This set enables task local storage for non-BPF_LSM programs.
It is common for tracing BPF program to access per-task data. Currently,
these data are stored in hash tables with pid as the key. In
bcc/libbpftools [1], 9 out of 23 tools use such hash tables. However,
hash table is not ideal for many use case. Task local storage provides
better usability and performance for BPF programs. Please refer to 4/4 for
some performance comparison of task local storage vs. hash table.
Changes v2 => v3:
1. Make the selftest more robust. (Andrii)
2. Small changes with runqslower. (Andrii)
3. Shortern CC list to make it easy for vger.
Changes v1 => v2:
1. Do not allocate task local storage when the task is being freed.
2. Revise the selftest and added a new test for a task being freed.
3. Minor changes in runqslower.
Song Liu (4):
bpf: enable task local storage for tracing programs
selftests/bpf: add non-BPF_LSM test for task local storage
bpf: runqslower: prefer using local vmlimux to generate vmlinux.h
bpf: runqslower: use task local storage
include/linux/bpf.h | 7 ++
include/linux/bpf_lsm.h | 22 ------
include/linux/bpf_types.h | 2 +-
include/linux/sched.h | 5 ++
kernel/bpf/Makefile | 3 +-
kernel/bpf/bpf_local_storage.c | 28 +++++---
kernel/bpf/bpf_lsm.c | 4 --
kernel/bpf/bpf_task_storage.c | 34 +++------
kernel/fork.c | 5 ++
kernel/trace/bpf_trace.c | 4 ++
tools/bpf/runqslower/Makefile | 5 +-
tools/bpf/runqslower/runqslower.bpf.c | 33 +++++----
.../bpf/prog_tests/task_local_storage.c | 69 +++++++++++++++++++
.../selftests/bpf/progs/task_local_storage.c | 64 +++++++++++++++++
.../bpf/progs/task_local_storage_exit_creds.c | 32 +++++++++
15 files changed, 239 insertions(+), 78 deletions(-)
create mode 100644 tools/testing/selftests/bpf/prog_tests/task_local_storage.c
create mode 100644 tools/testing/selftests/bpf/progs/task_local_storage.c
create mode 100644 tools/testing/selftests/bpf/progs/task_local_storage_exit_creds.c
--
2.24.1
Update the Makefile to prefer using $(O)/mvlinux, $(KBUILD_OUTPUT)/vmlinux
(for selftests) or ../../../vmlinux. These two files should have latest
definitions for vmlinux.h.
Signed-off-by: Song Liu <[email protected]>
---
tools/bpf/runqslower/Makefile | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/tools/bpf/runqslower/Makefile b/tools/bpf/runqslower/Makefile
index 4d5ca54fcd4c8..e6d1f85a3871f 100644
--- a/tools/bpf/runqslower/Makefile
+++ b/tools/bpf/runqslower/Makefile
@@ -19,7 +19,10 @@ CFLAGS := -g -Wall
# Try to detect best kernel BTF source
KERNEL_REL := $(shell uname -r)
-VMLINUX_BTF_PATHS := /sys/kernel/btf/vmlinux /boot/vmlinux-$(KERNEL_REL)
+VMLINUX_BTF_PATHS := $(if $(O),$(O)/vmlinux) \
+ $(if $(KBUILD_OUTPUT),$(KBUILD_OUTPUT)/vmlinux) \
+ ../../../vmlinux /sys/kernel/btf/vmlinux \
+ /boot/vmlinux-$(KERNEL_REL)
VMLINUX_BTF_PATH := $(or $(VMLINUX_BTF),$(firstword \
$(wildcard $(VMLINUX_BTF_PATHS))))
--
2.24.1
Task local storage is enabled for tracing programs. Add two tests for
task local storage without CONFIG_BPF_LSM.
The first test stores a value in sys_enter and read it back in sys_exit.
The second test checks whether the kernel allows allocating task local
storage in exit_creds() (which it should not).
Signed-off-by: Song Liu <[email protected]>
---
.../bpf/prog_tests/task_local_storage.c | 69 +++++++++++++++++++
.../selftests/bpf/progs/task_local_storage.c | 64 +++++++++++++++++
.../bpf/progs/task_local_storage_exit_creds.c | 32 +++++++++
3 files changed, 165 insertions(+)
create mode 100644 tools/testing/selftests/bpf/prog_tests/task_local_storage.c
create mode 100644 tools/testing/selftests/bpf/progs/task_local_storage.c
create mode 100644 tools/testing/selftests/bpf/progs/task_local_storage_exit_creds.c
diff --git a/tools/testing/selftests/bpf/prog_tests/task_local_storage.c b/tools/testing/selftests/bpf/prog_tests/task_local_storage.c
new file mode 100644
index 0000000000000..dbb7525cdd567
--- /dev/null
+++ b/tools/testing/selftests/bpf/prog_tests/task_local_storage.c
@@ -0,0 +1,69 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2021 Facebook */
+
+#define _GNU_SOURCE /* See feature_test_macros(7) */
+#include <unistd.h>
+#include <sys/syscall.h> /* For SYS_xxx definitions */
+#include <sys/types.h>
+#include <test_progs.h>
+#include "task_local_storage.skel.h"
+#include "task_local_storage_exit_creds.skel.h"
+
+static void test_sys_enter_exit(void)
+{
+ struct task_local_storage *skel;
+ int err;
+
+ skel = task_local_storage__open_and_load();
+ if (!ASSERT_OK_PTR(skel, "skel_open_and_load"))
+ return;
+
+ skel->bss->target_pid = syscall(SYS_gettid);
+
+ err = task_local_storage__attach(skel);
+ if (!ASSERT_OK(err, "skel_attach"))
+ goto out;
+
+ syscall(SYS_gettid);
+ syscall(SYS_gettid);
+
+ /* 3x syscalls: 1x attach and 2x gettid */
+ ASSERT_EQ(skel->bss->enter_cnt, 3, "enter_cnt");
+ ASSERT_EQ(skel->bss->exit_cnt, 3, "exit_cnt");
+ ASSERT_EQ(skel->bss->mismatch_cnt, 0, "mismatch_cnt");
+out:
+ task_local_storage__destroy(skel);
+}
+
+static void test_exit_creds(void)
+{
+ struct task_local_storage_exit_creds *skel;
+ int err;
+
+ skel = task_local_storage_exit_creds__open_and_load();
+ if (!ASSERT_OK_PTR(skel, "skel_open_and_load"))
+ return;
+
+ err = task_local_storage_exit_creds__attach(skel);
+ if (!ASSERT_OK(err, "skel_attach"))
+ goto out;
+
+ /* trigger at least one exit_creds() */
+ if (CHECK_FAIL(system("ls > /dev/null")))
+ goto out;
+
+ /* sync rcu to make sure exit_creds() is called for "ls" */
+ kern_sync_rcu();
+ ASSERT_EQ(skel->bss->valid_ptr_count, 0, "valid_ptr_count");
+ ASSERT_NEQ(skel->bss->null_ptr_count, 0, "null_ptr_count");
+out:
+ task_local_storage_exit_creds__destroy(skel);
+}
+
+void test_task_local_storage(void)
+{
+ if (test__start_subtest("sys_enter_exit"))
+ test_sys_enter_exit();
+ if (test__start_subtest("exit_creds"))
+ test_exit_creds();
+}
diff --git a/tools/testing/selftests/bpf/progs/task_local_storage.c b/tools/testing/selftests/bpf/progs/task_local_storage.c
new file mode 100644
index 0000000000000..80a0a20db88d2
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/task_local_storage.c
@@ -0,0 +1,64 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2021 Facebook */
+
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+char _license[] SEC("license") = "GPL";
+
+struct {
+ __uint(type, BPF_MAP_TYPE_TASK_STORAGE);
+ __uint(map_flags, BPF_F_NO_PREALLOC);
+ __type(key, int);
+ __type(value, long);
+} enter_id SEC(".maps");
+
+#define MAGIC_VALUE 0xabcd1234
+
+pid_t target_pid = 0;
+int mismatch_cnt = 0;
+int enter_cnt = 0;
+int exit_cnt = 0;
+
+SEC("tp_btf/sys_enter")
+int BPF_PROG(on_enter, struct pt_regs *regs, long id)
+{
+ struct task_struct *task;
+ long *ptr;
+
+ task = bpf_get_current_task_btf();
+ if (task->pid != target_pid)
+ return 0;
+
+ ptr = bpf_task_storage_get(&enter_id, task, 0,
+ BPF_LOCAL_STORAGE_GET_F_CREATE);
+ if (!ptr)
+ return 0;
+
+ __sync_fetch_and_add(&enter_cnt, 1);
+ *ptr = MAGIC_VALUE + enter_cnt;
+
+ return 0;
+}
+
+SEC("tp_btf/sys_exit")
+int BPF_PROG(on_exit, struct pt_regs *regs, long id)
+{
+ struct task_struct *task;
+ long *ptr;
+
+ task = bpf_get_current_task_btf();
+ if (task->pid != target_pid)
+ return 0;
+
+ ptr = bpf_task_storage_get(&enter_id, task, 0,
+ BPF_LOCAL_STORAGE_GET_F_CREATE);
+ if (!ptr)
+ return 0;
+
+ __sync_fetch_and_add(&exit_cnt, 1);
+ if (*ptr != MAGIC_VALUE + exit_cnt)
+ __sync_fetch_and_add(&mismatch_cnt, 1);
+ return 0;
+}
diff --git a/tools/testing/selftests/bpf/progs/task_local_storage_exit_creds.c b/tools/testing/selftests/bpf/progs/task_local_storage_exit_creds.c
new file mode 100644
index 0000000000000..81758c0aef993
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/task_local_storage_exit_creds.c
@@ -0,0 +1,32 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2021 Facebook */
+
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+char _license[] SEC("license") = "GPL";
+
+struct {
+ __uint(type, BPF_MAP_TYPE_TASK_STORAGE);
+ __uint(map_flags, BPF_F_NO_PREALLOC);
+ __type(key, int);
+ __type(value, __u64);
+} task_storage SEC(".maps");
+
+int valid_ptr_count = 0;
+int null_ptr_count = 0;
+
+SEC("fentry/exit_creds")
+int BPF_PROG(trace_exit_creds, struct task_struct *task)
+{
+ __u64 *ptr;
+
+ ptr = bpf_task_storage_get(&task_storage, task, 0,
+ BPF_LOCAL_STORAGE_GET_F_CREATE);
+ if (ptr)
+ __sync_fetch_and_add(&valid_ptr_count, 1);
+ else
+ __sync_fetch_and_add(&null_ptr_count, 1);
+ return 0;
+}
--
2.24.1
To access per-task data, BPF programs usually creates a hash table with
pid as the key. This is not ideal because:
1. The user need to estimate the proper size of the hash table, which may
be inaccurate;
2. Big hash tables are slow;
3. To clean up the data properly during task terminations, the user need
to write extra logic.
Task local storage overcomes these issues and offers a better option for
these per-task data. Task local storage is only available to BPF_LSM. Now
enable it for tracing programs.
Unlike LSM progreams, tracing programs can be called in IRQ contexts.
Helpers that accesses task local storage are updated to use
raw_spin_lock_irqsave() instead of raw_spin_lock_bh().
Tracing programs can attach to functions on the task free path, e.g.
exit_creds(). To avoid allocating task local storage after
bpf_task_storage_free(). bpf_task_storage_get() is updated to not allocate
new storage when the task is not refcounted (task->usage == 0).
Signed-off-by: Song Liu <[email protected]>
---
include/linux/bpf.h | 7 +++++++
include/linux/bpf_lsm.h | 22 ----------------------
include/linux/bpf_types.h | 2 +-
include/linux/sched.h | 5 +++++
kernel/bpf/Makefile | 3 +--
kernel/bpf/bpf_local_storage.c | 28 +++++++++++++++++-----------
kernel/bpf/bpf_lsm.c | 4 ----
kernel/bpf/bpf_task_storage.c | 34 +++++++++-------------------------
kernel/fork.c | 5 +++++
kernel/trace/bpf_trace.c | 4 ++++
10 files changed, 49 insertions(+), 65 deletions(-)
diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index 1aac2af12fed2..bd0682af59173 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -1488,6 +1488,7 @@ struct bpf_prog *bpf_prog_by_id(u32 id);
struct bpf_link *bpf_link_by_id(u32 id);
const struct bpf_func_proto *bpf_base_func_proto(enum bpf_func_id func_id);
+void bpf_task_storage_free(struct task_struct *task);
#else /* !CONFIG_BPF_SYSCALL */
static inline struct bpf_prog *bpf_prog_get(u32 ufd)
{
@@ -1673,6 +1674,10 @@ bpf_base_func_proto(enum bpf_func_id func_id)
{
return NULL;
}
+
+static inline void bpf_task_storage_free(struct task_struct *task)
+{
+}
#endif /* CONFIG_BPF_SYSCALL */
void __bpf_free_used_btfs(struct bpf_prog_aux *aux,
@@ -1874,6 +1879,8 @@ extern const struct bpf_func_proto bpf_per_cpu_ptr_proto;
extern const struct bpf_func_proto bpf_this_cpu_ptr_proto;
extern const struct bpf_func_proto bpf_ktime_get_coarse_ns_proto;
extern const struct bpf_func_proto bpf_sock_from_file_proto;
+extern const struct bpf_func_proto bpf_task_storage_get_proto;
+extern const struct bpf_func_proto bpf_task_storage_delete_proto;
const struct bpf_func_proto *bpf_tracing_func_proto(
enum bpf_func_id func_id, const struct bpf_prog *prog);
diff --git a/include/linux/bpf_lsm.h b/include/linux/bpf_lsm.h
index 0d1c33ace3987..479c101546ad1 100644
--- a/include/linux/bpf_lsm.h
+++ b/include/linux/bpf_lsm.h
@@ -38,21 +38,9 @@ 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 */
@@ -73,20 +61,10 @@ 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 99f7fd657d87a..b9edee336d804 100644
--- a/include/linux/bpf_types.h
+++ b/include/linux/bpf_types.h
@@ -109,8 +109,8 @@ 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_TASK_STORAGE, task_storage_map_ops)
BPF_MAP_TYPE(BPF_MAP_TYPE_CPUMAP, cpu_map_ops)
#if defined(CONFIG_XDP_SOCKETS)
BPF_MAP_TYPE(BPF_MAP_TYPE_XSKMAP, xsk_map_ops)
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 6e3a5eeec509a..4abc58922c23e 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -42,6 +42,7 @@ struct audit_context;
struct backing_dev_info;
struct bio_list;
struct blk_plug;
+struct bpf_local_storage;
struct capture_control;
struct cfs_rq;
struct fs_struct;
@@ -1348,6 +1349,10 @@ struct task_struct {
/* Used by LSM modules for access restriction: */
void *security;
#endif
+#ifdef CONFIG_BPF_SYSCALL
+ /* Used by BPF task local storage */
+ struct bpf_local_storage __rcu *bpf_storage;
+#endif
#ifdef CONFIG_GCC_PLUGIN_STACKLEAK
unsigned long lowest_stack;
diff --git a/kernel/bpf/Makefile b/kernel/bpf/Makefile
index d1249340fd6ba..ca995fdfa45e7 100644
--- a/kernel/bpf/Makefile
+++ b/kernel/bpf/Makefile
@@ -8,9 +8,8 @@ CFLAGS_core.o += $(call cc-disable-warning, override-init) $(cflags-nogcse-yy)
obj-$(CONFIG_BPF_SYSCALL) += syscall.o verifier.o inode.o helpers.o tnum.o bpf_iter.o map_iter.o task_iter.o prog_iter.o
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-$(CONFIG_BPF_SYSCALL) += local_storage.o queue_stack_maps.o ringbuf.o bpf_task_storage.o
obj-${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_local_storage.c b/kernel/bpf/bpf_local_storage.c
index dd5aedee99e73..9bd47ad2b26f1 100644
--- a/kernel/bpf/bpf_local_storage.c
+++ b/kernel/bpf/bpf_local_storage.c
@@ -140,17 +140,18 @@ static void __bpf_selem_unlink_storage(struct bpf_local_storage_elem *selem)
{
struct bpf_local_storage *local_storage;
bool free_local_storage = false;
+ unsigned long flags;
if (unlikely(!selem_linked_to_storage(selem)))
/* selem has already been unlinked from sk */
return;
local_storage = rcu_dereference(selem->local_storage);
- raw_spin_lock_bh(&local_storage->lock);
+ raw_spin_lock_irqsave(&local_storage->lock, flags);
if (likely(selem_linked_to_storage(selem)))
free_local_storage = bpf_selem_unlink_storage_nolock(
local_storage, selem, true);
- raw_spin_unlock_bh(&local_storage->lock);
+ raw_spin_unlock_irqrestore(&local_storage->lock, flags);
if (free_local_storage)
kfree_rcu(local_storage, rcu);
@@ -167,6 +168,7 @@ void bpf_selem_unlink_map(struct bpf_local_storage_elem *selem)
{
struct bpf_local_storage_map *smap;
struct bpf_local_storage_map_bucket *b;
+ unsigned long flags;
if (unlikely(!selem_linked_to_map(selem)))
/* selem has already be unlinked from smap */
@@ -174,21 +176,22 @@ void bpf_selem_unlink_map(struct bpf_local_storage_elem *selem)
smap = rcu_dereference(SDATA(selem)->smap);
b = select_bucket(smap, selem);
- raw_spin_lock_bh(&b->lock);
+ raw_spin_lock_irqsave(&b->lock, flags);
if (likely(selem_linked_to_map(selem)))
hlist_del_init_rcu(&selem->map_node);
- raw_spin_unlock_bh(&b->lock);
+ raw_spin_unlock_irqrestore(&b->lock, flags);
}
void bpf_selem_link_map(struct bpf_local_storage_map *smap,
struct bpf_local_storage_elem *selem)
{
struct bpf_local_storage_map_bucket *b = select_bucket(smap, selem);
+ unsigned long flags;
- raw_spin_lock_bh(&b->lock);
+ raw_spin_lock_irqsave(&b->lock, flags);
RCU_INIT_POINTER(SDATA(selem)->smap, smap);
hlist_add_head_rcu(&selem->map_node, &b->list);
- raw_spin_unlock_bh(&b->lock);
+ raw_spin_unlock_irqrestore(&b->lock, flags);
}
void bpf_selem_unlink(struct bpf_local_storage_elem *selem)
@@ -224,16 +227,18 @@ bpf_local_storage_lookup(struct bpf_local_storage *local_storage,
sdata = SDATA(selem);
if (cacheit_lockit) {
+ unsigned long flags;
+
/* spinlock is needed to avoid racing with the
* parallel delete. Otherwise, publishing an already
* deleted sdata to the cache will become a use-after-free
* problem in the next bpf_local_storage_lookup().
*/
- raw_spin_lock_bh(&local_storage->lock);
+ raw_spin_lock_irqsave(&local_storage->lock, flags);
if (selem_linked_to_storage(selem))
rcu_assign_pointer(local_storage->cache[smap->cache_idx],
sdata);
- raw_spin_unlock_bh(&local_storage->lock);
+ raw_spin_unlock_irqrestore(&local_storage->lock, flags);
}
return sdata;
@@ -327,6 +332,7 @@ bpf_local_storage_update(void *owner, struct bpf_local_storage_map *smap,
struct bpf_local_storage_data *old_sdata = NULL;
struct bpf_local_storage_elem *selem;
struct bpf_local_storage *local_storage;
+ unsigned long flags;
int err;
/* BPF_EXIST and BPF_NOEXIST cannot be both set */
@@ -374,7 +380,7 @@ bpf_local_storage_update(void *owner, struct bpf_local_storage_map *smap,
}
}
- raw_spin_lock_bh(&local_storage->lock);
+ raw_spin_lock_irqsave(&local_storage->lock, flags);
/* Recheck local_storage->list under local_storage->lock */
if (unlikely(hlist_empty(&local_storage->list))) {
@@ -428,11 +434,11 @@ bpf_local_storage_update(void *owner, struct bpf_local_storage_map *smap,
}
unlock:
- raw_spin_unlock_bh(&local_storage->lock);
+ raw_spin_unlock_irqrestore(&local_storage->lock, flags);
return SDATA(selem);
unlock_err:
- raw_spin_unlock_bh(&local_storage->lock);
+ raw_spin_unlock_irqrestore(&local_storage->lock, flags);
return ERR_PTR(err);
}
diff --git a/kernel/bpf/bpf_lsm.c b/kernel/bpf/bpf_lsm.c
index 70e5e0b6d69d0..cf4ebea95c993 100644
--- a/kernel/bpf/bpf_lsm.c
+++ b/kernel/bpf/bpf_lsm.c
@@ -115,10 +115,6 @@ bpf_lsm_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
return &bpf_spin_lock_proto;
case BPF_FUNC_spin_unlock:
return &bpf_spin_unlock_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;
case BPF_FUNC_bprm_opts_set:
return &bpf_bprm_opts_set_proto;
case BPF_FUNC_ima_inode_hash:
diff --git a/kernel/bpf/bpf_task_storage.c b/kernel/bpf/bpf_task_storage.c
index e0da0258b732d..2034019966d44 100644
--- a/kernel/bpf/bpf_task_storage.c
+++ b/kernel/bpf/bpf_task_storage.c
@@ -15,7 +15,6 @@
#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>
@@ -24,12 +23,8 @@ 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;
+ return &task->bpf_storage;
}
static struct bpf_local_storage_data *
@@ -38,13 +33,8 @@ task_storage_lookup(struct task_struct *task, struct bpf_map *map,
{
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);
+ task_storage = rcu_dereference(task->bpf_storage);
if (!task_storage)
return NULL;
@@ -57,16 +47,12 @@ 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;
+ unsigned long flags;
rcu_read_lock();
- local_storage = rcu_dereference(bsb->storage);
+ local_storage = rcu_dereference(task->bpf_storage);
if (!local_storage) {
rcu_read_unlock();
return;
@@ -81,7 +67,7 @@ void bpf_task_storage_free(struct task_struct *task)
* when unlinking elem from the local_storage->list and
* the map's bucket->list.
*/
- raw_spin_lock_bh(&local_storage->lock);
+ raw_spin_lock_irqsave(&local_storage->lock, flags);
hlist_for_each_entry_safe(selem, n, &local_storage->list, snode) {
/* Always unlink from map before unlinking from
* local_storage.
@@ -90,7 +76,7 @@ void bpf_task_storage_free(struct task_struct *task)
free_task_storage = bpf_selem_unlink_storage_nolock(
local_storage, selem, false);
}
- raw_spin_unlock_bh(&local_storage->lock);
+ raw_spin_unlock_irqrestore(&local_storage->lock, flags);
rcu_read_unlock();
/* free_task_storage should always be true as long as
@@ -225,11 +211,9 @@ BPF_CALL_4(bpf_task_storage_get, struct bpf_map *, map, struct task_struct *,
if (sdata)
return (unsigned long)sdata->data;
- /* This helper must only be called from places where the lifetime of the task
- * is guaranteed. Either by being refcounted or by being protected
- * by an RCU read-side critical section.
- */
- if (flags & BPF_LOCAL_STORAGE_GET_F_CREATE) {
+ /* only allocate new storage, when the task is refcounted */
+ if (refcount_read(&task->usage) &&
+ (flags & BPF_LOCAL_STORAGE_GET_F_CREATE)) {
sdata = bpf_local_storage_update(
task, (struct bpf_local_storage_map *)map, value,
BPF_NOEXIST);
diff --git a/kernel/fork.c b/kernel/fork.c
index 37720a6d04eaa..3eeebe07a6e9c 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -96,6 +96,7 @@
#include <linux/kasan.h>
#include <linux/scs.h>
#include <linux/io_uring.h>
+#include <linux/bpf.h>
#include <asm/pgalloc.h>
#include <linux/uaccess.h>
@@ -734,6 +735,7 @@ void __put_task_struct(struct task_struct *tsk)
cgroup_free(tsk);
task_numa_free(tsk, true);
security_task_free(tsk);
+ bpf_task_storage_free(tsk);
exit_creds(tsk);
delayacct_tsk_free(tsk);
put_signal_struct(tsk->signal);
@@ -2064,6 +2066,9 @@ static __latent_entropy struct task_struct *copy_process(
p->sequential_io = 0;
p->sequential_io_avg = 0;
#endif
+#ifdef CONFIG_BPF_SYSCALL
+ RCU_INIT_POINTER(p->bpf_storage, NULL);
+#endif
/* Perform scheduler related setup. Assign this task to a CPU. */
retval = sched_fork(clone_flags, p);
diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
index 6c0018abe68a0..9b12d92dde1d4 100644
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -1366,6 +1366,10 @@ bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
return &bpf_per_cpu_ptr_proto;
case BPF_FUNC_this_cpu_ptr:
return &bpf_this_cpu_ptr_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 NULL;
}
--
2.24.1
Replace hashtab with task local storage in runqslower. This improves the
performance of these BPF programs. The following table summarizes average
runtime of these programs, in nanoseconds:
task-local hash-prealloc hash-no-prealloc
handle__sched_wakeup 125 340 3124
handle__sched_wakeup_new 2812 1510 2998
handle__sched_switch 151 208 991
Note that, task local storage gives better performance than hashtab for
handle__sched_wakeup and handle__sched_switch. On the other hand, for
handle__sched_wakeup_new, task local storage is slower than hashtab with
prealloc. This is because handle__sched_wakeup_new accesses the data for
the first time, so it has to allocate the data for task local storage.
Once the initial allocation is done, subsequent accesses, as those in
handle__sched_wakeup, are much faster with task local storage. If we
disable hashtab prealloc, task local storage is much faster for all 3
functions.
Acked-by: Andrii Nakryiko <[email protected]>
Signed-off-by: Song Liu <[email protected]>
---
tools/bpf/runqslower/runqslower.bpf.c | 33 +++++++++++++++++----------
1 file changed, 21 insertions(+), 12 deletions(-)
diff --git a/tools/bpf/runqslower/runqslower.bpf.c b/tools/bpf/runqslower/runqslower.bpf.c
index 1f18a409f0443..645530ca7e985 100644
--- a/tools/bpf/runqslower/runqslower.bpf.c
+++ b/tools/bpf/runqslower/runqslower.bpf.c
@@ -11,9 +11,9 @@ const volatile __u64 min_us = 0;
const volatile pid_t targ_pid = 0;
struct {
- __uint(type, BPF_MAP_TYPE_HASH);
- __uint(max_entries, 10240);
- __type(key, u32);
+ __uint(type, BPF_MAP_TYPE_TASK_STORAGE);
+ __uint(map_flags, BPF_F_NO_PREALLOC);
+ __type(key, int);
__type(value, u64);
} start SEC(".maps");
@@ -25,15 +25,20 @@ struct {
/* record enqueue timestamp */
__always_inline
-static int trace_enqueue(u32 tgid, u32 pid)
+static int trace_enqueue(struct task_struct *t)
{
- u64 ts;
+ u32 pid = t->pid;
+ u64 *ptr;
if (!pid || (targ_pid && targ_pid != pid))
return 0;
- ts = bpf_ktime_get_ns();
- bpf_map_update_elem(&start, &pid, &ts, 0);
+ ptr = bpf_task_storage_get(&start, t, 0,
+ BPF_LOCAL_STORAGE_GET_F_CREATE);
+ if (!ptr)
+ return 0;
+
+ *ptr = bpf_ktime_get_ns();
return 0;
}
@@ -43,7 +48,7 @@ int handle__sched_wakeup(u64 *ctx)
/* TP_PROTO(struct task_struct *p) */
struct task_struct *p = (void *)ctx[0];
- return trace_enqueue(p->tgid, p->pid);
+ return trace_enqueue(p);
}
SEC("tp_btf/sched_wakeup_new")
@@ -52,7 +57,7 @@ int handle__sched_wakeup_new(u64 *ctx)
/* TP_PROTO(struct task_struct *p) */
struct task_struct *p = (void *)ctx[0];
- return trace_enqueue(p->tgid, p->pid);
+ return trace_enqueue(p);
}
SEC("tp_btf/sched_switch")
@@ -70,12 +75,16 @@ int handle__sched_switch(u64 *ctx)
/* ivcsw: treat like an enqueue event and store timestamp */
if (prev->state == TASK_RUNNING)
- trace_enqueue(prev->tgid, prev->pid);
+ trace_enqueue(prev);
pid = next->pid;
+ /* For pid mismatch, save a bpf_task_storage_get */
+ if (!pid || (targ_pid && targ_pid != pid))
+ return 0;
+
/* fetch timestamp and calculate delta */
- tsp = bpf_map_lookup_elem(&start, &pid);
+ tsp = bpf_task_storage_get(&start, next, 0, 0);
if (!tsp)
return 0; /* missed enqueue */
@@ -91,7 +100,7 @@ int handle__sched_switch(u64 *ctx)
bpf_perf_event_output(ctx, &events, BPF_F_CURRENT_CPU,
&event, sizeof(event));
- bpf_map_delete_elem(&start, &pid);
+ bpf_task_storage_delete(&start, next);
return 0;
}
--
2.24.1
On Thu, Jan 28, 2021 at 1:20 AM Song Liu <[email protected]> wrote:
>
> To access per-task data, BPF programs usually creates a hash table with
> pid as the key. This is not ideal because:
> 1. The user need to estimate the proper size of the hash table, which may
> be inaccurate;
> 2. Big hash tables are slow;
> 3. To clean up the data properly during task terminations, the user need
> to write extra logic.
>
> Task local storage overcomes these issues and offers a better option for
> these per-task data. Task local storage is only available to BPF_LSM. Now
> enable it for tracing programs.
>
> Unlike LSM progreams, tracing programs can be called in IRQ contexts.
nit: typo *programs
> Helpers that accesses task local storage are updated to use
nit: Helpers that access..
> raw_spin_lock_irqsave() instead of raw_spin_lock_bh().
>
> Tracing programs can attach to functions on the task free path, e.g.
> exit_creds(). To avoid allocating task local storage after
> bpf_task_storage_free(). bpf_task_storage_get() is updated to not allocate
> new storage when the task is not refcounted (task->usage == 0).
>
> Signed-off-by: Song Liu <[email protected]>
Acked-by: KP Singh <[email protected]>
Thanks for adding better commit descriptions :)
I think checking the usage before adding storage should work for the
task exit path (I could not think of cases where it would break).
Would also be nice to check with Martin and Hao about this.