGE_GET_F_CREATE** to specify
+ *		the initial value of a bpf_local_storage.  If *value* is
+ *		**NULL**, the new bpf_local_storage will be zero initialized.
+ *	Return
+ *		A bpf_local_storage pointer is returned on success.
+ *
+ *		**NULL** if not found or there was an error in adding
+ *		a new bpf_local_storage.
+ *
+ * long bpf_task_storage_delete(struct bpf_map *map, struct task_struct *task)
+ *	Description
+ *		Delete a bpf_local_storage from a *task*.
+ *	Return
+ *		0 on success.
+ *
+ *		**-ENOENT** if the bpf_local_storage cannot be found.
  */
 #define __BPF_FUNC_MAPPER(FN)		\
 	FN(unspec),			\
@@ -3900,6 +3937,8 @@ union bpf_attr {
 	FN(bpf_per_cpu_ptr),            \
 	FN(bpf_this_cpu_ptr),		\
 	FN(redirect_peer),		\
+	FN(task_storage_get),		\
+	FN(task_storage_delete),	\
 	/* */
 
 /* integer value in 'imm' field of BPF_CALL instruction selects which helper
-- 
2.29.1.341.ge80a0c044ae-goog

[PATCH bpf-next v2 2/8] libbpf: Add support for task local storage

[KP Singh]

From: KP Singh <kpsingh@google.com>

Signed-off-by: KP Singh <kpsingh@google.com>
---
 tools/lib/bpf/libbpf_probes.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/tools/lib/bpf/libbpf_probes.c b/tools/lib/bpf/libbpf_probes.c
index 5482a9b7ae2d..bed00ca194f0 100644
--- a/tools/lib/bpf/libbpf_probes.c
+++ b/tools/lib/bpf/libbpf_probes.c
@@ -1,6 +1,7 @@
 // SPDX-License-Identifier: (LGPL-2.1 OR BSD-2-Clause)
 /* Copyright (c) 2019 Netronome Systems, Inc. */
 
+#include "linux/bpf.h"
 #include <errno.h>
 #include <fcntl.h>
 #include <string.h>
@@ -230,6 +231,7 @@ bool bpf_probe_map_type(enum bpf_map_type map_type, __u32 ifindex)
 		break;
 	case BPF_MAP_TYPE_SK_STORAGE:
 	case BPF_MAP_TYPE_INODE_STORAGE:
+	case BPF_MAP_TYPE_TASK_STORAGE:
 		btf_key_type_id = 1;
 		btf_value_type_id = 3;
 		value_size = 8;
-- 
2.29.1.341.ge80a0c044ae-goog

[PATCH bpf-next v2 3/8] bpftool: Add support for task local storage

[KP Singh]

From: KP Singh <kpsingh@google.com>

Signed-off-by: KP Singh <kpsingh@google.com>
---
 tools/bpf/bpftool/Documentation/bpftool-map.rst | 3 ++-
 tools/bpf/bpftool/bash-completion/bpftool       | 2 +-
 tools/bpf/bpftool/map.c                         | 4 +++-
 3 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/tools/bpf/bpftool/Documentation/bpftool-map.rst b/tools/bpf/bpftool/Documentation/bpftool-map.rst
index dade10cdf295..3d52256ba75f 100644
--- a/tools/bpf/bpftool/Documentation/bpftool-map.rst
+++ b/tools/bpf/bpftool/Documentation/bpftool-map.rst
@@ -50,7 +50,8 @@ MAP COMMANDS
 |		| **lru_percpu_hash** | **lpm_trie** | **array_of_maps** | **hash_of_maps**
 |		| **devmap** | **devmap_hash** | **sockmap** | **cpumap** | **xskmap** | **sockhash**
 |		| **cgroup_storage** | **reuseport_sockarray** | **percpu_cgroup_storage**
-|		| **queue** | **stack** | **sk_storage** | **struct_ops** | **ringbuf** | **inode_storage** }
+|		| **queue** | **stack** | **sk_storage** | **struct_ops** | **ringbuf** | **inode_storage**
+		| **task_storage** }
 
 DESCRIPTION
 ===========
diff --git a/tools/bpf/bpftool/bash-completion/bpftool b/tools/bpf/bpftool/bash-completion/bpftool
index 3f1da30c4da6..fdffbc64c65c 100644
--- a/tools/bpf/bpftool/bash-completion/bpftool
+++ b/tools/bpf/bpftool/bash-completion/bpftool
@@ -705,7 +705,7 @@ _bpftool()
                                 hash_of_maps devmap devmap_hash sockmap cpumap \
                                 xskmap sockhash cgroup_storage reuseport_sockarray \
                                 percpu_cgroup_storage queue stack sk_storage \
-                                struct_ops inode_storage' -- \
+                                struct_ops inode_storage task_storage' -- \
                                                    "$cur" ) )
                             return 0
                             ;;
diff --git a/tools/bpf/bpftool/map.c b/tools/bpf/bpftool/map.c
index a7efbd84fbcc..b400364ee054 100644
--- a/tools/bpf/bpftool/map.c
+++ b/tools/bpf/bpftool/map.c
@@ -51,6 +51,7 @@ const char * const map_type_name[] = {
 	[BPF_MAP_TYPE_STRUCT_OPS]		= "struct_ops",
 	[BPF_MAP_TYPE_RINGBUF]			= "ringbuf",
 	[BPF_MAP_TYPE_INODE_STORAGE]		= "inode_storage",
+	[BPF_MAP_TYPE_TASK_STORAGE]		= "task_storage",
 };
 
 const size_t map_type_name_size = ARRAY_SIZE(map_type_name);
@@ -1464,7 +1465,8 @@ static int do_help(int argc, char **argv)
 		"                 lru_percpu_hash | lpm_trie | array_of_maps | hash_of_maps |\n"
 		"                 devmap | devmap_hash | sockmap | cpumap | xskmap | sockhash |\n"
 		"                 cgroup_storage | reuseport_sockarray | percpu_cgroup_storage |\n"
-		"                 queue | stack | sk_storage | struct_ops | ringbuf | inode_storage }\n"
+		"                 queue | stack | sk_storage | struct_ops | ringbuf | inode_storage |\n"
+		"		  task_storage }\n"
 		"       " HELP_SPEC_OPTIONS "\n"
 		"",
 		bin_name, argv[-2]);
-- 
2.29.1.341.ge80a0c044ae-goog

[PATCH bpf-next v2 4/8] bpf: Implement get_current_task_btf and RET_PTR_TO_BTF_ID

[KP Singh]

From: KP Singh <kpsingh@google.com>

The currently available bpf_get_current_task returns an unsigned integer
which can be used along with BPF_CORE_READ to read data from
the task_struct but still cannot be used as an input argument to a
helper that accepts an ARG_PTR_TO_BTF_ID of type task_struct.

In order to implement this helper a new return type, RET_PTR_TO_BTF_ID,
is added. This is similar to RET_PTR_TO_BTF_ID_OR_NULL but does not
require checking the nullness of returned pointer.

Signed-off-by: KP Singh <kpsingh@google.com>
---
 include/linux/bpf.h            |  1 +
 include/uapi/linux/bpf.h       |  9 +++++++++
 kernel/bpf/verifier.c          |  7 +++++--
 kernel/trace/bpf_trace.c       | 16 ++++++++++++++++
 tools/include/uapi/linux/bpf.h |  9 +++++++++
 5 files changed, 40 insertions(+), 2 deletions(-)

diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index 2fffd30e13ac..73d5381a5d5c 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -310,6 +310,7 @@ enum bpf_return_type {
 	RET_PTR_TO_BTF_ID_OR_NULL,	/* returns a pointer to a btf_id or NULL */
 	RET_PTR_TO_MEM_OR_BTF_ID_OR_NULL, /* returns a pointer to a valid memory or a btf_id or NULL */
 	RET_PTR_TO_MEM_OR_BTF_ID,	/* returns a pointer to a valid memory or a btf_id */
+	RET_PTR_TO_BTF_ID,		/* returns a pointer to a btf_id */
 };
 
 /* eBPF function prototype used by verifier to allow BPF_CALLs from eBPF programs
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index f4037b2161a6..9879d6793e90 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -3779,6 +3779,14 @@ union bpf_attr {
  *		0 on success.
  *
  *		**-ENOENT** if the bpf_local_storage cannot be found.
+ *
+ * struct task_struct *bpf_get_current_task_btf(void)
+ *	Description
+ *		Return a BTF pointer to the "current" task.
+ *		This pointer can also be used in helpers that accept an
+ *		*ARG_PTR_TO_BTF_ID* of type *task_struct*.
+ *	Return
+ *		Pointer to the current task.
  */
 #define __BPF_FUNC_MAPPER(FN)		\
 	FN(unspec),			\
@@ -3939,6 +3947,7 @@ union bpf_attr {
 	FN(redirect_peer),		\
 	FN(task_storage_get),		\
 	FN(task_storage_delete),	\
+	FN(get_current_task_btf),	\
 	/* */
 
 /* integer value in 'imm' field of BPF_CALL instruction selects which helper
diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index b0790876694f..314018e8fc12 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -5186,11 +5186,14 @@ static int check_helper_call(struct bpf_verifier_env *env, int func_id, int insn
 				PTR_TO_BTF_ID : PTR_TO_BTF_ID_OR_NULL;
 			regs[BPF_REG_0].btf_id = meta.ret_btf_id;
 		}
-	} else if (fn->ret_type == RET_PTR_TO_BTF_ID_OR_NULL) {
+	} else if (fn->ret_type == RET_PTR_TO_BTF_ID_OR_NULL ||
+		   fn->ret_type == RET_PTR_TO_BTF_ID) {
 		int ret_btf_id;
 
 		mark_reg_known_zero(env, regs, BPF_REG_0);
-		regs[BPF_REG_0].type = PTR_TO_BTF_ID_OR_NULL;
+		regs[BPF_REG_0].type = fn->ret_type == RET_PTR_TO_BTF_ID ?
+						     PTR_TO_BTF_ID :
+						     PTR_TO_BTF_ID_OR_NULL;
 		ret_btf_id = *fn->ret_btf_id;
 		if (ret_btf_id == 0) {
 			verbose(env, "invalid return type %d of func %s#%d\n",
diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
index 4517c8b66518..e4515b0f62a8 100644
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -1022,6 +1022,20 @@ const struct bpf_func_proto bpf_get_current_task_proto = {
 	.ret_type	= RET_INTEGER,
 };
 
+BPF_CALL_0(bpf_get_current_task_btf)
+{
+	return (unsigned long) current;
+}
+
+BTF_ID_LIST_SINGLE(bpf_get_current_btf_ids, struct, task_struct)
+
+static const struct bpf_func_proto bpf_get_current_task_btf_proto = {
+	.func		= bpf_get_current_task_btf,
+	.gpl_only	= true,
+	.ret_type	= RET_PTR_TO_BTF_ID,
+	.ret_btf_id	= &bpf_get_current_btf_ids[0],
+};
+
 BPF_CALL_2(bpf_current_task_under_cgroup, struct bpf_map *, map, u32, idx)
 {
 	struct bpf_array *array = container_of(map, struct bpf_array, map);
@@ -1265,6 +1279,8 @@ bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 		return &bpf_get_current_pid_tgid_proto;
 	case BPF_FUNC_get_current_task:
 		return &bpf_get_current_task_proto;
+	case BPF_FUNC_get_current_task_btf:
+		return &bpf_get_current_task_btf_proto;
 	case BPF_FUNC_get_current_uid_gid:
 		return &bpf_get_current_uid_gid_proto;
 	case BPF_FUNC_get_current_comm:
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index f4037b2161a6..9879d6793e90 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -3779,6 +3779,14 @@ union bpf_attr {
  *		0 on success.
  *
  *		**-ENOENT** if the bpf_local_storage cannot be found.
+ *
+ * struct task_struct *bpf_get_current_task_btf(void)
+ *	Description
+ *		Return a BTF pointer to the "current" task.
+ *		This pointer can also be used in helpers that accept an
+ *		*ARG_PTR_TO_BTF_ID* of type *task_struct*.
+ *	Return
+ *		Pointer to the current task.
  */
 #define __BPF_FUNC_MAPPER(FN)		\
 	FN(unspec),			\
@@ -3939,6 +3947,7 @@ union bpf_attr {
 	FN(redirect_peer),		\
 	FN(task_storage_get),		\
 	FN(task_storage_delete),	\
+	FN(get_current_task_btf),	\
 	/* */
 
 /* integer value in 'imm' field of BPF_CALL instruction selects which helper
-- 
2.29.1.341.ge80a0c044ae-goog

[PATCH bpf-next v2 5/8] bpf: Fix tests for local_storage

[KP Singh]

From: KP Singh <kpsingh@google.com>

The {inode,sk}_storage_result checking if the correct value was retrieved
was being clobbered unconditionally by the return value of the
bpf_{inode,sk}_storage_delete call.

Also, consistently use the newly added BPF_LOCAL_STORAGE_GET_F_CREATE
flag.

Fixes: cd324d7abb3d ("bpf: Add selftests for local_storage")
Signed-off-by: KP Singh <kpsingh@google.com>
---
 .../selftests/bpf/progs/local_storage.c       | 24 ++++++++++++-------
 1 file changed, 15 insertions(+), 9 deletions(-)

diff --git a/tools/testing/selftests/bpf/progs/local_storage.c b/tools/testing/selftests/bpf/progs/local_storage.c
index 0758ba229ae0..09529e33be98 100644
--- a/tools/testing/selftests/bpf/progs/local_storage.c
+++ b/tools/testing/selftests/bpf/progs/local_storage.c
@@ -58,20 +58,22 @@ int BPF_PROG(unlink_hook, struct inode *dir, struct dentry *victim)
 {
 	__u32 pid = bpf_get_current_pid_tgid() >> 32;
 	struct dummy_storage *storage;
+	int err;
 
 	if (pid != monitored_pid)
 		return 0;
 
 	storage = bpf_inode_storage_get(&inode_storage_map, victim->d_inode, 0,
-				     BPF_SK_STORAGE_GET_F_CREATE);
+					BPF_LOCAL_STORAGE_GET_F_CREATE);
 	if (!storage)
 		return 0;
 
-	if (storage->value == DUMMY_STORAGE_VALUE)
+	if (storage->value != DUMMY_STORAGE_VALUE)
 		inode_storage_result = -1;
 
-	inode_storage_result =
-		bpf_inode_storage_delete(&inode_storage_map, victim->d_inode);
+	err = bpf_inode_storage_delete(&inode_storage_map, victim->d_inode);
+	if (!err)
+		inode_storage_result = err;
 
 	return 0;
 }
@@ -82,19 +84,23 @@ int BPF_PROG(socket_bind, struct socket *sock, struct sockaddr *address,
 {
 	__u32 pid = bpf_get_current_pid_tgid() >> 32;
 	struct dummy_storage *storage;
+	int err;
 
 	if (pid != monitored_pid)
 		return 0;
 
 	storage = bpf_sk_storage_get(&sk_storage_map, sock->sk, 0,
-				     BPF_SK_STORAGE_GET_F_CREATE);
+				     BPF_LOCAL_STORAGE_GET_F_CREATE);
 	if (!storage)
 		return 0;
 
-	if (storage->value == DUMMY_STORAGE_VALUE)
+	if (storage->value != DUMMY_STORAGE_VALUE)
 		sk_storage_result = -1;
 
-	sk_storage_result = bpf_sk_storage_delete(&sk_storage_map, sock->sk);
+	err = bpf_sk_storage_delete(&sk_storage_map, sock->sk);
+	if (!err)
+		sk_storage_result = err;
+
 	return 0;
 }
 
@@ -109,7 +115,7 @@ int BPF_PROG(socket_post_create, struct socket *sock, int family, int type,
 		return 0;
 
 	storage = bpf_sk_storage_get(&sk_storage_map, sock->sk, 0,
-				     BPF_SK_STORAGE_GET_F_CREATE);
+				     BPF_LOCAL_STORAGE_GET_F_CREATE);
 	if (!storage)
 		return 0;
 
@@ -131,7 +137,7 @@ int BPF_PROG(file_open, struct file *file)
 		return 0;
 
 	storage = bpf_inode_storage_get(&inode_storage_map, file->f_inode, 0,
-				     BPF_LOCAL_STORAGE_GET_F_CREATE);
+					BPF_LOCAL_STORAGE_GET_F_CREATE);
 	if (!storage)
 		return 0;
 
-- 
2.29.1.341.ge80a0c044ae-goog

[PATCH bpf-next v2 6/8] bpf: Update selftests for local_storage to use vmlinux.h

[KP Singh]

From: KP Singh <kpsingh@google.com>

With the fixing of BTF pruning of embedded types being fixed, the test
can be simplified to use vmlinux.h

Signed-off-by: KP Singh <kpsingh@google.com>
---
 .../selftests/bpf/progs/local_storage.c       | 20 +------------------
 1 file changed, 1 insertion(+), 19 deletions(-)

diff --git a/tools/testing/selftests/bpf/progs/local_storage.c b/tools/testing/selftests/bpf/progs/local_storage.c
index 09529e33be98..ef3822bc7542 100644
--- a/tools/testing/selftests/bpf/progs/local_storage.c
+++ b/tools/testing/selftests/bpf/progs/local_storage.c
@@ -4,9 +4,8 @@
  * Copyright 2020 Google LLC.
  */
 
+#include "vmlinux.h"
 #include <errno.h>
-#include <linux/bpf.h>
-#include <stdbool.h>
 #include <bpf/bpf_helpers.h>
 #include <bpf/bpf_tracing.h>
 
@@ -36,23 +35,6 @@ struct {
 	__type(value, struct dummy_storage);
 } sk_storage_map SEC(".maps");
 
-/* TODO Use vmlinux.h once BTF pruning for embedded types is fixed.
- */
-struct sock {} __attribute__((preserve_access_index));
-struct sockaddr {} __attribute__((preserve_access_index));
-struct socket {
-	struct sock *sk;
-} __attribute__((preserve_access_index));
-
-struct inode {} __attribute__((preserve_access_index));
-struct dentry {
-	struct inode *d_inode;
-} __attribute__((preserve_access_index));
-struct file {
-	struct inode *f_inode;
-} __attribute__((preserve_access_index));
-
-
 SEC("lsm/inode_unlink")
 int BPF_PROG(unlink_hook, struct inode *dir, struct dentry *victim)
 {
-- 
2.29.1.341.ge80a0c044ae-goog

[PATCH bpf-next v2 7/8] bpf: Add tests for task_local_storage

[KP Singh]

From: KP Singh <kpsingh@google.com>

The test exercises the syscall based map operations by creating a pidfd
for the current process.

For verifying kernel / LSM functionality, the test implements a simple
MAC policy which denies an executable from unlinking itself. The LSM
program bprm_committed_creds sets a task_local_storage with a pointer to
the inode. This is then used to detect if the task is trying to unlink
itself in the inode_unlink LSM hook.

The test copies /bin/rm to /tmp and executes it in a child thread with
the intention of deleting itself. A successful test should prevent the
the running executable from deleting itself.

The temporary file is cleaned up later in the test.

Signed-off-by: KP Singh <kpsingh@google.com>
---
 .../bpf/prog_tests/test_local_storage.c       | 167 ++++++++++++++++--
 .../selftests/bpf/progs/local_storage.c       |  45 ++++-
 2 files changed, 194 insertions(+), 18 deletions(-)

diff --git a/tools/testing/selftests/bpf/prog_tests/test_local_storage.c b/tools/testing/selftests/bpf/prog_tests/test_local_storage.c
index 91cd6f357246..feba23f8848b 100644
--- a/tools/testing/selftests/bpf/prog_tests/test_local_storage.c
+++ b/tools/testing/selftests/bpf/prog_tests/test_local_storage.c
@@ -4,30 +4,149 @@
  * Copyright (C) 2020 Google LLC.
  */
 
+#define _GNU_SOURCE
+
+#include <asm-generic/errno-base.h>
+#include <unistd.h>
+#include <sys/stat.h>
 #include <test_progs.h>
 #include <linux/limits.h>
 
 #include "local_storage.skel.h"
 #include "network_helpers.h"
 
-int create_and_unlink_file(void)
+static inline int sys_pidfd_open(pid_t pid, unsigned int flags)
+{
+	return syscall(__NR_pidfd_open, pid, flags);
+}
+
+unsigned int duration;
+
+#define TEST_STORAGE_VALUE 0xbeefdead
+
+struct storage {
+	void *inode;
+	unsigned int value;
+	/* Lock ensures that spin locked versions of local stoage operations
+	 * also work, most operations in this tests are still single threaded
+	 */
+	struct bpf_spin_lock lock;
+};
+
+/* Copies an rm binary to a temp file. dest is a mkstemp template */
+int copy_rm(char *dest)
 {
-	char fname[PATH_MAX] = "/tmp/fileXXXXXX";
-	int fd;
+	int ret, fd_in, fd_out;
+	struct stat stat;
 
-	fd = mkstemp(fname);
-	if (fd < 0)
-		return fd;
+	fd_in = open("/bin/rm", O_RDONLY);
+	if (fd_in < 0)
+		return fd_in;
 
-	close(fd);
-	unlink(fname);
+	fd_out = mkstemp(dest);
+	if (fd_out < 0)
+		return fd_out;
+
+	ret = fstat(fd_in, &stat);
+	if (ret == -1)
+		return errno;
+
+	ret = copy_file_range(fd_in, NULL, fd_out, NULL, stat.st_size, 0);
+	if (ret == -1)
+		return errno;
+
+	/* Set executable permission on the copied file */
+	ret = chmod(dest, 0100);
+	if (ret == -1)
+		return errno;
+
+	close(fd_in);
+	close(fd_out);
 	return 0;
 }
 
+/* Fork and exec the provided rm binary and return the exit code of the
+ * forked process and its pid.
+ */
+int run_self_unlink(int *monitored_pid, const char *rm_path)
+{
+	int child_pid, child_status, ret;
+	int null_fd;
+
+	child_pid = fork();
+	if (child_pid == 0) {
+		null_fd = open("/dev/null", O_WRONLY);
+		dup2(null_fd, STDOUT_FILENO);
+		dup2(null_fd, STDERR_FILENO);
+		close(null_fd);
+
+		*monitored_pid = getpid();
+		/* Use the copied /usr/bin/rm to delete itself
+		 * /tmp/copy_of_rm /tmp/copy_of_rm.
+		 */
+		ret = execlp(rm_path, rm_path, rm_path, NULL);
+		if (ret)
+			exit(errno);
+	} else if (child_pid > 0) {
+		waitpid(child_pid, &child_status, 0);
+		return WEXITSTATUS(child_status);
+	}
+
+	return -EINVAL;
+}
+
+bool check_syscall_operations(int map_fd, int obj_fd)
+{
+	struct storage val = { .value = TEST_STORAGE_VALUE, .lock = { 0 } },
+		       lookup_val = { .value = 0, .lock = { 0 } };
+	int err;
+
+	/* Looking up an existing element should fail initially */
+	err = bpf_map_lookup_elem_flags(map_fd, &obj_fd, &lookup_val,
+					BPF_F_LOCK);
+	if (CHECK(!err || errno != ENOENT, "bpf_map_lookup_elem",
+		  "err:%d errno:%d\n", err, errno))
+		return false;
+
+	/* Create a new element */
+	err = bpf_map_update_elem(map_fd, &obj_fd, &val,
+				  BPF_NOEXIST | BPF_F_LOCK);
+	if (CHECK(err < 0, "bpf_map_update_elem", "err:%d errno:%d\n", err,
+		  errno))
+		return false;
+
+	/* Lookup the newly created element */
+	err = bpf_map_lookup_elem_flags(map_fd, &obj_fd, &lookup_val,
+					BPF_F_LOCK);
+	if (CHECK(err < 0, "bpf_map_lookup_elem", "err:%d errno:%d", err,
+		  errno))
+		return false;
+
+	/* Check the value of the newly created element */
+	if (CHECK(lookup_val.value != val.value, "bpf_map_lookup_elem",
+		  "value got = %x errno:%d", lookup_val.value, val.value))
+		return false;
+
+	err = bpf_map_delete_elem(map_fd, &obj_fd);
+	if (CHECK(err, "bpf_map_delete_elem()", "err:%d errno:%d\n", err,
+		  errno))
+		return false;
+
+	/* The lookup should fail, now that the element has been deleted */
+	err = bpf_map_lookup_elem_flags(map_fd, &obj_fd, &lookup_val,
+					BPF_F_LOCK);
+	if (CHECK(!err || errno != ENOENT, "bpf_map_lookup_elem",
+		  "err:%d errno:%d\n", err, errno))
+		return false;
+
+	return true;
+}
+
 void test_test_local_storage(void)
 {
+	char tmp_exec_path[PATH_MAX] = "/tmp/copy_of_rmXXXXXX";
+	int err, serv_sk = -1, task_fd = -1;
 	struct local_storage *skel = NULL;
-	int err, duration = 0, serv_sk = -1;
 
 	skel = local_storage__open_and_load();
 	if (CHECK(!skel, "skel_load", "lsm skeleton failed\n"))
@@ -37,10 +156,35 @@ void test_test_local_storage(void)
 	if (CHECK(err, "attach", "lsm attach failed: %d\n", err))
 		goto close_prog;
 
+	task_fd = sys_pidfd_open(getpid(), 0);
+	if (CHECK(task_fd < 0, "pidfd_open",
+		  "failed to get pidfd err:%d, errno:%d", task_fd, errno))
+		goto close_prog;
+
+	if (!check_syscall_operations(bpf_map__fd(skel->maps.task_storage_map),
+				      task_fd))
+		goto close_prog;
+
+	err = copy_rm(tmp_exec_path);
+	if (CHECK(err < 0, "copy_rm", "err %d errno %d\n", err, errno))
+		goto close_prog;
+
+	/* Sets skel->bss->monitored_pid to the pid of the forked child
+	 * forks a child process that executes tmp_exec_path and tries to
+	 * unlink its executable. This operation should be denied by the loaded
+	 * LSM program.
+	 */
+	err = run_self_unlink(&skel->bss->monitored_pid, tmp_exec_path);
+	if (CHECK(err != EPERM, "run_self_unlink", "err %d want EPERM\n", err))
+		goto close_prog;
+
+	/* Set the process being monitored to be the current process */
 	skel->bss->monitored_pid = getpid();
 
-	err = create_and_unlink_file();
-	if (CHECK(err < 0, "exec_cmd", "err %d errno %d\n", err, errno))
+	/* Remove the temporary created executable */
+	err = unlink(tmp_exec_path);
+	if (CHECK(err != 0, "unlink", "unable to unlink %s: %d", tmp_exec_path,
+		  errno))
 		goto close_prog;
 
 	CHECK(skel->data->inode_storage_result != 0, "inode_storage_result",
@@ -56,5 +200,6 @@ void test_test_local_storage(void)
 	close(serv_sk);
 
 close_prog:
+	close(task_fd);
 	local_storage__destroy(skel);
 }
diff --git a/tools/testing/selftests/bpf/progs/local_storage.c b/tools/testing/selftests/bpf/progs/local_storage.c
index ef3822bc7542..a4979982ce80 100644
--- a/tools/testing/selftests/bpf/progs/local_storage.c
+++ b/tools/testing/selftests/bpf/progs/local_storage.c
@@ -17,34 +17,50 @@ int monitored_pid = 0;
 int inode_storage_result = -1;
 int sk_storage_result = -1;
 
-struct dummy_storage {
+struct local_storage {
+	struct inode *exec_inode;
 	__u32 value;
+	struct bpf_spin_lock lock;
 };
 
 struct {
 	__uint(type, BPF_MAP_TYPE_INODE_STORAGE);
 	__uint(map_flags, BPF_F_NO_PREALLOC);
 	__type(key, int);
-	__type(value, struct dummy_storage);
+	__type(value, struct local_storage);
 } inode_storage_map SEC(".maps");
 
 struct {
 	__uint(type, BPF_MAP_TYPE_SK_STORAGE);
 	__uint(map_flags, BPF_F_NO_PREALLOC | BPF_F_CLONE);
 	__type(key, int);
-	__type(value, struct dummy_storage);
+	__type(value, struct local_storage);
 } sk_storage_map SEC(".maps");
 
+struct {
+	__uint(type, BPF_MAP_TYPE_TASK_STORAGE);
+	__uint(map_flags, BPF_F_NO_PREALLOC);
+	__type(key, int);
+	__type(value, struct local_storage);
+} task_storage_map SEC(".maps");
+
 SEC("lsm/inode_unlink")
 int BPF_PROG(unlink_hook, struct inode *dir, struct dentry *victim)
 {
 	__u32 pid = bpf_get_current_pid_tgid() >> 32;
-	struct dummy_storage *storage;
+	struct local_storage *storage;
 	int err;
 
 	if (pid != monitored_pid)
 		return 0;
 
+	storage = bpf_task_storage_get(&task_storage_map,
+				       bpf_get_current_task_btf(), 0, 0);
+
+	/* Don't let an executable delete itself */
+	if (storage && storage->exec_inode == victim->d_inode)
+		return -EPERM;
+
 	storage = bpf_inode_storage_get(&inode_storage_map, victim->d_inode, 0,
 					BPF_LOCAL_STORAGE_GET_F_CREATE);
 	if (!storage)
@@ -65,7 +81,7 @@ int BPF_PROG(socket_bind, struct socket *sock, struct sockaddr *address,
 	     int addrlen)
 {
 	__u32 pid = bpf_get_current_pid_tgid() >> 32;
-	struct dummy_storage *storage;
+	struct local_storage *storage;
 	int err;
 
 	if (pid != monitored_pid)
@@ -91,7 +107,7 @@ int BPF_PROG(socket_post_create, struct socket *sock, int family, int type,
 	     int protocol, int kern)
 {
 	__u32 pid = bpf_get_current_pid_tgid() >> 32;
-	struct dummy_storage *storage;
+	struct local_storage *storage;
 
 	if (pid != monitored_pid)
 		return 0;
@@ -110,7 +126,7 @@ SEC("lsm/file_open")
 int BPF_PROG(file_open, struct file *file)
 {
 	__u32 pid = bpf_get_current_pid_tgid() >> 32;
-	struct dummy_storage *storage;
+	struct local_storage *storage;
 
 	if (pid != monitored_pid)
 		return 0;
@@ -126,3 +142,18 @@ int BPF_PROG(file_open, struct file *file)
 	storage->value = DUMMY_STORAGE_VALUE;
 	return 0;
 }
+
+/* This uses the local storage to remember the inode of the binary that a
+ * process was originally executing.
+ */
+SEC("lsm/bprm_committed_creds")
+void BPF_PROG(exec, struct linux_binprm *bprm)
+{
+	struct local_storage *storage;
+
+	storage = bpf_task_storage_get(&task_storage_map,
+				       bpf_get_current_task_btf(), 0,
+				       BPF_LOCAL_STORAGE_GET_F_CREATE);
+	if (storage)
+		storage->exec_inode = bprm->file->f_inode;
+}
-- 
2.29.1.341.ge80a0c044ae-goog

[PATCH bpf-next v2 8/8] bpf: Exercise syscall operations for inode and sk storage

[KP Singh]

From: KP Singh <kpsingh@google.com>

Signed-off-by: KP Singh <kpsingh@google.com>
---
 .../bpf/prog_tests/test_local_storage.c          | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)

diff --git a/tools/testing/selftests/bpf/prog_tests/test_local_storage.c b/tools/testing/selftests/bpf/prog_tests/test_local_storage.c
index feba23f8848b..2e64baabb50d 100644
--- a/tools/testing/selftests/bpf/prog_tests/test_local_storage.c
+++ b/tools/testing/selftests/bpf/prog_tests/test_local_storage.c
@@ -145,7 +145,7 @@ bool check_syscall_operations(int map_fd, int obj_fd)
 void test_test_local_storage(void)
 {
 	char tmp_exec_path[PATH_MAX] = "/tmp/copy_of_rmXXXXXX";
-	int err, serv_sk = -1, task_fd = -1;
+	int err, serv_sk = -1, task_fd = -1, rm_fd = -1;
 	struct local_storage *skel = NULL;
 
 	skel = local_storage__open_and_load();
@@ -169,6 +169,15 @@ void test_test_local_storage(void)
 	if (CHECK(err < 0, "copy_rm", "err %d errno %d\n", err, errno))
 		goto close_prog;
 
+	rm_fd = open(tmp_exec_path, O_RDONLY);
+	if (CHECK(rm_fd < 0, "open", "failed to open %s err:%d, errno:%d",
+		  tmp_exec_path, rm_fd, errno))
+		goto close_prog;
+
+	if (!check_syscall_operations(bpf_map__fd(skel->maps.inode_storage_map),
+				      rm_fd))
+		goto close_prog;
+
 	/* Sets skel->bss->monitored_pid to the pid of the forked child
 	 * forks a child process that executes tmp_exec_path and tries to
 	 * unlink its executable. This operation should be denied by the loaded
@@ -197,9 +206,14 @@ void test_test_local_storage(void)
 	CHECK(skel->data->sk_storage_result != 0, "sk_storage_result",
 	      "sk_local_storage not set\n");
 
+	if (!check_syscall_operations(bpf_map__fd(skel->maps.sk_storage_map),
+				      serv_sk))
+		goto close_prog;
+
 	close(serv_sk);
 
 close_prog:
+	close(rm_fd);
 	close(task_fd);
 	local_storage__destroy(skel);
 }
-- 
2.29.1.341.ge80a0c044ae-goog

Re: [PATCH bpf-next v2 7/8] bpf: Add tests for task_local_storage

[Alexei Starovoitov]

On Tue, Nov 03, 2020 at 04:31:31PM +0100, KP Singh wrote:
> +
> +struct storage {
> +	void *inode;
> +	unsigned int value;
> +	/* Lock ensures that spin locked versions of local stoage operations
> +	 * also work, most operations in this tests are still single threaded
> +	 */
> +	struct bpf_spin_lock lock;
> +};

I think it's a good idea to test spin_lock in local_storage,
but it seems the test is not doing it fully.
It's only adding it to the storage, but the program is not accessing it.

Re: [PATCH bpf-next v2 7/8] bpf: Add tests for task_local_storage

[KP Singh]

On Tue, Nov 3, 2020 at 7:47 PM Alexei Starovoitov
<alexei.starovoitov@gmail.com> wrote:
>
> On Tue, Nov 03, 2020 at 04:31:31PM +0100, KP Singh wrote:
> > +
> > +struct storage {
> > +     void *inode;
> > +     unsigned int value;
> > +     /* Lock ensures that spin locked versions of local stoage operations
> > +      * also work, most operations in this tests are still single threaded
> > +      */
> > +     struct bpf_spin_lock lock;
> > +};
>
> I think it's a good idea to test spin_lock in local_storage,
> but it seems the test is not doing it fully.
> It's only adding it to the storage, but the program is not accessing it.

I added it here just to check if the offset calculations (map->spin_lock_off)
are correctly happening for these new maps.

As mentioned in the updates, I do intend to generalize
tools/testing/selftests/bpf/map_tests/sk_storage_map.c which already has
 the threading logic to exercise bpf_spin_lock in storage maps.

Hope this is an okay plan?

Re: [PATCH bpf-next v2 2/8] libbpf: Add support for task local storage

[Andrii Nakryiko]

On Tue, Nov 3, 2020 at 7:34 AM KP Singh <kpsingh@chromium.org> wrote:
>
> From: KP Singh <kpsingh@google.com>
>
> Signed-off-by: KP Singh <kpsingh@google.com>
> ---
>  tools/lib/bpf/libbpf_probes.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/tools/lib/bpf/libbpf_probes.c b/tools/lib/bpf/libbpf_probes.c
> index 5482a9b7ae2d..bed00ca194f0 100644
> --- a/tools/lib/bpf/libbpf_probes.c
> +++ b/tools/lib/bpf/libbpf_probes.c
> @@ -1,6 +1,7 @@
>  // SPDX-License-Identifier: (LGPL-2.1 OR BSD-2-Clause)
>  /* Copyright (c) 2019 Netronome Systems, Inc. */
>
> +#include "linux/bpf.h"

why "", not <>?

>  #include <errno.h>
>  #include <fcntl.h>
>  #include <string.h>
> @@ -230,6 +231,7 @@ bool bpf_probe_map_type(enum bpf_map_type map_type, __u32 ifindex)
>                 break;
>         case BPF_MAP_TYPE_SK_STORAGE:
>         case BPF_MAP_TYPE_INODE_STORAGE:
> +       case BPF_MAP_TYPE_TASK_STORAGE:
>                 btf_key_type_id = 1;
>                 btf_value_type_id = 3;
>                 value_size = 8;
> --
> 2.29.1.341.ge80a0c044ae-goog
>

Re: [PATCH bpf-next v2 2/8] libbpf: Add support for task local storage

[KP Singh]

On Tue, Nov 3, 2020 at 8:28 PM Andrii Nakryiko
<andrii.nakryiko@gmail.com> wrote:
>
> On Tue, Nov 3, 2020 at 7:34 AM KP Singh <kpsingh@chromium.org> wrote:
> >
> > From: KP Singh <kpsingh@google.com>
> >
> > Signed-off-by: KP Singh <kpsingh@google.com>
> > ---
> >  tools/lib/bpf/libbpf_probes.c | 2 ++
> >  1 file changed, 2 insertions(+)
> >
> > diff --git a/tools/lib/bpf/libbpf_probes.c b/tools/lib/bpf/libbpf_probes.c
> > index 5482a9b7ae2d..bed00ca194f0 100644
> > --- a/tools/lib/bpf/libbpf_probes.c
> > +++ b/tools/lib/bpf/libbpf_probes.c
> > @@ -1,6 +1,7 @@
> >  // SPDX-License-Identifier: (LGPL-2.1 OR BSD-2-Clause)
> >  /* Copyright (c) 2019 Netronome Systems, Inc. */
> >
> > +#include "linux/bpf.h"
>
> why "", not <>?

I need to disable this vscode feature where it tries to be oversmart
and adds includes. Fixed.

- KP

Re: [PATCH bpf-next v2 8/8] bpf: Exercise syscall operations for inode and sk storage

[Song Liu]

> On Nov 3, 2020, at 7:31 AM, KP Singh <kpsingh@chromium.org> wrote:
> 
> From: KP Singh <kpsingh@google.com>

A short commit log would be great...

> 
> Signed-off-by: KP Singh <kpsingh@google.com>
> ---
> .../bpf/prog_tests/test_local_storage.c          | 16 +++++++++++++++-
> 1 file changed, 15 insertions(+), 1 deletion(-)
> 
> diff --git a/tools/testing/selftests/bpf/prog_tests/test_local_storage.c b/tools/testing/selftests/bpf/prog_tests/test_local_storage.c
> index feba23f8848b..2e64baabb50d 100644
> --- a/tools/testing/selftests/bpf/prog_tests/test_local_storage.c
> +++ b/tools/testing/selftests/bpf/prog_tests/test_local_storage.c
> @@ -145,7 +145,7 @@ bool check_syscall_operations(int map_fd, int obj_fd)
> void test_test_local_storage(void)
> {
> 	char tmp_exec_path[PATH_MAX] = "/tmp/copy_of_rmXXXXXX";
> -	int err, serv_sk = -1, task_fd = -1;
> +	int err, serv_sk = -1, task_fd = -1, rm_fd = -1;
> 	struct local_storage *skel = NULL;
> 
> 	skel = local_storage__open_and_load();
> @@ -169,6 +169,15 @@ void test_test_local_storage(void)
> 	if (CHECK(err < 0, "copy_rm", "err %d errno %d\n", err, errno))
> 		goto close_prog;
> 
> +	rm_fd = open(tmp_exec_path, O_RDONLY);
> +	if (CHECK(rm_fd < 0, "open", "failed to open %s err:%d, errno:%d",
> +		  tmp_exec_path, rm_fd, errno))
> +		goto close_prog;
> +
> +	if (!check_syscall_operations(bpf_map__fd(skel->maps.inode_storage_map),
> +				      rm_fd))
> +		goto close_prog;
> +
> 	/* Sets skel->bss->monitored_pid to the pid of the forked child
> 	 * forks a child process that executes tmp_exec_path and tries to
> 	 * unlink its executable. This operation should be denied by the loaded
> @@ -197,9 +206,14 @@ void test_test_local_storage(void)
> 	CHECK(skel->data->sk_storage_result != 0, "sk_storage_result",
> 	      "sk_local_storage not set\n");
> 
> +	if (!check_syscall_operations(bpf_map__fd(skel->maps.sk_storage_map),
> +				      serv_sk))
> +		goto close_prog;

We shouldn't need this goto, otherwise we may leak serv_sk. 

> +
> 	close(serv_sk);
> 
> close_prog:
> +	close(rm_fd);
> 	close(task_fd);
> 	local_storage__destroy(skel);
> }
> -- 
> 2.29.1.341.ge80a0c044ae-goog
> 

Re: [PATCH bpf-next v2 8/8] bpf: Exercise syscall operations for inode and sk storage

[KP Singh]

On Tue, Nov 3, 2020 at 11:32 PM Song Liu <songliubraving@fb.com> wrote:
>
>
>
> > On Nov 3, 2020, at 7:31 AM, KP Singh <kpsingh@chromium.org> wrote:
> >
> > From: KP Singh <kpsingh@google.com>
>
> A short commit log would be great...

Sure :) No excuses for not having one, will add it in the next revision.

- KP

[...]

> > +                                   serv_sk))
> > +             goto close_prog;
>
> We shouldn't need this goto, otherwise we may leak serv_sk.

Good point, I will just move the close(serv_sk); along with the other
descriptor clean up.

>
> > +
> >       close(serv_sk);
> >
> > close_prog:
> > +     close(rm_fd);
> >       close(task_fd);
> >       local_storage__destroy(skel);
> > }
> > --
> > 2.29.1.341.ge80a0c044ae-goog
> >
>

Re: [PATCH bpf-next v2 1/8] bpf: Implement task local storage

[Song Liu]

> On Nov 3, 2020, at 7:31 AM, KP Singh <kpsingh@chromium.org> wrote:
> 
> From: KP Singh <kpsingh@google.com>
> 
> Similar to bpf_local_storage for sockets and inodes add local storage
> for task_struct.
> 
> The life-cycle of storage is managed with the life-cycle of the
> task_struct.  i.e. the storage is destroyed along with the owning task
> with a callback to the bpf_task_storage_free from the task_free LSM
> hook.
> 
> The BPF LSM allocates an __rcu pointer to the bpf_local_storage in
> the security blob which are now stackable and can co-exist with other
> LSMs.
> 
> The userspace map operations can be done by using a pid fd as a key
> passed to the lookup, update and delete operations.
> 
> Signed-off-by: KP Singh <kpsingh@google.com>

Acked-by: Song Liu <songliubraving@fb.com>

with a few nits:

> ---
> include/linux/bpf_lsm.h        |  23 +++
> include/linux/bpf_types.h      |   1 +
> include/uapi/linux/bpf.h       |  39 ++++
> kernel/bpf/Makefile            |   1 +
> kernel/bpf/bpf_lsm.c           |   4 +
> kernel/bpf/bpf_task_storage.c  | 313 +++++++++++++++++++++++++++++++++
> kernel/bpf/syscall.c           |   3 +-
> kernel/bpf/verifier.c          |  10 ++
> security/bpf/hooks.c           |   2 +
> tools/include/uapi/linux/bpf.h |  39 ++++
> 10 files changed, 434 insertions(+), 1 deletion(-)
> create mode 100644 kernel/bpf/bpf_task_storage.c
> 
> diff --git a/include/linux/bpf_lsm.h b/include/linux/bpf_lsm.h
> index aaacb6aafc87..326cb68a3632 100644
> --- a/include/linux/bpf_lsm.h
> +++ b/include/linux/bpf_lsm.h
> @@ -7,6 +7,7 @@
> #ifndef _LINUX_BPF_LSM_H
> #define _LINUX_BPF_LSM_H
> 
> +#include "linux/sched.h"

vscode?

> #include <linux/bpf.h>
> #include <linux/lsm_hooks.h>
> 
> @@ -35,9 +36,21 @@ static inline struct bpf_storage_blob *bpf_inode(
> 	return inode->i_security + bpf_lsm_blob_sizes.lbs_inode;
> }

[...]

> index 000000000000..f5ed5eedc532
> --- /dev/null
> +++ b/kernel/bpf/bpf_task_storage.c
> @@ -0,0 +1,313 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Copyright (c) 2019 Facebook

nit: I guess we shouldn't say 2019 Facebook 

> + * Copyright 2020 Google LLC.
> + */
> +
> +#include "linux/pid.h"
> +#include "linux/sched.h"
> +#include <linux/rculist.h>
> +#include <linux/list.h>
> +#include <linux/hash.h>
> +#include <linux/types.h>

[...]

> +}
> +
> +BPF_CALL_2(bpf_task_storage_delete, struct bpf_map *, map, struct task_struct *,
> +	   task)
> +{
> +	/* This helper must only called from where the task is guaranteed
> +	 * to have a refcount and cannot be freed.
> +	 */
> +	return task_storage_delete(task, map);
> +}
> +
> +static int notsupp_get_next_key(struct bpf_map *map, void *key, void *next_key)
> +{
> +	return -ENOTSUPP;
> +}

This is the third copy of notsupp_get_next_key(). We can probably move it to bpf.h. 

[...]

Re: [PATCH bpf-next v2 3/8] bpftool: Add support for task local storage

[Song Liu]

> On Nov 3, 2020, at 7:31 AM, KP Singh <kpsingh@chromium.org> wrote:
> 
> From: KP Singh <kpsingh@google.com>
> 
> Signed-off-by: KP Singh <kpsingh@google.com>

LGTM, except that commit log is missing (also for 2/8). 

Acked-by: Song Liu <songliubraving@fb.com>

> ---
> tools/bpf/bpftool/Documentation/bpftool-map.rst | 3 ++-
> tools/bpf/bpftool/bash-completion/bpftool       | 2 +-
> tools/bpf/bpftool/map.c                         | 4 +++-
> 3 files changed, 6 insertions(+), 3 deletions(-)
> 
> diff --git a/tools/bpf/bpftool/Documentation/bpftool-map.rst b/tools/bpf/bpftool/Documentation/bpftool-map.rst
> index dade10cdf295..3d52256ba75f 100644
> --- a/tools/bpf/bpftool/Documentation/bpftool-map.rst
> +++ b/tools/bpf/bpftool/Documentation/bpftool-map.rst
> @@ -50,7 +50,8 @@ MAP COMMANDS
> |		| **lru_percpu_hash** | **lpm_trie** | **array_of_maps** | **hash_of_maps**
> |		| **devmap** | **devmap_hash** | **sockmap** | **cpumap** | **xskmap** | **sockhash**
> |		| **cgroup_storage** | **reuseport_sockarray** | **percpu_cgroup_storage**
> -|		| **queue** | **stack** | **sk_storage** | **struct_ops** | **ringbuf** | **inode_storage** }
> +|		| **queue** | **stack** | **sk_storage** | **struct_ops** | **ringbuf** | **inode_storage**
> +		| **task_storage** }
> 
> DESCRIPTION
> ===========
> diff --git a/tools/bpf/bpftool/bash-completion/bpftool b/tools/bpf/bpftool/bash-completion/bpftool
> index 3f1da30c4da6..fdffbc64c65c 100644
> --- a/tools/bpf/bpftool/bash-completion/bpftool
> +++ b/tools/bpf/bpftool/bash-completion/bpftool
> @@ -705,7 +705,7 @@ _bpftool()
>                                 hash_of_maps devmap devmap_hash sockmap cpumap \
>                                 xskmap sockhash cgroup_storage reuseport_sockarray \
>                                 percpu_cgroup_storage queue stack sk_storage \
> -                                struct_ops inode_storage' -- \
> +                                struct_ops inode_storage task_storage' -- \
>                                                    "$cur" ) )
>                             return 0
>                             ;;
> diff --git a/tools/bpf/bpftool/map.c b/tools/bpf/bpftool/map.c
> index a7efbd84fbcc..b400364ee054 100644
> --- a/tools/bpf/bpftool/map.c
> +++ b/tools/bpf/bpftool/map.c
> @@ -51,6 +51,7 @@ const char * const map_type_name[] = {
> 	[BPF_MAP_TYPE_STRUCT_OPS]		= "struct_ops",
> 	[BPF_MAP_TYPE_RINGBUF]			= "ringbuf",
> 	[BPF_MAP_TYPE_INODE_STORAGE]		= "inode_storage",
> +	[BPF_MAP_TYPE_TASK_STORAGE]		= "task_storage",
> };
> 
> const size_t map_type_name_size = ARRAY_SIZE(map_type_name);
> @@ -1464,7 +1465,8 @@ static int do_help(int argc, char **argv)
> 		"                 lru_percpu_hash | lpm_trie | array_of_maps | hash_of_maps |\n"
> 		"                 devmap | devmap_hash | sockmap | cpumap | xskmap | sockhash |\n"
> 		"                 cgroup_storage | reuseport_sockarray | percpu_cgroup_storage |\n"
> -		"                 queue | stack | sk_storage | struct_ops | ringbuf | inode_storage }\n"
> +		"                 queue | stack | sk_storage | struct_ops | ringbuf | inode_storage |\n"
> +		"		  task_storage }\n"
> 		"       " HELP_SPEC_OPTIONS "\n"
> 		"",
> 		bin_name, argv[-2]);
> -- 
> 2.29.1.341.ge80a0c044ae-goog
> 

Re: [PATCH bpf-next v2 1/8] bpf: Implement task local storage

[KP Singh]

On Wed, Nov 4, 2020 at 12:47 AM Song Liu <songliubraving@fb.com> wrote:
>
>
> > On Nov 3, 2020, at 7:31 AM, KP Singh <kpsingh@chromium.org> wrote:
> >
> > From: KP Singh <kpsingh@google.com>
> >
> > Similar to bpf_local_storage for sockets and inodes add local storage
> > for task_struct.
> >
> > The life-cycle of storage is managed with the life-cycle of the
> > task_struct.  i.e. the storage is destroyed along with the owning task
> > with a callback to the bpf_task_storage_free from the task_free LSM
> > hook.
> >
> > The BPF LSM allocates an __rcu pointer to the bpf_local_storage in
> > the security blob which are now stackable and can co-exist with other
> > LSMs.
> >
> > The userspace map operations can be done by using a pid fd as a key
> > passed to the lookup, update and delete operations.
> >
> > Signed-off-by: KP Singh <kpsingh@google.com>
>
> Acked-by: Song Liu <songliubraving@fb.com>
>
> with a few nits:
>
> > ---
> > include/linux/bpf_lsm.h        |  23 +++
> > include/linux/bpf_types.h      |   1 +
> > include/uapi/linux/bpf.h       |  39 ++++
> > kernel/bpf/Makefile            |   1 +
> > kernel/bpf/bpf_lsm.c           |   4 +
> > kernel/bpf/bpf_task_storage.c  | 313 +++++++++++++++++++++++++++++++++
> > kernel/bpf/syscall.c           |   3 +-
> > kernel/bpf/verifier.c          |  10 ++
> > security/bpf/hooks.c           |   2 +
> > tools/include/uapi/linux/bpf.h |  39 ++++
> > 10 files changed, 434 insertions(+), 1 deletion(-)
> > create mode 100644 kernel/bpf/bpf_task_storage.c
> >
> > diff --git a/include/linux/bpf_lsm.h b/include/linux/bpf_lsm.h
> > index aaacb6aafc87..326cb68a3632 100644
> > --- a/include/linux/bpf_lsm.h
> > +++ b/include/linux/bpf_lsm.h
> > @@ -7,6 +7,7 @@
> > #ifndef _LINUX_BPF_LSM_H
> > #define _LINUX_BPF_LSM_H
> >
> > +#include "linux/sched.h"
>
> vscode?

Yep, turns out it was clangd (which I use in vscode) and I needed to
pass "--header-insertion=never"
when starting the clangd server, I fixed all other instances as well.

- KP

Re: [PATCH bpf-next v2 4/8] bpf: Implement get_current_task_btf and RET_PTR_TO_BTF_ID

[Song Liu]

> On Nov 3, 2020, at 7:31 AM, KP Singh <kpsingh@chromium.org> wrote:
> 
> From: KP Singh <kpsingh@google.com>
> 
> The currently available bpf_get_current_task returns an unsigned integer
> which can be used along with BPF_CORE_READ to read data from
> the task_struct but still cannot be used as an input argument to a
> helper that accepts an ARG_PTR_TO_BTF_ID of type task_struct.
> 
> In order to implement this helper a new return type, RET_PTR_TO_BTF_ID,
> is added. This is similar to RET_PTR_TO_BTF_ID_OR_NULL but does not
> require checking the nullness of returned pointer.
> 
> Signed-off-by: KP Singh <kpsingh@google.com>

Acked-by: Song Liu <songliubraving@fb.com>

> ---
> include/linux/bpf.h            |  1 +
> include/uapi/linux/bpf.h       |  9 +++++++++
> kernel/bpf/verifier.c          |  7 +++++--
> kernel/trace/bpf_trace.c       | 16 ++++++++++++++++
> tools/include/uapi/linux/bpf.h |  9 +++++++++
> 5 files changed, 40 insertions(+), 2 deletions(-)
> 
> diff --git a/include/linux/bpf.h b/include/linux/bpf.h
> index 2fffd30e13ac..73d5381a5d5c 100644
> --- a/include/linux/bpf.h
> +++ b/include/linux/bpf.h
> @@ -310,6 +310,7 @@ enum bpf_return_type {
> 	RET_PTR_TO_BTF_ID_OR_NULL,	/* returns a pointer to a btf_id or NULL */
> 	RET_PTR_TO_MEM_OR_BTF_ID_OR_NULL, /* returns a pointer to a valid memory or a btf_id or NULL */
> 	RET_PTR_TO_MEM_OR_BTF_ID,	/* returns a pointer to a valid memory or a btf_id */
> +	RET_PTR_TO_BTF_ID,		/* returns a pointer to a btf_id */
> };
> 
> /* eBPF function prototype used by verifier to allow BPF_CALLs from eBPF programs
> diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> index f4037b2161a6..9879d6793e90 100644
> --- a/include/uapi/linux/bpf.h
> +++ b/include/uapi/linux/bpf.h
> @@ -3779,6 +3779,14 @@ union bpf_attr {
>  *		0 on success.
>  *
>  *		**-ENOENT** if the bpf_local_storage cannot be found.
> + *
> + * struct task_struct *bpf_get_current_task_btf(void)
> + *	Description
> + *		Return a BTF pointer to the "current" task.
> + *		This pointer can also be used in helpers that accept an
> + *		*ARG_PTR_TO_BTF_ID* of type *task_struct*.
> + *	Return
> + *		Pointer to the current task.
>  */
> #define __BPF_FUNC_MAPPER(FN)		\
> 	FN(unspec),			\
> @@ -3939,6 +3947,7 @@ union bpf_attr {
> 	FN(redirect_peer),		\
> 	FN(task_storage_get),		\
> 	FN(task_storage_delete),	\
> +	FN(get_current_task_btf),	\
> 	/* */
> 
> /* integer value in 'imm' field of BPF_CALL instruction selects which helper
> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index b0790876694f..314018e8fc12 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c
> @@ -5186,11 +5186,14 @@ static int check_helper_call(struct bpf_verifier_env *env, int func_id, int insn
> 				PTR_TO_BTF_ID : PTR_TO_BTF_ID_OR_NULL;
> 			regs[BPF_REG_0].btf_id = meta.ret_btf_id;
> 		}
> -	} else if (fn->ret_type == RET_PTR_TO_BTF_ID_OR_NULL) {
> +	} else if (fn->ret_type == RET_PTR_TO_BTF_ID_OR_NULL ||
> +		   fn->ret_type == RET_PTR_TO_BTF_ID) {
> 		int ret_btf_id;
> 
> 		mark_reg_known_zero(env, regs, BPF_REG_0);
> -		regs[BPF_REG_0].type = PTR_TO_BTF_ID_OR_NULL;
> +		regs[BPF_REG_0].type = fn->ret_type == RET_PTR_TO_BTF_ID ?
> +						     PTR_TO_BTF_ID :
> +						     PTR_TO_BTF_ID_OR_NULL;
> 		ret_btf_id = *fn->ret_btf_id;
> 		if (ret_btf_id == 0) {
> 			verbose(env, "invalid return type %d of func %s#%d\n",
> diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
> index 4517c8b66518..e4515b0f62a8 100644
> --- a/kernel/trace/bpf_trace.c
> +++ b/kernel/trace/bpf_trace.c
> @@ -1022,6 +1022,20 @@ const struct bpf_func_proto bpf_get_current_task_proto = {
> 	.ret_type	= RET_INTEGER,
> };
> 
> +BPF_CALL_0(bpf_get_current_task_btf)
> +{
> +	return (unsigned long) current;
> +}
> +
> +BTF_ID_LIST_SINGLE(bpf_get_current_btf_ids, struct, task_struct)
> +
> +static const struct bpf_func_proto bpf_get_current_task_btf_proto = {
> +	.func		= bpf_get_current_task_btf,
> +	.gpl_only	= true,
> +	.ret_type	= RET_PTR_TO_BTF_ID,
> +	.ret_btf_id	= &bpf_get_current_btf_ids[0],
> +};
> +
> BPF_CALL_2(bpf_current_task_under_cgroup, struct bpf_map *, map, u32, idx)
> {
> 	struct bpf_array *array = container_of(map, struct bpf_array, map);
> @@ -1265,6 +1279,8 @@ bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
> 		return &bpf_get_current_pid_tgid_proto;
> 	case BPF_FUNC_get_current_task:
> 		return &bpf_get_current_task_proto;
> +	case BPF_FUNC_get_current_task_btf:
> +		return &bpf_get_current_task_btf_proto;
> 	case BPF_FUNC_get_current_uid_gid:
> 		return &bpf_get_current_uid_gid_proto;
> 	case BPF_FUNC_get_current_comm:
> diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
> index f4037b2161a6..9879d6793e90 100644
> --- a/tools/include/uapi/linux/bpf.h
> +++ b/tools/include/uapi/linux/bpf.h
> @@ -3779,6 +3779,14 @@ union bpf_attr {
>  *		0 on success.
>  *
>  *		**-ENOENT** if the bpf_local_storage cannot be found.
> + *
> + * struct task_struct *bpf_get_current_task_btf(void)
> + *	Description
> + *		Return a BTF pointer to the "current" task.
> + *		This pointer can also be used in helpers that accept an
> + *		*ARG_PTR_TO_BTF_ID* of type *task_struct*.
> + *	Return
> + *		Pointer to the current task.
>  */
> #define __BPF_FUNC_MAPPER(FN)		\
> 	FN(unspec),			\
> @@ -3939,6 +3947,7 @@ union bpf_attr {
> 	FN(redirect_peer),		\
> 	FN(task_storage_get),		\
> 	FN(task_storage_delete),	\
> +	FN(get_current_task_btf),	\
> 	/* */
> 
> /* integer value in 'imm' field of BPF_CALL instruction selects which helper
> -- 
> 2.29.1.341.ge80a0c044ae-goog
> 

Re: [PATCH bpf-next v2 7/8] bpf: Add tests for task_local_storage

[KP Singh]

On Tue, Nov 3, 2020 at 7:59 PM KP Singh <kpsingh@chromium.org> wrote:
>
> On Tue, Nov 3, 2020 at 7:47 PM Alexei Starovoitov
> <alexei.starovoitov@gmail.com> wrote:
> >
> > On Tue, Nov 03, 2020 at 04:31:31PM +0100, KP Singh wrote:
> > > +
> > > +struct storage {
> > > +     void *inode;
> > > +     unsigned int value;
> > > +     /* Lock ensures that spin locked versions of local stoage operations
> > > +      * also work, most operations in this tests are still single threaded
> > > +      */
> > > +     struct bpf_spin_lock lock;
> > > +};
> >
> > I think it's a good idea to test spin_lock in local_storage,
> > but it seems the test is not doing it fully.
> > It's only adding it to the storage, but the program is not accessing it.
>
> I added it here just to check if the offset calculations (map->spin_lock_off)
> are correctly happening for these new maps.
>
> As mentioned in the updates, I do intend to generalize
> tools/testing/selftests/bpf/map_tests/sk_storage_map.c which already has
>  the threading logic to exercise bpf_spin_lock in storage maps.
>

Actually, after I added simple bpf_spin_{lock, unlock} to the test programs, I
ended up realizing that we have not exposed spin locks to LSM programs
for now, this is because they inherit the tracing helpers.

I saw the docs mention that these are not exposed to tracing programs due to
insufficient preemption checks. Do you think it would be okay to allow them
for LSM programs?


- KP

> Hope this is an okay plan?

Re: [PATCH bpf-next v2 5/8] bpf: Fix tests for local_storage

[Song Liu]

> On Nov 3, 2020, at 7:31 AM, KP Singh <kpsingh@chromium.org> wrote:
> 
> From: KP Singh <kpsingh@google.com>
> 
> The {inode,sk}_storage_result checking if the correct value was retrieved
> was being clobbered unconditionally by the return value of the
> bpf_{inode,sk}_storage_delete call.
> 
> Also, consistently use the newly added BPF_LOCAL_STORAGE_GET_F_CREATE
> flag.
> 
> Fixes: cd324d7abb3d ("bpf: Add selftests for local_storage")
> Signed-off-by: KP Singh <kpsingh@google.com>

Acked-by: Song Liu <songliubraving@fb.com>

[...]

Re: [PATCH bpf-next v2 6/8] bpf: Update selftests for local_storage to use vmlinux.h

[Song Liu]

> On Nov 3, 2020, at 7:31 AM, KP Singh <kpsingh@chromium.org> wrote:
> 
> From: KP Singh <kpsingh@google.com>
> 
> With the fixing of BTF pruning of embedded types being fixed, the test
> can be simplified to use vmlinux.h
> 
> Signed-off-by: KP Singh <kpsingh@google.com>

Acked-by: Song Liu <songliubraving@fb.com>

> ---
> .../selftests/bpf/progs/local_storage.c       | 20 +------------------
> 1 file changed, 1 insertion(+), 19 deletions(-)
> 
> diff --git a/tools/testing/selftests/bpf/progs/local_storage.c b/tools/testing/selftests/bpf/progs/local_storage.c
> index 09529e33be98..ef3822bc7542 100644
> --- a/tools/testing/selftests/bpf/progs/local_storage.c
> +++ b/tools/testing/selftests/bpf/progs/local_storage.c
> @@ -4,9 +4,8 @@
>  * Copyright 2020 Google LLC.
>  */
> 
> +#include "vmlinux.h"
> #include <errno.h>
> -#include <linux/bpf.h>
> -#include <stdbool.h>
> #include <bpf/bpf_helpers.h>
> #include <bpf/bpf_tracing.h>
> 
> @@ -36,23 +35,6 @@ struct {
> 	__type(value, struct dummy_storage);
> } sk_storage_map SEC(".maps");
> 
> -/* TODO Use vmlinux.h once BTF pruning for embedded types is fixed.
> - */
> -struct sock {} __attribute__((preserve_access_index));
> -struct sockaddr {} __attribute__((preserve_access_index));
> -struct socket {
> -	struct sock *sk;
> -} __attribute__((preserve_access_index));
> -
> -struct inode {} __attribute__((preserve_access_index));
> -struct dentry {
> -	struct inode *d_inode;
> -} __attribute__((preserve_access_index));
> -struct file {
> -	struct inode *f_inode;
> -} __attribute__((preserve_access_index));
> -
> -
> SEC("lsm/inode_unlink")
> int BPF_PROG(unlink_hook, struct inode *dir, struct dentry *victim)
> {
> -- 
> 2.29.1.341.ge80a0c044ae-goog
> 

Re: [PATCH bpf-next v2 7/8] bpf: Add tests for task_local_storage

[Alexei Starovoitov]

On Tue, Nov 3, 2020 at 4:05 PM KP Singh <kpsingh@chromium.org> wrote:
>
> On Tue, Nov 3, 2020 at 7:59 PM KP Singh <kpsingh@chromium.org> wrote:
> >
> > On Tue, Nov 3, 2020 at 7:47 PM Alexei Starovoitov
> > <alexei.starovoitov@gmail.com> wrote:
> > >
> > > On Tue, Nov 03, 2020 at 04:31:31PM +0100, KP Singh wrote:
> > > > +
> > > > +struct storage {
> > > > +     void *inode;
> > > > +     unsigned int value;
> > > > +     /* Lock ensures that spin locked versions of local stoage operations
> > > > +      * also work, most operations in this tests are still single threaded
> > > > +      */
> > > > +     struct bpf_spin_lock lock;
> > > > +};
> > >
> > > I think it's a good idea to test spin_lock in local_storage,
> > > but it seems the test is not doing it fully.
> > > It's only adding it to the storage, but the program is not accessing it.
> >
> > I added it here just to check if the offset calculations (map->spin_lock_off)
> > are correctly happening for these new maps.
> >
> > As mentioned in the updates, I do intend to generalize
> > tools/testing/selftests/bpf/map_tests/sk_storage_map.c which already has
> >  the threading logic to exercise bpf_spin_lock in storage maps.
> >
>
> Actually, after I added simple bpf_spin_{lock, unlock} to the test programs, I
> ended up realizing that we have not exposed spin locks to LSM programs
> for now, this is because they inherit the tracing helpers.
>
> I saw the docs mention that these are not exposed to tracing programs due to
> insufficient preemption checks. Do you think it would be okay to allow them
> for LSM programs?

hmm. Isn't it allowed already?
The verifier does:
        if ((is_tracing_prog_type(prog_type) ||
             prog_type == BPF_PROG_TYPE_SOCKET_FILTER) &&
            map_value_has_spin_lock(map)) {
                verbose(env, "tracing progs cannot use bpf_spin_lock yet\n");
                return -EINVAL;
        }

BPF_PROG_TYPE_LSM is not in this list.

Re: [PATCH bpf-next v2 7/8] bpf: Add tests for task_local_storage

[KP Singh]

[...]

> >
> > I saw the docs mention that these are not exposed to tracing programs due to
> > insufficient preemption checks. Do you think it would be okay to allow them
> > for LSM programs?
>
> hmm. Isn't it allowed already?
> The verifier does:
>         if ((is_tracing_prog_type(prog_type) ||
>              prog_type == BPF_PROG_TYPE_SOCKET_FILTER) &&
>             map_value_has_spin_lock(map)) {
>                 verbose(env, "tracing progs cannot use bpf_spin_lock yet\n");
>                 return -EINVAL;
>         }
>
> BPF_PROG_TYPE_LSM is not in this list.

The verifier does not have any problem, it's just that the helpers are not
exposed to LSM programs via bpf_lsm_func_proto.

So all we need is:

diff --git a/kernel/bpf/bpf_lsm.c b/kernel/bpf/bpf_lsm.c
index 61f8cc52fd5b..93383df2140b 100644
--- a/kernel/bpf/bpf_lsm.c
+++ b/kernel/bpf/bpf_lsm.c
@@ -63,6 +63,10 @@ bpf_lsm_func_proto(enum bpf_func_id func_id, const
struct bpf_prog *prog)
                return &bpf_task_storage_get_proto;
        case BPF_FUNC_task_storage_delete:
                return &bpf_task_storage_delete_proto;
+       case BPF_FUNC_spin_lock:
+               return &bpf_spin_lock_proto;
+       case BPF_FUNC_spin_unlock:
+               return &bpf_spin_unlock_proto;
        default:
                return tracing_prog_func_proto(func_id, prog);
        }

Re: [PATCH bpf-next v2 7/8] bpf: Add tests for task_local_storage

[Alexei Starovoitov]

On Tue, Nov 3, 2020 at 5:55 PM KP Singh <kpsingh@chromium.org> wrote:
>
> [...]
>
> > >
> > > I saw the docs mention that these are not exposed to tracing programs due to
> > > insufficient preemption checks. Do you think it would be okay to allow them
> > > for LSM programs?
> >
> > hmm. Isn't it allowed already?
> > The verifier does:
> >         if ((is_tracing_prog_type(prog_type) ||
> >              prog_type == BPF_PROG_TYPE_SOCKET_FILTER) &&
> >             map_value_has_spin_lock(map)) {
> >                 verbose(env, "tracing progs cannot use bpf_spin_lock yet\n");
> >                 return -EINVAL;
> >         }
> >
> > BPF_PROG_TYPE_LSM is not in this list.
>
> The verifier does not have any problem, it's just that the helpers are not
> exposed to LSM programs via bpf_lsm_func_proto.
>
> So all we need is:
>
> diff --git a/kernel/bpf/bpf_lsm.c b/kernel/bpf/bpf_lsm.c
> index 61f8cc52fd5b..93383df2140b 100644
> --- a/kernel/bpf/bpf_lsm.c
> +++ b/kernel/bpf/bpf_lsm.c
> @@ -63,6 +63,10 @@ bpf_lsm_func_proto(enum bpf_func_id func_id, const
> struct bpf_prog *prog)
>                 return &bpf_task_storage_get_proto;
>         case BPF_FUNC_task_storage_delete:
>                 return &bpf_task_storage_delete_proto;
> +       case BPF_FUNC_spin_lock:
> +               return &bpf_spin_lock_proto;
> +       case BPF_FUNC_spin_unlock:
> +               return &bpf_spin_unlock_proto;

Ahh. Yes. That should do it. Right now I don't see concerns with safety
of the bpf_spin_lock in bpf_lsm progs.

Re: [PATCH bpf-next v2 7/8] bpf: Add tests for task_local_storage

[John Fastabend]

Alexei Starovoitov wrote:
> On Tue, Nov 3, 2020 at 5:55 PM KP Singh <kpsingh@chromium.org> wrote:
> >
> > [...]
> >
> > > >
> > > > I saw the docs mention that these are not exposed to tracing programs due to
> > > > insufficient preemption checks. Do you think it would be okay to allow them
> > > > for LSM programs?
> > >
> > > hmm. Isn't it allowed already?
> > > The verifier does:
> > >         if ((is_tracing_prog_type(prog_type) ||
> > >              prog_type == BPF_PROG_TYPE_SOCKET_FILTER) &&
> > >             map_value_has_spin_lock(map)) {
> > >                 verbose(env, "tracing progs cannot use bpf_spin_lock yet\n");
> > >                 return -EINVAL;
> > >         }
> > >
> > > BPF_PROG_TYPE_LSM is not in this list.\f
> >
> > The verifier does not have any problem, it's just that the helpers are not
> > exposed to LSM programs via bpf_lsm_func_proto.
> >
> > So all we need is:
> >
> > diff --git a/kernel/bpf/bpf_lsm.c b/kernel/bpf/bpf_lsm.c
> > index 61f8cc52fd5b..93383df2140b 100644
> > --- a/kernel/bpf/bpf_lsm.c
> > +++ b/kernel/bpf/bpf_lsm.c
> > @@ -63,6 +63,10 @@ bpf_lsm_func_proto(enum bpf_func_id func_id, const
> > struct bpf_prog *prog)
> >                 return &bpf_task_storage_get_proto;
> >         case BPF_FUNC_task_storage_delete:
> >                 return &bpf_task_storage_delete_proto;
> > +       case BPF_FUNC_spin_lock:
> > +               return &bpf_spin_lock_proto;
> > +       case BPF_FUNC_spin_unlock:
> > +               return &bpf_spin_unlock_proto;
> 
> Ahh. Yes. That should do it. Right now I don't see concerns with safety
> of the bpf_spin_lock in bpf_lsm progs.

What about sleepable lsm hooks? Normally we wouldn't expect to sleep with
a spinlock held. Should we have a check to ensure programs bpf_spin_lock
are not also sleepable?

Re: [PATCH bpf-next v2 7/8] bpf: Add tests for task_local_storage

[KP Singh]

[...]

> > Ahh. Yes. That should do it. Right now I don't see concerns with safety
> > of the bpf_spin_lock in bpf_lsm progs.
>
> What about sleepable lsm hooks? Normally we wouldn't expect to sleep with
> a spinlock held. Should we have a check to ensure programs bpf_spin_lock
> are not also sleepable?

Thanks. Yes, I added that to my patch:

diff --git a/kernel/bpf/bpf_lsm.c b/kernel/bpf/bpf_lsm.c
index 61f8cc52fd5b..93383df2140b 100644
--- a/kernel/bpf/bpf_lsm.c
+++ b/kernel/bpf/bpf_lsm.c
@@ -63,6 +63,10 @@ bpf_lsm_func_proto(enum bpf_func_id func_id, const
struct bpf_prog *prog)
                return &bpf_task_storage_get_proto;
        case BPF_FUNC_task_storage_delete:
                return &bpf_task_storage_delete_proto;
+       case BPF_FUNC_spin_lock:
+               return &bpf_spin_lock_proto;
+       case BPF_FUNC_spin_unlock:
+               return &bpf_spin_unlock_proto;
        default:
                return tracing_prog_func_proto(func_id, prog);
        }
diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 314018e8fc12..8892f7ba2041 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -9739,6 +9739,23 @@ static int check_map_prog_compatibility(struct
bpf_verifier_env *env,
                return -EINVAL;
        }

+       if (map_value_has_spin_lock(map)) {
+               if (prog_type == BPF_PROG_TYPE_SOCKET_FILTER) {
+                       verbose(env, "socket filter progs cannot use
bpf_spin_lock yet\n");
+                       return -EINVAL;
+               }
+
+               if (is_tracing_prog_type(prog_type)) {
+                       verbose(env, "tracing progs cannot use
bpf_spin_lock yet\n");
+                       return -EINVAL;
+               }
+
+               if (prog->aux->sleepable) {
+                       verbose(env, "sleepable progs cannot use
bpf_spin_lock\n");
+                       return -EINVAL;
+               }
+       }
+

Re: [PATCH bpf-next v2 7/8] bpf: Add tests for task_local_storage

[KP Singh]

On Wed, Nov 4, 2020 at 12:03 PM KP Singh <kpsingh@chromium.org> wrote:
>
> [...]
>
> > > Ahh. Yes. That should do it. Right now I don't see concerns with safety
> > > of the bpf_spin_lock in bpf_lsm progs.
> >
> > What about sleepable lsm hooks? Normally we wouldn't expect to sleep with
> > a spinlock held. Should we have a check to ensure programs bpf_spin_lock
> > are not also sleepable?
>
> Thanks. Yes, I added that to my patch:
>
> diff --git a/kernel/bpf/bpf_lsm.c b/kernel/bpf/bpf_lsm.c
> index 61f8cc52fd5b..93383df2140b 100644
> --- a/kernel/bpf/bpf_lsm.c
> +++ b/kernel/bpf/bpf_lsm.c
> @@ -63,6 +63,10 @@ bpf_lsm_func_proto(enum bpf_func_id func_id, const
> struct bpf_prog *prog)
>                 return &bpf_task_storage_get_proto;
>         case BPF_FUNC_task_storage_delete:
>                 return &bpf_task_storage_delete_proto;
> +       case BPF_FUNC_spin_lock:
> +               return &bpf_spin_lock_proto;
> +       case BPF_FUNC_spin_unlock:
> +               return &bpf_spin_unlock_proto;
>         default:
>                 return tracing_prog_func_proto(func_id, prog);
>         }
> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index 314018e8fc12..8892f7ba2041 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c
> @@ -9739,6 +9739,23 @@ static int check_map_prog_compatibility(struct
> bpf_verifier_env *env,
>                 return -EINVAL;
>         }
>
> +       if (map_value_has_spin_lock(map)) {
> +               if (prog_type == BPF_PROG_TYPE_SOCKET_FILTER) {
> +                       verbose(env, "socket filter progs cannot use
> bpf_spin_lock yet\n");
> +                       return -EINVAL;
> +               }
> +
> +               if (is_tracing_prog_type(prog_type)) {
> +                       verbose(env, "tracing progs cannot use
> bpf_spin_lock yet\n");
> +                       return -EINVAL;
> +               }
> +
> +               if (prog->aux->sleepable) {
> +                       verbose(env, "sleepable progs cannot use
> bpf_spin_lock\n");

I think this can still be "yet" as it's doable; we can disable/enable
preemption in the helpers
and then have the verifier track that no sleepable helper is called
when a spin lock is held.
I would, however, prefer if we do it in a subsequent patch.

- KP

> +                       return -EINVAL;
> +               }
> +       }
> +