Skip to contentSkip to conversation

bpf: implement bpf iterator for kernel data

PATCH In review Yonghong Song 2 participants 42 messages May 9, 2020 – May 12, 2020

Subsystems (14)
lore
Share & tools
lore
#0 lore raw

v4 cover bpf: implement bpf iterator for kernel data

Motivation:
The current way to dump kernel data structures mostly:
1. /proc system
2. various specific tools like "ss" which requires kernel support.
3. drgn
The dropback for the first two is that whenever you want to dump more, you need change the kernel. For example, Martin wants to dump socket local storage with "ss". Kernel change is needed for it to work ([1]). This is also the direct motivation for this work.

drgn ([2]) solves this proble nicely and no kernel change is not needed. But since drgn is not able to verify the validity of a particular pointer value, it might present the wrong results in rare cases.

In this patch set, we introduce bpf iterator. Initial kernel changes are still needed for interested kernel data, but a later data structure change will not require kernel changes any more. bpf program itself can adapt to new data structure changes. This will give certain flexibility with guaranteed correctness.

In this patch set, kernel seq_ops is used to facilitate iterating through kernel data, similar to current /proc and many other lossless kernel dumping facilities. In the future, different iterators can be implemented to trade off losslessness for other criteria e.g. no repeated object visits, etc.

User Interface:
1. Similar to prog/map/link, the iterator can be pinned into a path within a bpffs mount point.
2. The bpftool command can pin an iterator to a file bpftool iter pin <bpf_prog.o> <path>
3. Use cat <path> to dump the contents.
Use rm -f <path> to remove the pinned iterator.
4. The anonymous iterator can be created as well.

Please see patch #19 andd #20 for bpf programs and bpf iterator output examples.

Note that certain iterators are namespace aware. For example, task and task_file targets only iterate through current pid namespace. ipv6_route and netlink will iterate through current net namespace.

Please see individual patches for implementation details.

Performance:
The bpf iterator provides in-kernel aggregation abilities
for kernel data. This can greatly improve performance
compared to e.g., iterating all process directories under /proc.
For example, I did an experiment on my VM with an application forking different number of tasks and each forked process opening various number of files. The following is the result with the latency with unit of microseconds:

    # of forked tasks   # of open files    # of bpf_prog calls  # latency (us)
    100                 100                11503                7586
    1000                1000               1013203              709513
    10000               100                1130203              764519

The number of bpf_prog calls may be more than forked tasks multipled by open files since there are other tasks running on the system. The bpf program is a do-nothing program. One millions of bpf calls takes less than one second.

Although the initial motivation is from Martin's sk_local_storage, this patch didn't implement tcp6 sockets and sk_local_storage. The /proc/net/tcp6 involves three types of sockets, timewait, request and tcp6 sockets. Some kind of type casting or other mechanism is needed to handle all these socket types in one bpf program. This will be addressed in future work.

Currently, we do not support kernel data generated under module. This requires some BTF work.

More work for more iterators, e.g., tcp, udp, bpf_map elements, etc.

Changelog:
v3 -> v4:
- in bpf_seq_read(), if start() failed with an error, return that error to user space (Andrii)
- in bpf_seq_printf(), if reading kernel memory failed for %s and %p{i,I}{4,6}, set buffer to empty string or address 0. Documented this behavior in uapi header (Andrii)
- fix a few error handling issues for bpftool (Andrii)
- A few other minor fixes and cosmetic changes.
v2 -> v3:
- add bpf_iter_unreg_target() to unregister a target, used in the error path of the __init functions.
- handle err != 0 before handling overflow (Andrii)
- reference count "task" for task_file target (Andrii)
- remove some redundancy for bpf_map/task/task_file targets
- add bpf_iter_unreg_target() in ip6_route_cleanup()
- Handling "%%" format in bpf_seq_printf() (Andrii)
- implement auto-attach for bpf_iter in libbpf (Andrii)
- add macros offsetof and container_of in bpf_helpers.h (Andrii)
- add tests for auto-attach and program-return-1 cases
- some other minor fixes
v1 -> v2:
- removed target_feature, using callback functions instead
- checking target to ensure program specified btf_id supported (Martin)
- link_create change with new changes from Andrii
- better handling of btf_iter vs. seq_file private data (Martin, Andrii)
- implemented bpf_seq_read() (Andrii, Alexei)
- percpu buffer for bpf_seq_printf() (Andrii)
- better syntax for BPF_SEQ_PRINTF macro (Andrii)
- bpftool fixes (Quentin)
- a lot of other fixes
RFC v2 -> v1:
- rename bpfdump to bpf_iter
- use bpffs instead of a new file system
- use bpf_link to streamline and simplify iterator creation.

References:
[1]: https://lore.kernel.org/bpf/[email protected]
[2]: https://github.com/osandov/drgn

Yonghong Song (21):
bpf: implement an interface to register bpf_iter targets
bpf: allow loading of a bpf_iter program
bpf: support bpf tracing/iter programs for BPF_LINK_CREATE
bpf: support bpf tracing/iter programs for BPF_LINK_UPDATE
bpf: implement bpf_seq_read() for bpf iterator
bpf: create anonymous bpf iterator
bpf: create file bpf iterator
bpf: implement common macros/helpers for target iterators
bpf: add bpf_map iterator
net: bpf: add netlink and ipv6_route bpf_iter targets
bpf: add task and task/file iterator targets
bpf: add PTR_TO_BTF_ID_OR_NULL support
bpf: add bpf_seq_printf and bpf_seq_write helpers
bpf: handle spilled PTR_TO_BTF_ID properly when checking
stack_boundary
bpf: support variable length array in tracing programs tools/libbpf: add bpf_iter support
tools/libpf: add offsetof/container_of macro in bpf_helpers.h tools/bpftool: add bpf_iter support for bptool
tools/bpf: selftests: add iterator programs for ipv6_route and netlink tools/bpf: selftests: add iter progs for bpf_map/task/task_file tools/bpf: selftests: add bpf_iter selftests

 fs/proc/proc_net.c                            |  19 +
 include/linux/bpf.h                           |  36 ++
 include/linux/bpf_types.h                     |   1 +
 include/linux/proc_fs.h                       |   3 +
 include/uapi/linux/bpf.h                      |  47 +-
 kernel/bpf/Makefile                           |   2 +-
 kernel/bpf/bpf_iter.c                         | 530 ++++++++++++++++++
 kernel/bpf/btf.c                              |  42 +-
 kernel/bpf/inode.c                            |   5 +-
 kernel/bpf/map_iter.c                         |  97 ++++
 kernel/bpf/syscall.c                          |  59 ++
 kernel/bpf/task_iter.c                        | 333 +++++++++++
 kernel/bpf/verifier.c                         |  42 +-
 kernel/trace/bpf_trace.c                      | 214 +++++++
 net/ipv6/ip6_fib.c                            |  65 ++-
 net/ipv6/route.c                              |  37 ++
 net/netlink/af_netlink.c                      |  87 ++-
 scripts/bpf_helpers_doc.py                    |   2 +
 .../bpftool/Documentation/bpftool-iter.rst    |  83 +++
 tools/bpf/bpftool/bash-completion/bpftool     |  13 +
 tools/bpf/bpftool/iter.c                      |  88 +++
 tools/bpf/bpftool/link.c                      |   1 +
 tools/bpf/bpftool/main.c                      |   3 +-
 tools/bpf/bpftool/main.h                      |   1 +
 tools/include/uapi/linux/bpf.h                |  47 +-
 tools/lib/bpf/bpf.c                           |  10 +
 tools/lib/bpf/bpf.h                           |   2 +
 tools/lib/bpf/bpf_helpers.h                   |  14 +
 tools/lib/bpf/bpf_tracing.h                   |  16 +
 tools/lib/bpf/libbpf.c                        |  52 ++
 tools/lib/bpf/libbpf.h                        |   9 +
 tools/lib/bpf/libbpf.map                      |   2 +
 .../selftests/bpf/prog_tests/bpf_iter.c       | 409 ++++++++++++++
 .../selftests/bpf/progs/bpf_iter_bpf_map.c    |  28 +
 .../selftests/bpf/progs/bpf_iter_ipv6_route.c |  62 ++
 .../selftests/bpf/progs/bpf_iter_netlink.c    |  66 +++
 .../selftests/bpf/progs/bpf_iter_task.c       |  25 +
 .../selftests/bpf/progs/bpf_iter_task_file.c  |  26 +
 .../selftests/bpf/progs/bpf_iter_test_kern1.c |   4 +
 .../selftests/bpf/progs/bpf_iter_test_kern2.c |   4 +
 .../selftests/bpf/progs/bpf_iter_test_kern3.c |  18 +
 .../selftests/bpf/progs/bpf_iter_test_kern4.c |  52 ++
 .../bpf/progs/bpf_iter_test_kern_common.h     |  22 +
 43 files changed, 2664 insertions(+), 14 deletions(-)
 create mode 100644 kernel/bpf/bpf_iter.c
 create mode 100644 kernel/bpf/map_iter.c
 create mode 100644 kernel/bpf/task_iter.c
 create mode 100644 tools/bpf/bpftool/Documentation/bpftool-iter.rst
 create mode 100644 tools/bpf/bpftool/iter.c
 create mode 100644 tools/testing/selftests/bpf/prog_tests/bpf_iter.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_bpf_map.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_ipv6_route.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_netlink.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_task.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_task_file.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_test_kern1.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_test_kern2.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_test_kern3.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_test_kern4.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_test_kern_common.h
Signature · 1 line
2.24.1
#1 lore raw

PATCH v4 1/21 bpf: implement an interface to register bpf_iter targets

The target can call bpf_iter_reg_target() to register itself. The needed information:

  target:           target name
  seq_ops:          the seq_file operations for the target
  init_seq_private  target callback to initialize seq_priv during file open
  fini_seq_private  target callback to clean up seq_priv during file release
  seq_priv_size:    the private_data size needed by the seq_file

operations

The target name represents a target which provides a seq_ops for iterating objects.

The target can provide two callback functions, init_seq_private and fini_seq_private, called during file open/release time. For example, /proc/net/{tcp6, ipv6_route, netlink, ...}, net name space needs to be setup properly during file open and released properly during file release.

Function bpf_iter_unreg_target() is also implemented to unregister a particular target.

---
 include/linux/bpf.h   | 15 +++++++++++
 kernel/bpf/Makefile   |  2 +-
 kernel/bpf/bpf_iter.c | 59 +++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 75 insertions(+), 1 deletion(-)
 create mode 100644 kernel/bpf/bpf_iter.c
include/linux/bpf.h +15 -0
index 1262ec460ab3..40c78b86fe38 100644
@@ -31,6 +31,7 @@ struct seq_file;
 struct btf;
 struct btf_type;
 struct exception_table_entry;
+struct seq_operations;
 
 extern struct idr btf_idr;
 extern spinlock_t btf_idr_lock;
@@ -1126,6 +1127,20 @@ struct bpf_link *bpf_link_get_from_fd(u32 ufd);
 int bpf_obj_pin_user(u32 ufd, const char __user *pathname);
 int bpf_obj_get_user(const char __user *pathname, int flags);
 
+typedef int (*bpf_iter_init_seq_priv_t)(void *private_data);
+typedef void (*bpf_iter_fini_seq_priv_t)(void *private_data);
+
+struct bpf_iter_reg {
+	const char *target;
+	const struct seq_operations *seq_ops;
+	bpf_iter_init_seq_priv_t init_seq_private;
+	bpf_iter_fini_seq_priv_t fini_seq_private;
+	u32 seq_priv_size;
+};
+
+int bpf_iter_reg_target(struct bpf_iter_reg *reg_info);
+void bpf_iter_unreg_target(const char *target);
+
 int bpf_percpu_hash_copy(struct bpf_map *map, void *key, void *value);
 int bpf_percpu_array_copy(struct bpf_map *map, void *key, void *value);
 int bpf_percpu_hash_update(struct bpf_map *map, void *key, void *value,
kernel/bpf/Makefile +1 -1
index f2d7be596966..6a8b0febd3f6 100644
@@ -2,7 +2,7 @@
 obj-y := core.o
 CFLAGS_core.o += $(call cc-disable-warning, override-init)
 
-obj-$(CONFIG_BPF_SYSCALL) += syscall.o verifier.o inode.o helpers.o tnum.o
+obj-$(CONFIG_BPF_SYSCALL) += syscall.o verifier.o inode.o helpers.o tnum.o bpf_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
 obj-$(CONFIG_BPF_SYSCALL) += disasm.o
kernel/bpf/bpf_iter.c new file +59 -0

59 lines not on this page · raw message

Signature · 1 line
2.24.1
#2 lore raw

PATCH v4 2/21 bpf: allow loading of a bpf_iter program

In reply to Yonghong Song

A bpf_iter program is a tracing program with attach type BPF_TRACE_ITER. The load attribute
attach_btf_id
is used by the verifier against a particular kernel function, which represents a target, e.g., __bpf_iter__bpf_map for target bpf_map which is implemented later.

The program return value must be 0 or 1 for now.
0 : successful, except potential seq_file buffer overflow
which is handled by seq_file reader.
1 : request to restart the same object

In the future, other return values may be used for filtering or teminating the iterator.

---
 include/linux/bpf.h            |  3 +++
 include/uapi/linux/bpf.h       |  1 +
 kernel/bpf/bpf_iter.c          | 36 ++++++++++++++++++++++++++++++++++
 kernel/bpf/verifier.c          | 21 ++++++++++++++++++++
 tools/include/uapi/linux/bpf.h |  1 +
 5 files changed, 62 insertions(+)
include/linux/bpf.h +3 -0
index 40c78b86fe38..f28bdd714754 100644
@@ -1127,6 +1127,8 @@ struct bpf_link *bpf_link_get_from_fd(u32 ufd);
 int bpf_obj_pin_user(u32 ufd, const char __user *pathname);
 int bpf_obj_get_user(const char __user *pathname, int flags);
 
+#define BPF_ITER_FUNC_PREFIX "__bpf_iter__"
+
 typedef int (*bpf_iter_init_seq_priv_t)(void *private_data);
 typedef void (*bpf_iter_fini_seq_priv_t)(void *private_data);
 
@@ -1140,6 +1142,7 @@ struct bpf_iter_reg {
 
 int bpf_iter_reg_target(struct bpf_iter_reg *reg_info);
 void bpf_iter_unreg_target(const char *target);
+bool bpf_iter_prog_supported(struct bpf_prog *prog);
 
 int bpf_percpu_hash_copy(struct bpf_map *map, void *key, void *value);
 int bpf_percpu_array_copy(struct bpf_map *map, void *key, void *value);
include/uapi/linux/bpf.h +1 -0
index 6e5e7caa3739..c8a5325cc8d0 100644
@@ -218,6 +218,7 @@ enum bpf_attach_type {
 	BPF_TRACE_FEXIT,
 	BPF_MODIFY_RETURN,
 	BPF_LSM_MAC,
+	BPF_TRACE_ITER,
 	__MAX_BPF_ATTACH_TYPE
 };
 
kernel/bpf/bpf_iter.c +36 -0
index 5a8119d17d14..dec182d8395a 100644
@@ -12,6 +12,7 @@ struct bpf_iter_target_info {
 	bpf_iter_init_seq_priv_t init_seq_private;
 	bpf_iter_fini_seq_priv_t fini_seq_private;
 	u32 seq_priv_size;
+	u32 btf_id;	/* cached value */
 };
 
 static struct list_head targets = LIST_HEAD_INIT(targets);
@@ -57,3 +58,38 @@ void bpf_iter_unreg_target(const char *target)
 
 	WARN_ON(found == false);
 }
+
+static void cache_btf_id(struct bpf_iter_target_info *tinfo,
+			 struct bpf_prog *prog)
+{
+	tinfo->btf_id = prog->aux->attach_btf_id;
+}
+
+bool bpf_iter_prog_supported(struct bpf_prog *prog)
+{
+	const char *attach_fname = prog->aux->attach_func_name;
+	u32 prog_btf_id = prog->aux->attach_btf_id;
+	const char *prefix = BPF_ITER_FUNC_PREFIX;
+	struct bpf_iter_target_info *tinfo;
+	int prefix_len = strlen(prefix);
+	bool supported = false;
+
+	if (strncmp(attach_fname, prefix, prefix_len))
+		return false;
+
+	mutex_lock(&targets_mutex);
+	list_for_each_entry(tinfo, &targets, list) {
+		if (tinfo->btf_id && tinfo->btf_id == prog_btf_id) {
+			supported = true;
+			break;
+		}
+		if (!strcmp(attach_fname + prefix_len, tinfo->target)) {
+			cache_btf_id(tinfo, prog);
+			supported = true;
+			break;
+		}
+	}
+	mutex_unlock(&targets_mutex);
+
+	return supported;
+}
kernel/bpf/verifier.c +21 -0
index 70ad009577f8..d725ff7d11db 100644
@@ -7101,6 +7101,10 @@ static int check_return_code(struct bpf_verifier_env *env)
 			return 0;
 		range = tnum_const(0);
 		break;
+	case BPF_PROG_TYPE_TRACING:
+		if (env->prog->expected_attach_type != BPF_TRACE_ITER)
+			return 0;
+		break;
 	default:
 		return 0;
 	}
@@ -10481,6 +10485,7 @@ static int check_attach_btf_id(struct bpf_verifier_env *env)
 	struct bpf_prog *tgt_prog = prog->aux->linked_prog;
 	u32 btf_id = prog->aux->attach_btf_id;
 	const char prefix[] = "btf_trace_";
+	struct btf_func_model fmodel;
 	int ret = 0, subprog = -1, i;
 	struct bpf_trampoline *tr;
 	const struct btf_type *t;
@@ -10622,6 +10627,22 @@ static int check_attach_btf_id(struct bpf_verifier_env *env)
 		prog->aux->attach_func_proto = t;
 		prog->aux->attach_btf_trace = true;
 		return 0;
+	case BPF_TRACE_ITER:
+		if (!btf_type_is_func(t)) {
+			verbose(env, "attach_btf_id %u is not a function\n",
+				btf_id);
+			return -EINVAL;
+		}
+		t = btf_type_by_id(btf, t->type);
+		if (!btf_type_is_func_proto(t))
+			return -EINVAL;
+		prog->aux->attach_func_name = tname;
+		prog->aux->attach_func_proto = t;
+		if (!bpf_iter_prog_supported(prog))
+			return -EINVAL;
+		ret = btf_distill_func_proto(&env->log, btf, t,
+					     tname, &fmodel);
+		return ret;
 	default:
 		if (!prog_extension)
 			return -EINVAL;
tools/include/uapi/linux/bpf.h +1 -0
index 6e5e7caa3739..c8a5325cc8d0 100644
@@ -218,6 +218,7 @@ enum bpf_attach_type {
 	BPF_TRACE_FEXIT,
 	BPF_MODIFY_RETURN,
 	BPF_LSM_MAC,
+	BPF_TRACE_ITER,
 	__MAX_BPF_ATTACH_TYPE
 };
 
Signature · 1 line
2.24.1
#3 lore raw

Re: [PATCH bpf-next v4 02/21] bpf: allow loading of a bpf_iter program

kernel/bpf/verifier.c static int check_return_code(struct bpf_verifier_env *env) · line 7107

Quoted 12 lines · Yonghong Song
kernel/bpf/verifier.c +4 -0
index 70ad009577f8..d725ff7d11db 100644
@@ -7101,6 +7101,10 @@ static int check_return_code(struct bpf_verifier_env *env)
 			return 0;
 		range = tnum_const(0);
 		break;
+	case BPF_PROG_TYPE_TRACING:
+		if (env->prog->expected_attach_type != BPF_TRACE_ITER)
+			return 0;
+		break;

Not related to this set, but I just noticed that I managed to forget to add this check for fentry/fexit/freplace.
While it's not too late let's enforce return 0 for them ? Could you follow up with a patch for bpf tree?

#4 lore raw

Re: [PATCH bpf-next v4 02/21] bpf: allow loading of a bpf_iter program

Quoted 10 lines · Alexei Starovoitov
Quoted 9 lines · Yonghong Song
kernel/bpf/verifier.c +1 -0
index 70ad009577f8..d725ff7d11db 100644
@@ -7101,6 +7101,10 @@ static int check_return_code(struct bpf_verifier_env *env)
  			return 0;
  		range = tnum_const(0);
  		break;
+	case BPF_PROG_TYPE_TRACING:
Quoted 8 lines

kernel/bpf/verifier.c static int check_return_code(struct bpf_verifier_env *env) · line 7107

Quoted 3 lines
kernel/bpf/verifier.c +3 -0
+		if (env->prog->expected_attach_type != BPF_TRACE_ITER)
+			return 0;
+		break;

Not related to this set, but I just noticed that I managed to forget to add this check for fentry/fexit/freplace.
While it's not too late let's enforce return 0 for them ? Could you follow up with a patch for bpf tree?

Sure. I can have a followup patch for this.

#5 lore raw

Re: [PATCH bpf-next v4 02/21] bpf: allow loading of a bpf_iter program

In reply to Alexei Starovoitov
Quoted 10 lines · Alexei Starovoitov
Quoted 9 lines · Yonghong Song
kernel/bpf/verifier.c +1 -0
index 70ad009577f8..d725ff7d11db 100644
@@ -7101,6 +7101,10 @@ static int check_return_code(struct bpf_verifier_env *env)
  			return 0;
  		range = tnum_const(0);
  		break;
+	case BPF_PROG_TYPE_TRACING:
Quoted 8 lines

kernel/bpf/verifier.c static int check_return_code(struct bpf_verifier_env *env) · line 7107

Quoted 3 lines
kernel/bpf/verifier.c +3 -0
+		if (env->prog->expected_attach_type != BPF_TRACE_ITER)
+			return 0;
+		break;

Not related to this set, but I just noticed that I managed to forget to add this check for fentry/fexit/freplace.
While it's not too late let's enforce return 0 for them ? Could you follow up with a patch for bpf tree?

Just want to double check. In selftests, we have

SEC("fentry/__set_task_comm")
int BPF_PROG(prog4, struct task_struct *tsk, const char *buf, bool exec)

{
        return !tsk;
}

SEC("fexit/__set_task_comm")
int BPF_PROG(prog5, struct task_struct *tsk, const char *buf, bool exec)

{
        return !tsk;
}

fentry/fexit may returrn 1. What is the intention here? Does this mean we should allow [0, 1] instead of [0, 0]?

For freplace, we have

__u64 test_get_skb_len = 0;
SEC("freplace/get_skb_len")
int new_get_skb_len(struct __sk_buff *skb)

{
        int len = skb->len;

        if (len != 74)
                return 0;
        test_get_skb_len = 1;
        return 74; /* original get_skb_len() returns skb->len */
}

That means freplace may return arbitrary values depending on what to replace?

#6 lore raw

Re: [PATCH bpf-next v4 02/21] bpf: allow loading of a bpf_iter program

Quoted 30 lines · Yonghong Song
Quoted 10 lines · Alexei Starovoitov
Quoted 9 lines · Yonghong Song
kernel/bpf/verifier.c +1 -0
index 70ad009577f8..d725ff7d11db 100644
@@ -7101,6 +7101,10 @@ static int check_return_code(struct bpf_verifier_env *env)
  			return 0;
  		range = tnum_const(0);
  		break;
+	case BPF_PROG_TYPE_TRACING:
Quoted 8 lines

kernel/bpf/verifier.c static int check_return_code(struct bpf_verifier_env *env) · line 7107

Quoted 3 lines
kernel/bpf/verifier.c +3 -0
+		if (env->prog->expected_attach_type != BPF_TRACE_ITER)
+			return 0;
+		break;

Not related to this set, but I just noticed that I managed to forget to add this check for fentry/fexit/freplace.
While it's not too late let's enforce return 0 for them ? Could you follow up with a patch for bpf tree?

Just want to double check. In selftests, we have

SEC("fentry/__set_task_comm")
int BPF_PROG(prog4, struct task_struct *tsk, const char *buf, bool exec)

{
        return !tsk;
}
Quoted 8 lines

SEC("fexit/__set_task_comm")
int BPF_PROG(prog5, struct task_struct *tsk, const char *buf, bool exec)

{
        return !tsk;
}

fentry/fexit may returrn 1. What is the intention here? Does this mean we should allow [0, 1] instead of [0, 0]?

Argh. I missed that bit when commit ac065870d9282 tweaked the return value. For fentry/exit the return value is ignored by trampoline. imo it's misleading to users and should be rejected by the verifier. so [0,0] for fentry/fexit

Quoted 14 lines

For freplace, we have

__u64 test_get_skb_len = 0;
SEC("freplace/get_skb_len")
int new_get_skb_len(struct __sk_buff *skb)

{
        int len = skb->len;

        if (len != 74)
                return 0;
        test_get_skb_len = 1;
        return 74; /* original get_skb_len() returns skb->len */
}
Quoted 2 lines

That means freplace may return arbitrary values depending on what to replace?

yes. freplace and fmod_ret can return anything.

#7 lore raw

Re: [PATCH bpf-next v4 02/21] bpf: allow loading of a bpf_iter program

Quoted 37 lines · Alexei Starovoitov
Quoted 36 lines · Yonghong Song
Quoted 10 lines · Alexei Starovoitov
Quoted 9 lines · Yonghong Song
kernel/bpf/verifier.c +1 -0
index 70ad009577f8..d725ff7d11db 100644
@@ -7101,6 +7101,10 @@ static int check_return_code(struct bpf_verifier_env *env)
   			return 0;
   		range = tnum_const(0);
   		break;
+	case BPF_PROG_TYPE_TRACING:
Quoted 8 lines

kernel/bpf/verifier.c static int check_return_code(struct bpf_verifier_env *env) · line 7107

Quoted 3 lines
kernel/bpf/verifier.c +3 -0
+		if (env->prog->expected_attach_type != BPF_TRACE_ITER)
+			return 0;
+		break;

Not related to this set, but I just noticed that I managed to forget to add this check for fentry/fexit/freplace.
While it's not too late let's enforce return 0 for them ? Could you follow up with a patch for bpf tree?

Just want to double check. In selftests, we have

SEC("fentry/__set_task_comm")
int BPF_PROG(prog4, struct task_struct *tsk, const char *buf, bool exec)

{
         return !tsk;
}

SEC("fexit/__set_task_comm")
int BPF_PROG(prog5, struct task_struct *tsk, const char *buf, bool exec)

{
         return !tsk;
}
Quoted 7 lines
Quoted 2 lines

fentry/fexit may returrn 1. What is the intention here? Does this mean we should allow [0, 1] instead of [0, 0]?

Argh. I missed that bit when commit ac065870d9282 tweaked the return value. For fentry/exit the return value is ignored by trampoline. imo it's misleading to users and should be rejected by the verifier. so [0,0] for fentry/fexit

Sounds good. Will craft patch to enforce fentry/fexit with [0,0] then. Thanks!

Quoted 20 lines
Quoted 15 lines

For freplace, we have

__u64 test_get_skb_len = 0;
SEC("freplace/get_skb_len")
int new_get_skb_len(struct __sk_buff *skb)

{
         int len = skb->len;

         if (len != 74)
                 return 0;
         test_get_skb_len = 1;
         return 74; /* original get_skb_len() returns skb->len */
}
Quoted 2 lines

That means freplace may return arbitrary values depending on what to replace?

yes. freplace and fmod_ret can return anything.

#8 lore raw

PATCH v4 3/21 bpf: support bpf tracing/iter programs for BPF_LINK_CREATE

In reply to Yonghong Song

Given a bpf program, the step to create an anonymous bpf iterator is:
- create a bpf_iter_link, which combines bpf program and the target. In the future, there could be more information recorded in the link. A link_fd will be returned to the user space.
- create an anonymous bpf iterator with the given link_fd.

The bpf_iter_link can be pinned to bpffs mount file system to create a file based bpf iterator as well.

The benefit to use of bpf_iter_link:
- using bpf link simplifies design and implementation as bpf link is used for other tracing bpf programs.
- for file based bpf iterator, bpf_iter_link provides a standard way to replace underlying bpf programs.
- for both anonymous and free based iterators, bpf link query capability can be leveraged.

The patch added support of tracing/iter programs for BPF_LINK_CREATE. A new link type BPF_LINK_TYPE_ITER is added to facilitate link querying. Currently, only prog_id is needed, so there is no additional in-kernel show_fdinfo() and fill_link_info() hook is needed for BPF_LINK_TYPE_ITER link.

---
 include/linux/bpf.h            |  1 +
 include/linux/bpf_types.h      |  1 +
 include/uapi/linux/bpf.h       |  1 +
 kernel/bpf/bpf_iter.c          | 62 ++++++++++++++++++++++++++++++++++
 kernel/bpf/syscall.c           | 14 ++++++++
 tools/include/uapi/linux/bpf.h |  1 +
 6 files changed, 80 insertions(+)
include/linux/bpf.h +1 -0
index f28bdd714754..e93d2d33c82c 100644
@@ -1143,6 +1143,7 @@ struct bpf_iter_reg {
 int bpf_iter_reg_target(struct bpf_iter_reg *reg_info);
 void bpf_iter_unreg_target(const char *target);
 bool bpf_iter_prog_supported(struct bpf_prog *prog);
+int bpf_iter_link_attach(const union bpf_attr *attr, struct bpf_prog *prog);
 
 int bpf_percpu_hash_copy(struct bpf_map *map, void *key, void *value);
 int bpf_percpu_array_copy(struct bpf_map *map, void *key, void *value);
include/linux/bpf_types.h +1 -0
index 8345cdf553b8..29d22752fc87 100644
@@ -124,3 +124,4 @@ BPF_LINK_TYPE(BPF_LINK_TYPE_TRACING, tracing)
 #ifdef CONFIG_CGROUP_BPF
 BPF_LINK_TYPE(BPF_LINK_TYPE_CGROUP, cgroup)
 #endif
+BPF_LINK_TYPE(BPF_LINK_TYPE_ITER, iter)
include/uapi/linux/bpf.h +1 -0
index c8a5325cc8d0..1e8dfff5d5d4 100644
@@ -229,6 +229,7 @@ enum bpf_link_type {
 	BPF_LINK_TYPE_RAW_TRACEPOINT = 1,
 	BPF_LINK_TYPE_TRACING = 2,
 	BPF_LINK_TYPE_CGROUP = 3,
+	BPF_LINK_TYPE_ITER = 4,
 
 	MAX_BPF_LINK_TYPE,
 };
kernel/bpf/bpf_iter.c +62 -0

71 lines not on this page · raw message

kernel/bpf/syscall.c +14 -0
index bb1ab7da6103..6ffe2d8fb6c7 100644
@@ -2729,6 +2729,8 @@ attach_type_to_prog_type(enum bpf_attach_type attach_type)
 	case BPF_CGROUP_GETSOCKOPT:
 	case BPF_CGROUP_SETSOCKOPT:
 		return BPF_PROG_TYPE_CGROUP_SOCKOPT;
+	case BPF_TRACE_ITER:
+		return BPF_PROG_TYPE_TRACING;
 	default:
 		return BPF_PROG_TYPE_UNSPEC;
 	}
@@ -3729,6 +3731,15 @@ static int bpf_map_do_batch(const union bpf_attr *attr,
 	return err;
 }
 
+static int tracing_bpf_link_attach(const union bpf_attr *attr, struct bpf_prog *prog)
+{
+	if (attr->link_create.attach_type == BPF_TRACE_ITER &&
+	    prog->expected_attach_type == BPF_TRACE_ITER)
+		return bpf_iter_link_attach(attr, prog);
+
+	return -EINVAL;
+}
+
 #define BPF_LINK_CREATE_LAST_FIELD link_create.flags
 static int link_create(union bpf_attr *attr)
 {
@@ -3765,6 +3776,9 @@ static int link_create(union bpf_attr *attr)
 	case BPF_PROG_TYPE_CGROUP_SOCKOPT:
 		ret = cgroup_bpf_link_attach(attr, prog);
 		break;
+	case BPF_PROG_TYPE_TRACING:
+		ret = tracing_bpf_link_attach(attr, prog);
+		break;
 	default:
 		ret = -EINVAL;
 	}
tools/include/uapi/linux/bpf.h +1 -0
index c8a5325cc8d0..1e8dfff5d5d4 100644
@@ -229,6 +229,7 @@ enum bpf_link_type {
 	BPF_LINK_TYPE_RAW_TRACEPOINT = 1,
 	BPF_LINK_TYPE_TRACING = 2,
 	BPF_LINK_TYPE_CGROUP = 3,
+	BPF_LINK_TYPE_ITER = 4,
 
 	MAX_BPF_LINK_TYPE,
 };
Signature · 1 line
2.24.1
#9 lore raw

PATCH v4 4/21 bpf: support bpf tracing/iter programs for BPF_LINK_UPDATE

In reply to Yonghong Song

Added BPF_LINK_UPDATE support for tracing/iter programs. This way, a file based bpf iterator, which holds a reference to the link, can have its bpf program updated without creating new files.

---
 kernel/bpf/bpf_iter.c | 31 +++++++++++++++++++++++++++++++
 1 file changed, 31 insertions(+)
kernel/bpf/bpf_iter.c +31 -0
index 03f5832909db..0542a243b78c 100644
@@ -23,6 +23,9 @@ struct bpf_iter_link {
 static struct list_head targets = LIST_HEAD_INIT(targets);
 static DEFINE_MUTEX(targets_mutex);
 
+/* protect bpf_iter_link changes */
+static DEFINE_MUTEX(link_mutex);
+
 int bpf_iter_reg_target(struct bpf_iter_reg *reg_info)
 {
 	struct bpf_iter_target_info *tinfo;
@@ -111,9 +114,37 @@ static void bpf_iter_link_dealloc(struct bpf_link *link)
 	kfree(iter_link);
 }
 
+static int bpf_iter_link_replace(struct bpf_link *link,
+				 struct bpf_prog *new_prog,
+				 struct bpf_prog *old_prog)
+{
+	int ret = 0;
+
+	mutex_lock(&link_mutex);
+	if (old_prog && link->prog != old_prog) {
+		ret = -EPERM;
+		goto out_unlock;
+	}
+
+	if (link->prog->type != new_prog->type ||
+	    link->prog->expected_attach_type != new_prog->expected_attach_type ||
+	    link->prog->aux->attach_btf_id != new_prog->aux->attach_btf_id) {
+		ret = -EINVAL;
+		goto out_unlock;
+	}
+
+	old_prog = xchg(&link->prog, new_prog);
+	bpf_prog_put(old_prog);
+
+out_unlock:
+	mutex_unlock(&link_mutex);
+	return ret;
+}
+
 static const struct bpf_link_ops bpf_iter_link_lops = {
 	.release = bpf_iter_link_release,
 	.dealloc = bpf_iter_link_dealloc,
+	.update_prog = bpf_iter_link_replace,
 };
 
 int bpf_iter_link_attach(const union bpf_attr *attr, struct bpf_prog *prog)
Signature · 1 line
2.24.1
#10 lore raw

PATCH v4 5/21 bpf: implement bpf_seq_read() for bpf iterator

In reply to Yonghong Song

bpf iterator uses seq_file to provide a lossless way to transfer data to user space. But we want to call bpf program after all objects have been traversed, and bpf program may write additional data to the seq_file buffer. The current seq_read() does not work for this use case.

Besides allowing stop() function to write to the buffer, the bpf_seq_read() also fixed the buffer size to one page. If any single call of show() or stop() will emit data more than one page to cause overflow, -E2BIG error code will be returned to user space.

---
 kernel/bpf/bpf_iter.c | 123 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 123 insertions(+)
kernel/bpf/bpf_iter.c +123 -0
index 0542a243b78c..832973ee80fa 100644
@@ -26,6 +26,129 @@ static DEFINE_MUTEX(targets_mutex);
 /* protect bpf_iter_link changes */
 static DEFINE_MUTEX(link_mutex);
 
+/* bpf_seq_read, a customized and simpler version for bpf iterator.
+ * no_llseek is assumed for this file.
+ * The following are differences from seq_read():
+ *  . fixed buffer size (PAGE_SIZE)
+ *  . assuming no_llseek
+ *  . stop() may call bpf program, handling potential overflow there
+ */
+static ssize_t bpf_seq_read(struct file *file, char __user *buf, size_t size,
+			    loff_t *ppos)
+{
+	struct seq_file *seq = file->private_data;
+	size_t n, offs, copied = 0;
+	int err = 0;
+	void *p;
+
+	mutex_lock(&seq->lock);
+
+	if (!seq->buf) {
+		seq->size = PAGE_SIZE;
+		seq->buf = kmalloc(seq->size, GFP_KERNEL);
+		if (!seq->buf) {
+			err = -ENOMEM;
+			goto done;
+		}
+	}
+
+	if (seq->count) {
+		n = min(seq->count, size);
+		err = copy_to_user(buf, seq->buf + seq->from, n);
+		if (err) {
+			err = -EFAULT;
+			goto done;
+		}
+		seq->count -= n;
+		seq->from += n;
+		copied = n;
+		goto done;
+	}
+
+	seq->from = 0;
+	p = seq->op->start(seq, &seq->index);
+	if (!p)
+		goto stop;
+	if (IS_ERR(p)) {
+		err = PTR_ERR(p);
+		seq->op->stop(seq, p);
+		seq->count = 0;
+		goto done;
+	}
+
+	err = seq->op->show(seq, p);
+	if (err > 0) {
+		seq->count = 0;
+	} else if (err < 0 || seq_has_overflowed(seq)) {
+		if (!err)
+			err = -E2BIG;
+		seq->op->stop(seq, p);
+		seq->count = 0;
+		goto done;
+	}
+
+	while (1) {
+		loff_t pos = seq->index;
+
+		offs = seq->count;
+		p = seq->op->next(seq, p, &seq->index);
+		if (pos == seq->index) {
+			pr_info_ratelimited("buggy seq_file .next function %ps "
+				"did not updated position index\n",
+				seq->op->next);
+			seq->index++;
+		}
+
+		if (IS_ERR_OR_NULL(p))
+			break;
+
+		if (seq->count >= size)
+			break;
+
+		err = seq->op->show(seq, p);
+		if (err > 0) {
+			seq->count = offs;
+		} else if (err < 0 || seq_has_overflowed(seq)) {
+			seq->count = offs;
+			if (offs == 0) {
+				if (!err)
+					err = -E2BIG;
+				seq->op->stop(seq, p);
+				goto done;
+			}
+			break;
+		}
+	}
+stop:
+	offs = seq->count;
+	/* bpf program called if !p */
+	seq->op->stop(seq, p);
+	if (!p && seq_has_overflowed(seq)) {
+		seq->count = offs;
+		if (offs == 0) {
+			err = -E2BIG;
+			goto done;
+		}
+	}
+
+	n = min(seq->count, size);
+	err = copy_to_user(buf, seq->buf, n);
+	if (err) {
+		err = -EFAULT;
+		goto done;
+	}
+	copied = n;
+	seq->count -= n;
+	seq->from = n;
+done:
+	if (!copied)
+		copied = err;
+	else
+		*ppos += copied;
+	mutex_unlock(&seq->lock);
+	return copied;
+}
+
 int bpf_iter_reg_target(struct bpf_iter_reg *reg_info)
 {
 	struct bpf_iter_target_info *tinfo;
Signature · 1 line
2.24.1
#11 lore raw

Re: [PATCH bpf-next v4 05/21] bpf: implement bpf_seq_read() for bpf iterator

kernel/bpf/bpf_iter.c static DEFINE_MUTEX(targets_mutex); · line 110

Quoted 4 lines · Yonghong Song
kernel/bpf/bpf_iter.c +4 -0
+
+		err = seq->op->show(seq, p);
+		if (err > 0) {
+			seq->count = offs;

as far as I can see this condition can never happen.
I understand that seq_read() has this logic, but four iterators implemented don't exercise this path.
I guess it's ok to keep it, but may be add warn_once so we notice when things change?

#12 lore raw

Re: [PATCH bpf-next v4 05/21] bpf: implement bpf_seq_read() for bpf iterator

Quoted 11 lines · Alexei Starovoitov

kernel/bpf/bpf_iter.c static DEFINE_MUTEX(targets_mutex); · line 110

Quoted 4 lines · Yonghong Song
kernel/bpf/bpf_iter.c +4 -0
+
+		err = seq->op->show(seq, p);
+		if (err > 0) {
+			seq->count = offs;

as far as I can see this condition can never happen.
I understand that seq_read() has this logic, but four iterators implemented don't exercise this path.
I guess it's ok to keep it, but may be add warn_once so we notice when things change?

Yes, it won't happen with our current bpf return values. I keep it to be compatible with seq_read() and for potential future use in case we want to enable this.
we will add a warn_once here.

#13 lore raw

PATCH v4 6/21 bpf: create anonymous bpf iterator

In reply to Yonghong Song

A new bpf command BPF_ITER_CREATE is added.

The anonymous bpf iterator is seq_file based.
The seq_file private data are referenced by targets. The bpf_iter infrastructure allocated additional space at seq_file->private before the space used by targets to store some meta data, e.g.,
prog: prog to run
session_id: an unique id for each opened seq_file
seq_num: how many times bpf programs are queried in this session
done_stop: an internal state to decide whether bpf program

              should be called in seq_ops->stop() or not

The seq_num will start from 0 for valid objects. The bpf program may see the same seq_num more than once if
- seq_file buffer overflow happens and the same object is retried by bpf_seq_read(), or
- the bpf program explicitly requests a retry of the same object

Since module is not supported for bpf_iter, all target registeration happens at __init time, so there is no need to change bpf_iter_unreg_target() as it is used mostly in error path of the init function at which time no bpf iterators have been created yet.

---
 include/linux/bpf.h            |   1 +
 include/uapi/linux/bpf.h       |   6 ++
 kernel/bpf/bpf_iter.c          | 129 +++++++++++++++++++++++++++++++++
 kernel/bpf/syscall.c           |  26 +++++++
 tools/include/uapi/linux/bpf.h |   6 ++
 5 files changed, 168 insertions(+)
include/linux/bpf.h +1 -0
index e93d2d33c82c..80b1b9d8a638 100644
@@ -1144,6 +1144,7 @@ int bpf_iter_reg_target(struct bpf_iter_reg *reg_info);
 void bpf_iter_unreg_target(const char *target);
 bool bpf_iter_prog_supported(struct bpf_prog *prog);
 int bpf_iter_link_attach(const union bpf_attr *attr, struct bpf_prog *prog);
+int bpf_iter_new_fd(struct bpf_link *link);
 
 int bpf_percpu_hash_copy(struct bpf_map *map, void *key, void *value);
 int bpf_percpu_array_copy(struct bpf_map *map, void *key, void *value);
include/uapi/linux/bpf.h +6 -0
index 1e8dfff5d5d4..708763f702e1 100644
@@ -116,6 +116,7 @@ enum bpf_cmd {
 	BPF_LINK_GET_FD_BY_ID,
 	BPF_LINK_GET_NEXT_ID,
 	BPF_ENABLE_STATS,
+	BPF_ITER_CREATE,
 };
 
 enum bpf_map_type {
@@ -614,6 +615,11 @@ union bpf_attr {
 		__u32		type;
 	} enable_stats;
 
+	struct { /* struct used by BPF_ITER_CREATE command */
+		__u32		link_fd;
+		__u32		flags;
+	} iter_create;
+
 } __attribute__((aligned(8)));
 
 /* The description below is an attempt at providing documentation to eBPF
kernel/bpf/bpf_iter.c +129 -0

156 lines not on this page · raw message

kernel/bpf/syscall.c +26 -0
index 6ffe2d8fb6c7..a293e88ee01a 100644
@@ -3941,6 +3941,29 @@ static int bpf_enable_stats(union bpf_attr *attr)
 	return -EINVAL;
 }
 
+#define BPF_ITER_CREATE_LAST_FIELD iter_create.flags
+
+static int bpf_iter_create(union bpf_attr *attr)
+{
+	struct bpf_link *link;
+	int err;
+
+	if (CHECK_ATTR(BPF_ITER_CREATE))
+		return -EINVAL;
+
+	if (attr->iter_create.flags)
+		return -EINVAL;
+
+	link = bpf_link_get_from_fd(attr->iter_create.link_fd);
+	if (IS_ERR(link))
+		return PTR_ERR(link);
+
+	err = bpf_iter_new_fd(link);
+	bpf_link_put(link);
+
+	return err;
+}
+
 SYSCALL_DEFINE3(bpf, int, cmd, union bpf_attr __user *, uattr, unsigned int, size)
 {
 	union bpf_attr attr;
@@ -4068,6 +4091,9 @@ SYSCALL_DEFINE3(bpf, int, cmd, union bpf_attr __user *, uattr, unsigned int, siz
 	case BPF_ENABLE_STATS:
 		err = bpf_enable_stats(&attr);
 		break;
+	case BPF_ITER_CREATE:
+		err = bpf_iter_create(&attr);
+		break;
 	default:
 		err = -EINVAL;
 		break;
tools/include/uapi/linux/bpf.h +6 -0
index 1e8dfff5d5d4..708763f702e1 100644
@@ -116,6 +116,7 @@ enum bpf_cmd {
 	BPF_LINK_GET_FD_BY_ID,
 	BPF_LINK_GET_NEXT_ID,
 	BPF_ENABLE_STATS,
+	BPF_ITER_CREATE,
 };
 
 enum bpf_map_type {
@@ -614,6 +615,11 @@ union bpf_attr {
 		__u32		type;
 	} enable_stats;
 
+	struct { /* struct used by BPF_ITER_CREATE command */
+		__u32		link_fd;
+		__u32		flags;
+	} iter_create;
+
 } __attribute__((aligned(8)));
 
 /* The description below is an attempt at providing documentation to eBPF
Signature · 1 line
2.24.1
#14 lore raw

PATCH v4 7/21 bpf: create file bpf iterator

In reply to Yonghong Song

To produce a file bpf iterator, the fd must be corresponding to a link_fd assocciated with a trace/iter program. When the pinned file is opened, a seq_file will be generated.

---
 include/linux/bpf.h   |  2 ++
 kernel/bpf/bpf_iter.c | 17 ++++++++++++++++-
 kernel/bpf/inode.c    |  5 ++++-
 3 files changed, 22 insertions(+), 2 deletions(-)
include/linux/bpf.h +2 -0
index 80b1b9d8a638..b06653ab3476 100644
@@ -1022,6 +1022,7 @@ static inline void bpf_enable_instrumentation(void)
 
 extern const struct file_operations bpf_map_fops;
 extern const struct file_operations bpf_prog_fops;
+extern const struct file_operations bpf_iter_fops;
 
 #define BPF_PROG_TYPE(_id, _name, prog_ctx_type, kern_ctx_type) \
 	extern const struct bpf_prog_ops _name ## _prog_ops; \
@@ -1145,6 +1146,7 @@ void bpf_iter_unreg_target(const char *target);
 bool bpf_iter_prog_supported(struct bpf_prog *prog);
 int bpf_iter_link_attach(const union bpf_attr *attr, struct bpf_prog *prog);
 int bpf_iter_new_fd(struct bpf_link *link);
+bool bpf_link_is_iter(struct bpf_link *link);
 
 int bpf_percpu_hash_copy(struct bpf_map *map, void *key, void *value);
 int bpf_percpu_array_copy(struct bpf_map *map, void *key, void *value);
kernel/bpf/bpf_iter.c +16 -1
index e7129b57865f..090f09b0eacb 100644
@@ -39,6 +39,8 @@ static DEFINE_MUTEX(link_mutex);
 /* incremented on every opened seq_file */
 static atomic64_t session_id;
 
+static int prepare_seq_file(struct file *file, struct bpf_iter_link *link);
+
 /* bpf_seq_read, a customized and simpler version for bpf iterator.
  * no_llseek is assumed for this file.
  * The following are differences from seq_read():
@@ -162,6 +164,13 @@ static ssize_t bpf_seq_read(struct file *file, char __user *buf, size_t size,
 	return copied;
 }
 
+static int iter_open(struct inode *inode, struct file *file)
+{
+	struct bpf_iter_link *link = inode->i_private;
+
+	return prepare_seq_file(file, link);
+}
+
 static int iter_release(struct inode *inode, struct file *file)
 {
 	struct bpf_iter_priv_data *iter_priv;
@@ -183,7 +192,8 @@ static int iter_release(struct inode *inode, struct file *file)
 	return seq_release_private(inode, file);
 }
 
-static const struct file_operations bpf_iter_fops = {
+const struct file_operations bpf_iter_fops = {
+	.open		= iter_open,
 	.llseek		= no_llseek,
 	.read		= bpf_seq_read,
 	.release	= iter_release,
@@ -310,6 +320,11 @@ static const struct bpf_link_ops bpf_iter_link_lops = {
 	.update_prog = bpf_iter_link_replace,
 };
 
+bool bpf_link_is_iter(struct bpf_link *link)
+{
+	return link->ops == &bpf_iter_link_lops;
+}
+
 int bpf_iter_link_attach(const union bpf_attr *attr, struct bpf_prog *prog)
 {
 	struct bpf_link_primer link_primer;
kernel/bpf/inode.c +4 -1
index 95087d9f4ed3..fb878ba3f22f 100644
@@ -358,8 +358,11 @@ static int bpf_mkmap(struct dentry *dentry, umode_t mode, void *arg)
 
 static int bpf_mklink(struct dentry *dentry, umode_t mode, void *arg)
 {
+	struct bpf_link *link = arg;
+
 	return bpf_mkobj_ops(dentry, mode, arg, &bpf_link_iops,
-			     &bpffs_obj_fops);
+			     bpf_link_is_iter(link) ?
+			     &bpf_iter_fops : &bpffs_obj_fops);
 }
 
 static struct dentry *
Signature · 1 line
2.24.1
#15 lore raw

PATCH v4 8/21 bpf: implement common macros/helpers for target iterators

In reply to Yonghong Song

Macro DEFINE_BPF_ITER_FUNC is implemented so target can define an init function to capture the BTF type which represents the target.

The bpf_iter_meta is a structure holding meta data, common to all targets in the bpf program.

Additional marker functions are called before or after bpf_seq_read() show()/next()/stop() callback functions to help calculate precise seq_num and whether call bpf_prog inside stop().

Two functions, bpf_iter_get_info() and bpf_iter_run_prog(), are implemented so target can get needed information from bpf_iter infrastructure and can run the program.

---
 include/linux/bpf.h   | 11 ++++++
 kernel/bpf/bpf_iter.c | 85 ++++++++++++++++++++++++++++++++++++++++---
 2 files changed, 91 insertions(+), 5 deletions(-)
include/linux/bpf.h +11 -0
index b06653ab3476..ffe0b9b669bf 100644
@@ -1129,6 +1129,9 @@ int bpf_obj_pin_user(u32 ufd, const char __user *pathname);
 int bpf_obj_get_user(const char __user *pathname, int flags);
 
 #define BPF_ITER_FUNC_PREFIX "__bpf_iter__"
+#define DEFINE_BPF_ITER_FUNC(target, args...)			\
+	extern int __bpf_iter__ ## target(args);		\
+	int __init __bpf_iter__ ## target(args) { return 0; }
 
 typedef int (*bpf_iter_init_seq_priv_t)(void *private_data);
 typedef void (*bpf_iter_fini_seq_priv_t)(void *private_data);
@@ -1141,12 +1144,20 @@ struct bpf_iter_reg {
 	u32 seq_priv_size;
 };
 
+struct bpf_iter_meta {
+	__bpf_md_ptr(struct seq_file *, seq);
+	u64 session_id;
+	u64 seq_num;
+};
+
 int bpf_iter_reg_target(struct bpf_iter_reg *reg_info);
 void bpf_iter_unreg_target(const char *target);
 bool bpf_iter_prog_supported(struct bpf_prog *prog);
 int bpf_iter_link_attach(const union bpf_attr *attr, struct bpf_prog *prog);
 int bpf_iter_new_fd(struct bpf_link *link);
 bool bpf_link_is_iter(struct bpf_link *link);
+struct bpf_prog *bpf_iter_get_info(struct bpf_iter_meta *meta, bool in_stop);
+int bpf_iter_run_prog(struct bpf_prog *prog, void *ctx);
 
 int bpf_percpu_hash_copy(struct bpf_map *map, void *key, void *value);
 int bpf_percpu_array_copy(struct bpf_map *map, void *key, void *value);
kernel/bpf/bpf_iter.c +80 -5

117 lines not on this page · raw message

Signature · 1 line
2.24.1
#16 lore raw

PATCH v4 9/21 bpf: add bpf_map iterator

In reply to Yonghong Song

Implement seq_file operations to traverse all bpf_maps.

---
 include/linux/bpf.h   |  1 +
 kernel/bpf/Makefile   |  2 +-
 kernel/bpf/map_iter.c | 97 +++++++++++++++++++++++++++++++++++++++++++
 kernel/bpf/syscall.c  | 19 +++++++++
 4 files changed, 118 insertions(+), 1 deletion(-)
 create mode 100644 kernel/bpf/map_iter.c
include/linux/bpf.h +1 -0
index ffe0b9b669bf..363ab0751967 100644
@@ -1082,6 +1082,7 @@ int  generic_map_update_batch(struct bpf_map *map,
 int  generic_map_delete_batch(struct bpf_map *map,
 			      const union bpf_attr *attr,
 			      union bpf_attr __user *uattr);
+struct bpf_map *bpf_map_get_curr_or_next(u32 *id);
 
 extern int sysctl_unprivileged_bpf_disabled;
 
kernel/bpf/Makefile +1 -1
index 6a8b0febd3f6..b2b5eefc5254 100644
@@ -2,7 +2,7 @@
 obj-y := core.o
 CFLAGS_core.o += $(call cc-disable-warning, override-init)
 
-obj-$(CONFIG_BPF_SYSCALL) += syscall.o verifier.o inode.o helpers.o tnum.o bpf_iter.o
+obj-$(CONFIG_BPF_SYSCALL) += syscall.o verifier.o inode.o helpers.o tnum.o bpf_iter.o map_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
 obj-$(CONFIG_BPF_SYSCALL) += disasm.o
kernel/bpf/map_iter.c new file +97 -0

97 lines not on this page · raw message

kernel/bpf/syscall.c +19 -0
index a293e88ee01a..de2a75500233 100644
@@ -2934,6 +2934,25 @@ static int bpf_obj_get_next_id(const union bpf_attr *attr,
 	return err;
 }
 
+struct bpf_map *bpf_map_get_curr_or_next(u32 *id)
+{
+	struct bpf_map *map;
+
+	spin_lock_bh(&map_idr_lock);
+again:
+	map = idr_get_next(&map_idr, id);
+	if (map) {
+		map = __bpf_map_inc_not_zero(map, false);
+		if (IS_ERR(map)) {
+			(*id)++;
+			goto again;
+		}
+	}
+	spin_unlock_bh(&map_idr_lock);
+
+	return map;
+}
+
 #define BPF_PROG_GET_FD_BY_ID_LAST_FIELD prog_id
 
 struct bpf_prog *bpf_prog_by_id(u32 id)
Signature · 1 line
2.24.1
#17 lore raw

PATCH v4 10/21 net: bpf: add netlink and ipv6_route bpf_iter targets

In reply to Yonghong Song

This patch added netlink and ipv6_route targets, using the same seq_ops (except show() and minor changes for stop()) for /proc/net/{netlink,ipv6_route}.

The net namespace for these targets are the current net namespace at file open stage, similar to /proc/net/{netlink,ipv6_route} reference counting the net namespace at seq_file open stage.

Since module is not supported for now, ipv6_route is supported only if the IPV6 is built-in, i.e., not compiled as a module. The restriction can be lifted once module is properly supported for bpf_iter.

---
 fs/proc/proc_net.c       | 19 +++++++++
 include/linux/proc_fs.h  |  3 ++
 net/ipv6/ip6_fib.c       | 65 +++++++++++++++++++++++++++++-
 net/ipv6/route.c         | 37 +++++++++++++++++
 net/netlink/af_netlink.c | 87 +++++++++++++++++++++++++++++++++++++++-
 5 files changed, 207 insertions(+), 4 deletions(-)
fs/proc/proc_net.c +19 -0
index 4888c5224442..dba63b2429f0 100644
@@ -98,6 +98,25 @@ static const struct proc_ops proc_net_seq_ops = {
 	.proc_release	= seq_release_net,
 };
 
+int bpf_iter_init_seq_net(void *priv_data)
+{
+#ifdef CONFIG_NET_NS
+	struct seq_net_private *p = priv_data;
+
+	p->net = get_net(current->nsproxy->net_ns);
+#endif
+	return 0;
+}
+
+void bpf_iter_fini_seq_net(void *priv_data)
+{
+#ifdef CONFIG_NET_NS
+	struct seq_net_private *p = priv_data;
+
+	put_net(p->net);
+#endif
+}
+
 struct proc_dir_entry *proc_create_net_data(const char *name, umode_t mode,
 		struct proc_dir_entry *parent, const struct seq_operations *ops,
 		unsigned int state_size, void *data)
include/linux/proc_fs.h +3 -0
index 45c05fd9c99d..03953c59807d 100644
@@ -105,6 +105,9 @@ struct proc_dir_entry *proc_create_net_single_write(const char *name, umode_t mo
 						    void *data);
 extern struct pid *tgid_pidfd_to_pid(const struct file *file);
 
+extern int bpf_iter_init_seq_net(void *priv_data);
+extern void bpf_iter_fini_seq_net(void *priv_data);
+
 #ifdef CONFIG_PROC_PID_ARCH_STATUS
 /*
  * The architecture which selects CONFIG_PROC_PID_ARCH_STATUS must
net/ipv6/ip6_fib.c +63 -2

83 lines not on this page · raw message

net/ipv6/route.c +37 -0
index 3912aac7854d..25f6d3e619d0 100644
@@ -6393,6 +6393,30 @@ void __init ip6_route_init_special_entries(void)
   #endif
 }
 
+#if IS_BUILTIN(CONFIG_IPV6)
+#if defined(CONFIG_BPF_SYSCALL) && defined(CONFIG_PROC_FS)
+DEFINE_BPF_ITER_FUNC(ipv6_route, struct bpf_iter_meta *meta, struct fib6_info *rt)
+
+static int __init bpf_iter_register(void)
+{
+	struct bpf_iter_reg reg_info = {
+		.target			= "ipv6_route",
+		.seq_ops		= &ipv6_route_seq_ops,
+		.init_seq_private	= bpf_iter_init_seq_net,
+		.fini_seq_private	= bpf_iter_fini_seq_net,
+		.seq_priv_size		= sizeof(struct ipv6_route_iter),
+	};
+
+	return bpf_iter_reg_target(&reg_info);
+}
+
+static void bpf_iter_unregister(void)
+{
+	bpf_iter_unreg_target("ipv6_route");
+}
+#endif
+#endif
+
 int __init ip6_route_init(void)
 {
 	int ret;
@@ -6455,6 +6479,14 @@ int __init ip6_route_init(void)
 	if (ret)
 		goto out_register_late_subsys;
 
+#if IS_BUILTIN(CONFIG_IPV6)
+#if defined(CONFIG_BPF_SYSCALL) && defined(CONFIG_PROC_FS)
+	ret = bpf_iter_register();
+	if (ret)
+		goto out_register_late_subsys;
+#endif
+#endif
+
 	for_each_possible_cpu(cpu) {
 		struct uncached_list *ul = per_cpu_ptr(&rt6_uncached_list, cpu);
 
@@ -6487,6 +6519,11 @@ int __init ip6_route_init(void)
 
 void ip6_route_cleanup(void)
 {
+#if IS_BUILTIN(CONFIG_IPV6)
+#if defined(CONFIG_BPF_SYSCALL) && defined(CONFIG_PROC_FS)
+	bpf_iter_unregister();
+#endif
+#endif
 	unregister_netdevice_notifier(&ip6_route_dev_notifier);
 	unregister_pernet_subsys(&ip6_route_net_late_ops);
 	fib6_rules_cleanup();
net/netlink/af_netlink.c +85 -2

117 lines not on this page · raw message

Signature · 1 line
2.24.1
#18 lore raw

PATCH v4 11/21 bpf: add task and task/file iterator targets

In reply to Yonghong Song

Only the tasks belonging to "current" pid namespace are enumerated.

For task/file target, the bpf program will have access to
struct task_struct *task
u32 fd
struct file *file
where fd/file is an open file for the task.

---
 kernel/bpf/Makefile    |   2 +-
 kernel/bpf/task_iter.c | 333 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 334 insertions(+), 1 deletion(-)
 create mode 100644 kernel/bpf/task_iter.c
kernel/bpf/Makefile +1 -1
index b2b5eefc5254..37b2d8620153 100644
@@ -2,7 +2,7 @@
 obj-y := core.o
 CFLAGS_core.o += $(call cc-disable-warning, override-init)
 
-obj-$(CONFIG_BPF_SYSCALL) += syscall.o verifier.o inode.o helpers.o tnum.o bpf_iter.o map_iter.o
+obj-$(CONFIG_BPF_SYSCALL) += syscall.o verifier.o inode.o helpers.o tnum.o bpf_iter.o map_iter.o task_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
 obj-$(CONFIG_BPF_SYSCALL) += disasm.o
kernel/bpf/task_iter.c new file +333 -0

333 lines not on this page · raw message

Signature · 1 line
2.24.1
#19 lore raw

PATCH v4 12/21 bpf: add PTR_TO_BTF_ID_OR_NULL support

In reply to Yonghong Song

Add bpf_reg_type PTR_TO_BTF_ID_OR_NULL support. For tracing/iter program, the bpf program context definition, e.g., for previous bpf_map target, looks like

  struct bpf_iter__bpf_map {
    struct bpf_iter_meta *meta;
    struct bpf_map *map;

};

The kernel guarantees that meta is not NULL, but map pointer maybe NULL. The NULL map indicates that all objects have been traversed, so bpf program can take proper action, e.g., do final aggregation and/or send final report to user space.

Add btf_id_or_null_non0_off to prog->aux structure, to indicate that if the context access offset is not 0, set to PTR_TO_BTF_ID_OR_NULL instead of PTR_TO_BTF_ID. This bit is set for tracing/iter program.

---
 include/linux/bpf.h   |  2 ++
 kernel/bpf/btf.c      |  5 ++++-
 kernel/bpf/verifier.c | 16 ++++++++++++----
 3 files changed, 18 insertions(+), 5 deletions(-)
include/linux/bpf.h +2 -0
index 363ab0751967..cf4b6e44f2bc 100644
@@ -320,6 +320,7 @@ enum bpf_reg_type {
 	PTR_TO_TP_BUFFER,	 /* reg points to a writable raw tp's buffer */
 	PTR_TO_XDP_SOCK,	 /* reg points to struct xdp_sock */
 	PTR_TO_BTF_ID,		 /* reg points to kernel struct */
+	PTR_TO_BTF_ID_OR_NULL,	 /* reg points to kernel struct or NULL */
 };
 
 /* The information passed from prog-specific *_is_valid_access
@@ -658,6 +659,7 @@ struct bpf_prog_aux {
 	bool offload_requested;
 	bool attach_btf_trace; /* true if attaching to BTF-enabled raw tp */
 	bool func_proto_unreliable;
+	bool btf_id_or_null_non0_off;
 	enum bpf_tramp_prog_type trampoline_prog_type;
 	struct bpf_trampoline *trampoline;
 	struct hlist_node tramp_hlist;
kernel/bpf/btf.c +4 -1
index a2cfba89a8e1..c490fbde22d4 100644
@@ -3790,7 +3790,10 @@ bool btf_ctx_access(int off, int size, enum bpf_access_type type,
 		return true;
 
 	/* this is a pointer to another type */
-	info->reg_type = PTR_TO_BTF_ID;
+	if (off != 0 && prog->aux->btf_id_or_null_non0_off)
+		info->reg_type = PTR_TO_BTF_ID_OR_NULL;
+	else
+		info->reg_type = PTR_TO_BTF_ID;
 
 	if (tgt_prog) {
 		ret = btf_translate_to_vmlinux(log, btf, t, tgt_prog->type, arg);
kernel/bpf/verifier.c +12 -4
index d725ff7d11db..36b2a38a06fe 100644
@@ -398,7 +398,8 @@ static bool reg_type_may_be_null(enum bpf_reg_type type)
 	return type == PTR_TO_MAP_VALUE_OR_NULL ||
 	       type == PTR_TO_SOCKET_OR_NULL ||
 	       type == PTR_TO_SOCK_COMMON_OR_NULL ||
-	       type == PTR_TO_TCP_SOCK_OR_NULL;
+	       type == PTR_TO_TCP_SOCK_OR_NULL ||
+	       type == PTR_TO_BTF_ID_OR_NULL;
 }
 
 static bool reg_may_point_to_spin_lock(const struct bpf_reg_state *reg)
@@ -483,6 +484,7 @@ static const char * const reg_type_str[] = {
 	[PTR_TO_TP_BUFFER]	= "tp_buffer",
 	[PTR_TO_XDP_SOCK]	= "xdp_sock",
 	[PTR_TO_BTF_ID]		= "ptr_",
+	[PTR_TO_BTF_ID_OR_NULL]	= "ptr_or_null_",
 };
 
 static char slot_type_char[] = {
@@ -543,7 +545,7 @@ static void print_verifier_state(struct bpf_verifier_env *env,
 			/* reg->off should be 0 for SCALAR_VALUE */
 			verbose(env, "%lld", reg->var_off.value + reg->off);
 		} else {
-			if (t == PTR_TO_BTF_ID)
+			if (t == PTR_TO_BTF_ID || t == PTR_TO_BTF_ID_OR_NULL)
 				verbose(env, "%s", kernel_type_name(reg->btf_id));
 			verbose(env, "(id=%d", reg->id);
 			if (reg_type_may_be_refcounted_or_null(t))
@@ -2139,6 +2141,7 @@ static bool is_spillable_regtype(enum bpf_reg_type type)
 	case PTR_TO_TCP_SOCK_OR_NULL:
 	case PTR_TO_XDP_SOCK:
 	case PTR_TO_BTF_ID:
+	case PTR_TO_BTF_ID_OR_NULL:
 		return true;
 	default:
 		return false;
@@ -2659,7 +2662,7 @@ static int check_ctx_access(struct bpf_verifier_env *env, int insn_idx, int off,
 		 */
 		*reg_type = info.reg_type;
 
-		if (*reg_type == PTR_TO_BTF_ID)
+		if (*reg_type == PTR_TO_BTF_ID || *reg_type == PTR_TO_BTF_ID_OR_NULL)
 			*btf_id = info.btf_id;
 		else
 			env->insn_aux_data[insn_idx].ctx_field_size = info.ctx_field_size;
@@ -3243,7 +3246,8 @@ static int check_mem_access(struct bpf_verifier_env *env, int insn_idx, u32 regn
 				 * a sub-register.
 				 */
 				regs[value_regno].subreg_def = DEF_NOT_SUBREG;
-				if (reg_type == PTR_TO_BTF_ID)
+				if (reg_type == PTR_TO_BTF_ID ||
+				    reg_type == PTR_TO_BTF_ID_OR_NULL)
 					regs[value_regno].btf_id = btf_id;
 			}
 			regs[value_regno].type = reg_type;
@@ -6572,6 +6576,8 @@ static void mark_ptr_or_null_reg(struct bpf_func_state *state,
 			reg->type = PTR_TO_SOCK_COMMON;
 		} else if (reg->type == PTR_TO_TCP_SOCK_OR_NULL) {
 			reg->type = PTR_TO_TCP_SOCK;
+		} else if (reg->type == PTR_TO_BTF_ID_OR_NULL) {
+			reg->type = PTR_TO_BTF_ID;
 		}
 		if (is_null) {
 			/* We don't need id and ref_obj_id from this point
@@ -8429,6 +8435,7 @@ static bool reg_type_mismatch_ok(enum bpf_reg_type type)
 	case PTR_TO_TCP_SOCK_OR_NULL:
 	case PTR_TO_XDP_SOCK:
 	case PTR_TO_BTF_ID:
+	case PTR_TO_BTF_ID_OR_NULL:
 		return false;
 	default:
 		return true;
@@ -10640,6 +10647,7 @@ static int check_attach_btf_id(struct bpf_verifier_env *env)
 		prog->aux->attach_func_proto = t;
 		if (!bpf_iter_prog_supported(prog))
 			return -EINVAL;
+		prog->aux->btf_id_or_null_non0_off = true;
 		ret = btf_distill_func_proto(&env->log, btf, t,
 					     tname, &fmodel);
 		return ret;
Signature · 1 line
2.24.1
#20 lore raw

Re: [PATCH bpf-next v4 12/21] bpf: add PTR_TO_BTF_ID_OR_NULL support

kernel/bpf/btf.c bool btf_ctx_access(int off, int size, enum bpf_access_type type, · line 3796

Quoted 13 lines · Yonghong Song
kernel/bpf/btf.c +4 -1
index a2cfba89a8e1..c490fbde22d4 100644
@@ -3790,7 +3790,10 @@ bool btf_ctx_access(int off, int size, enum bpf_access_type type,
 		return true;
 
 	/* this is a pointer to another type */
-	info->reg_type = PTR_TO_BTF_ID;
+	if (off != 0 && prog->aux->btf_id_or_null_non0_off)
+		info->reg_type = PTR_TO_BTF_ID_OR_NULL;
+	else
+		info->reg_type = PTR_TO_BTF_ID;

I think the verifier should be smarter than this.
It's too specific and inflexible. All ctx fields of bpf_iter execpt first will be such ? let's figure out a different way to tell verifier about this. How about using typedef with specific suffix? Like:
typedef struct bpf_map *bpf_map_or_null;
struct bpf_iter__bpf_map {
struct bpf_iter_meta *meta;
bpf_map_or_null map;
};
or use a union with specific second member? Like:
struct bpf_iter__bpf_map {
struct bpf_iter_meta *meta;

   union {
     struct bpf_map *map;
     long null;

};
};

#21 lore raw

Re: [PATCH bpf-next v4 12/21] bpf: add PTR_TO_BTF_ID_OR_NULL support

Quoted 31 lines · Alexei Starovoitov

kernel/bpf/btf.c bool btf_ctx_access(int off, int size, enum bpf_access_type type, · line 3796

Quoted 13 lines · Yonghong Song
kernel/bpf/btf.c +4 -1
index a2cfba89a8e1..c490fbde22d4 100644
@@ -3790,7 +3790,10 @@ bool btf_ctx_access(int off, int size, enum bpf_access_type type,
  		return true;
 
  	/* this is a pointer to another type */
-	info->reg_type = PTR_TO_BTF_ID;
+	if (off != 0 && prog->aux->btf_id_or_null_non0_off)
+		info->reg_type = PTR_TO_BTF_ID_OR_NULL;
+	else
+		info->reg_type = PTR_TO_BTF_ID;

I think the verifier should be smarter than this.
It's too specific and inflexible. All ctx fields of bpf_iter execpt first will be such ? let's figure out a different way to tell verifier about this. How about using typedef with specific suffix? Like:
typedef struct bpf_map *bpf_map_or_null;

  struct bpf_iter__bpf_map {
    struct bpf_iter_meta *meta;
    bpf_map_or_null map;
  };
or use a union with specific second member? Like:
  struct bpf_iter__bpf_map {
    struct bpf_iter_meta *meta;
    union {
      struct bpf_map *map;
      long null;
    };
Quoted 1 line

};

I have an alternative approach to refactor this for future support for map elements as well.

For example, for bpf_map_elements iterator the prog context type can be

    struct bpf_iter_bpf_map_elem {
 	struct bpf_iter_meta *meta;
	strruct bpf_map *map;
	<key type>  *key;
	<value type> *val;

};

target will pass the following information to bpf_iter registration:
arg 1: PTR_TO_BTF_ID
arg 2: PTR_TO_BTF_ID_OR_NULL
arg 3: PTR_TO_BUFFER
arg 4: PTR_TO_BUFFER

verifier will retrieve the reg_type from target.

#22 lore raw

Re: [PATCH bpf-next v4 12/21] bpf: add PTR_TO_BTF_ID_OR_NULL support

Quoted 49 lines · Yonghong Song
Quoted 31 lines · Alexei Starovoitov

kernel/bpf/btf.c bool btf_ctx_access(int off, int size, enum bpf_access_type type, · line 3796

Quoted 13 lines · Yonghong Song
kernel/bpf/btf.c +4 -1
index a2cfba89a8e1..c490fbde22d4 100644
@@ -3790,7 +3790,10 @@ bool btf_ctx_access(int off, int size, enum bpf_access_type type,
             return true;
 
     /* this is a pointer to another type */
-    info->reg_type = PTR_TO_BTF_ID;
+    if (off != 0 && prog->aux->btf_id_or_null_non0_off)
+            info->reg_type = PTR_TO_BTF_ID_OR_NULL;
+    else
+            info->reg_type = PTR_TO_BTF_ID;

I think the verifier should be smarter than this.
It's too specific and inflexible. All ctx fields of bpf_iter execpt first will be such ? let's figure out a different way to tell verifier about this. How about using typedef with specific suffix? Like:
typedef struct bpf_map *bpf_map_or_null;

  struct bpf_iter__bpf_map {
    struct bpf_iter_meta *meta;
    bpf_map_or_null map;
  };
or use a union with specific second member? Like:
  struct bpf_iter__bpf_map {
    struct bpf_iter_meta *meta;
    union {
      struct bpf_map *map;
      long null;
    };
Quoted 1 line

};

I have an alternative approach to refactor this for future support for map elements as well.

For example, for bpf_map_elements iterator the prog context type can be

     struct bpf_iter_bpf_map_elem {
        struct bpf_iter_meta *meta;
        strruct bpf_map *map;
        <key type>  *key;
        <value type> *val;
    };
Quoted 7 lines

target will pass the following information to bpf_iter registration:
arg 1: PTR_TO_BTF_ID
arg 2: PTR_TO_BTF_ID_OR_NULL
arg 3: PTR_TO_BUFFER
arg 4: PTR_TO_BUFFER

verifier will retrieve the reg_type from target.

you mean to introduce something like 'struct bpf_func_proto' that describes types of helpers, but instead something similar to clarify the types in ctx ? That should work. Thanks

#23 lore raw

Re: [PATCH bpf-next v4 12/21] bpf: add PTR_TO_BTF_ID_OR_NULL support

Quoted 56 lines · Alexei Starovoitov
Quoted 55 lines · Yonghong Song
Quoted 19 lines · Alexei Starovoitov

kernel/bpf/btf.c bool btf_ctx_access(int off, int size, enum bpf_access_type type, · line 3796

Quoted 13 lines · Yonghong Song
kernel/bpf/btf.c +4 -1
index a2cfba89a8e1..c490fbde22d4 100644
@@ -3790,7 +3790,10 @@ bool btf_ctx_access(int off, int size, enum bpf_access_type type,
              return true;
 
      /* this is a pointer to another type */
-    info->reg_type = PTR_TO_BTF_ID;
+    if (off != 0 && prog->aux->btf_id_or_null_non0_off)
+            info->reg_type = PTR_TO_BTF_ID_OR_NULL;
+    else
+            info->reg_type = PTR_TO_BTF_ID;

I think the verifier should be smarter than this.
It's too specific and inflexible. All ctx fields of bpf_iter execpt first will be such ? let's figure out a different way to tell verifier about this. How about using typedef with specific suffix? Like:

Quoted 13 lines
typedef struct bpf_map *bpf_map_or_null;
   struct bpf_iter__bpf_map {
     struct bpf_iter_meta *meta;
     bpf_map_or_null map;
   };
or use a union with specific second member? Like:
   struct bpf_iter__bpf_map {
     struct bpf_iter_meta *meta;
     union {
       struct bpf_map *map;
       long null;
     };
   };

I have an alternative approach to refactor this for future support for map elements as well.

For example, for bpf_map_elements iterator the prog context type can be

      struct bpf_iter_bpf_map_elem {
         struct bpf_iter_meta *meta;
         strruct bpf_map *map;
         <key type>  *key;
         <value type> *val;
     };

target will pass the following information to bpf_iter registration:
arg 1: PTR_TO_BTF_ID
arg 2: PTR_TO_BTF_ID_OR_NULL
arg 3: PTR_TO_BUFFER
arg 4: PTR_TO_BUFFER

Quoted 5 lines
Quoted 1 line

verifier will retrieve the reg_type from target.

you mean to introduce something like 'struct bpf_func_proto' that describes types of helpers, but instead something similar to clarify the types in ctx ? That should work. Thanks

Yes, this is what I think will be extensible.

#24 lore raw

PATCH v4 13/21 bpf: add bpf_seq_printf and bpf_seq_write helpers

In reply to Yonghong Song

Two helpers bpf_seq_printf and bpf_seq_write, are added for writing data to the seq_file buffer.

bpf_seq_printf supports common format string flag/width/type fields so at least I can get identical results for netlink and ipv6_route targets.

For bpf_seq_printf and bpf_seq_write, return value -EOVERFLOW specifically indicates a write failure due to overflow, which means the object will be repeated in the next bpf invocation if object collection stays the same. Note that if the object collection is changed, depending how collection traversal is done, even if the object still in the collection, it may not be visited.

For bpf_seq_printf, format %s, %p{i,I}{4,6} needs to read kernel memory. Reading kernel memory may fail in the following two cases:
- invalid kernel address, or
- valid kernel address but requiring a major fault
If reading kernel memory failed, the %s string will be an empty string and %p{i,I}{4,6} will be all 0. Not returning error to bpf program is consistent with what bpf_trace_printk() does for now.

bpf_seq_printf may return -EBUSY meaning that internal percpu buffer for memory copy of strings or other pointees is not available. Bpf program can return 1 to indicate it wants the same object to be repeated. Right now, this should not happen on no-RT kernels since migrate_disable(), which guards bpf prog call, calls preempt_disable().

---
 include/uapi/linux/bpf.h       |  39 +++++-
 kernel/trace/bpf_trace.c       | 214 +++++++++++++++++++++++++++++++++
 scripts/bpf_helpers_doc.py     |   2 +
 tools/include/uapi/linux/bpf.h |  39 +++++-
 4 files changed, 292 insertions(+), 2 deletions(-)
include/uapi/linux/bpf.h +38 -1
index 708763f702e1..9d1932e23cec 100644
@@ -3077,6 +3077,41 @@ union bpf_attr {
  * 		See: clock_gettime(CLOCK_BOOTTIME)
  * 	Return
  * 		Current *ktime*.
+ *
+ * int bpf_seq_printf(struct seq_file *m, const char *fmt, u32 fmt_size, const void *data, u32 data_len)
+ * 	Description
+ * 		seq_printf uses seq_file seq_printf() to print out the format string.
+ * 		The *m* represents the seq_file. The *fmt* and *fmt_size* are for
+ * 		the format string itself. The *data* and *data_len* are format string
+ * 		arguments. The *data* are a u64 array and corresponding format string
+ * 		values are stored in the array. For strings and pointers where pointees
+ * 		are accessed, only the pointer values are stored in the *data* array.
+ * 		The *data_len* is the *data* size in term of bytes.
+ *
+ *		Formats **%s**, **%p{i,I}{4,6}** requires to read kernel memory.
+ *		Reading kernel memory may fail due to either invalid address or
+ *		valid address but requiring a major memory fault. If reading kernel memory
+ *		fails, the string for **%s** will be an empty string, and the ip
+ *		address for **%p{i,I}{4,6}** will be 0. Not returning error to
+ *		bpf program is consistent with what bpf_trace_printk() does for now.
+ * 	Return
+ * 		0 on success, or a negative errno in case of failure.
+ *
+ *		* **-EBUSY**		Percpu memory copy buffer is busy, can try again
+ *					by returning 1 from bpf program.
+ *		* **-EINVAL**		Invalid arguments, or invalid/unsupported formats.
+ *		* **-E2BIG**		Too many format specifiers.
+ *		* **-EOVERFLOW**	Overflow happens, the same object will be tried again.
+ *
+ * int bpf_seq_write(struct seq_file *m, const void *data, u32 len)
+ * 	Description
+ * 		seq_write uses seq_file seq_write() to write the data.
+ * 		The *m* represents the seq_file. The *data* and *len* represent the
+ *		data to write in bytes.
+ * 	Return
+ * 		0 on success, or a negative errno in case of failure.
+ *
+ *		* **-EOVERFLOW**	Overflow happens, the same object will be tried again.
  */
 #define __BPF_FUNC_MAPPER(FN)		\
 	FN(unspec),			\
@@ -3204,7 +3239,9 @@ union bpf_attr {
 	FN(get_netns_cookie),		\
 	FN(get_current_ancestor_cgroup_id),	\
 	FN(sk_assign),			\
-	FN(ktime_get_boot_ns),
+	FN(ktime_get_boot_ns),		\
+	FN(seq_printf),			\
+	FN(seq_write),
 
 /* integer value in 'imm' field of BPF_CALL instruction selects which helper
  * function eBPF program intends to call
kernel/trace/bpf_trace.c +214 -0

226 lines not on this page · raw message

scripts/bpf_helpers_doc.py +2 -0
index f43d193aff3a..ded304c96a05 100755
@@ -414,6 +414,7 @@ class PrinterHelpers(Printer):
             'struct sk_reuseport_md',
             'struct sockaddr',
             'struct tcphdr',
+            'struct seq_file',
 
             'struct __sk_buff',
             'struct sk_msg_md',
@@ -450,6 +451,7 @@ class PrinterHelpers(Printer):
             'struct sk_reuseport_md',
             'struct sockaddr',
             'struct tcphdr',
+            'struct seq_file',
     }
     mapped_types = {
             'u8': '__u8',
tools/include/uapi/linux/bpf.h +38 -1
index 708763f702e1..9d1932e23cec 100644
@@ -3077,6 +3077,41 @@ union bpf_attr {
  * 		See: clock_gettime(CLOCK_BOOTTIME)
  * 	Return
  * 		Current *ktime*.
+ *
+ * int bpf_seq_printf(struct seq_file *m, const char *fmt, u32 fmt_size, const void *data, u32 data_len)
+ * 	Description
+ * 		seq_printf uses seq_file seq_printf() to print out the format string.
+ * 		The *m* represents the seq_file. The *fmt* and *fmt_size* are for
+ * 		the format string itself. The *data* and *data_len* are format string
+ * 		arguments. The *data* are a u64 array and corresponding format string
+ * 		values are stored in the array. For strings and pointers where pointees
+ * 		are accessed, only the pointer values are stored in the *data* array.
+ * 		The *data_len* is the *data* size in term of bytes.
+ *
+ *		Formats **%s**, **%p{i,I}{4,6}** requires to read kernel memory.
+ *		Reading kernel memory may fail due to either invalid address or
+ *		valid address but requiring a major memory fault. If reading kernel memory
+ *		fails, the string for **%s** will be an empty string, and the ip
+ *		address for **%p{i,I}{4,6}** will be 0. Not returning error to
+ *		bpf program is consistent with what bpf_trace_printk() does for now.
+ * 	Return
+ * 		0 on success, or a negative errno in case of failure.
+ *
+ *		* **-EBUSY**		Percpu memory copy buffer is busy, can try again
+ *					by returning 1 from bpf program.
+ *		* **-EINVAL**		Invalid arguments, or invalid/unsupported formats.
+ *		* **-E2BIG**		Too many format specifiers.
+ *		* **-EOVERFLOW**	Overflow happens, the same object will be tried again.
+ *
+ * int bpf_seq_write(struct seq_file *m, const void *data, u32 len)
+ * 	Description
+ * 		seq_write uses seq_file seq_write() to write the data.
+ * 		The *m* represents the seq_file. The *data* and *len* represent the
+ *		data to write in bytes.
+ * 	Return
+ * 		0 on success, or a negative errno in case of failure.
+ *
+ *		* **-EOVERFLOW**	Overflow happens, the same object will be tried again.
  */
 #define __BPF_FUNC_MAPPER(FN)		\
 	FN(unspec),			\
@@ -3204,7 +3239,9 @@ union bpf_attr {
 	FN(get_netns_cookie),		\
 	FN(get_current_ancestor_cgroup_id),	\
 	FN(sk_assign),			\
-	FN(ktime_get_boot_ns),
+	FN(ktime_get_boot_ns),		\
+	FN(seq_printf),			\
+	FN(seq_write),
 
 /* integer value in 'imm' field of BPF_CALL instruction selects which helper
  * function eBPF program intends to call
Signature · 1 line
2.24.1
#25 lore raw

PATCH v4 14/21 bpf: handle spilled PTR_TO_BTF_ID properly when checking stack_boundary

In reply to Yonghong Song

This specifically to handle the case like below:
// ptr below is a socket ptr identified by PTR_TO_BTF_ID u64 param[2] = { ptr, val };
bpf_seq_printf(seq, fmt, sizeof(fmt), param, sizeof(param));

In this case, the 16 bytes stack for "param" contains:
8 bytes for ptr with spilled PTR_TO_BTF_ID 8 bytes for val as STACK_MISC

The current verifier will complain the ptr should not be visible to the helper.
...
16: (7b) *(u64 *)(r10 -64) = r2
18: (7b) *(u64 *)(r10 -56) = r1
19: (bf) r4 = r10
;
20: (07) r4 += -64
; BPF_SEQ_PRINTF(seq, fmt1, (long)s, s->sk_protocol); 21: (bf) r1 = r6
22: (18) r2 = 0xffffa8d00018605a
24: (b4) w3 = 10
25: (b4) w5 = 16
26: (85) call bpf_seq_printf#125

    R0=inv(id=0) R1_w=ptr_seq_file(id=0,off=0,imm=0)
    R2_w=map_value(id=0,off=90,ks=4,vs=144,imm=0) R3_w=inv10
    R4_w=fp-64 R5_w=inv16 R6=ptr_seq_file(id=0,off=0,imm=0)
    R7=ptr_netlink_sock(id=0,off=0,imm=0) R10=fp0 fp-56_w=mmmmmmmm
    fp-64_w=ptr_

last_idx 26 first_idx 13
regs=8 stack=0 before 25: (b4) w5 = 16 regs=8 stack=0 before 24: (b4) w3 = 10 invalid indirect read from stack off -64+0 size 16

---
 kernel/bpf/verifier.c | 5 +++++
 1 file changed, 5 insertions(+)
kernel/bpf/verifier.c +5 -0
index 36b2a38a06fe..2a1826c76bb6 100644
@@ -3494,6 +3494,11 @@ static int check_stack_boundary(struct bpf_verifier_env *env, int regno,
 			*stype = STACK_MISC;
 			goto mark;
 		}
+
+		if (state->stack[spi].slot_type[0] == STACK_SPILL &&
+		    state->stack[spi].spilled_ptr.type == PTR_TO_BTF_ID)
+			goto mark;
+
 		if (state->stack[spi].slot_type[0] == STACK_SPILL &&
 		    state->stack[spi].spilled_ptr.type == SCALAR_VALUE) {
 			__mark_reg_unknown(env, &state->stack[spi].spilled_ptr);
Signature · 1 line
2.24.1
#26 lore raw

PATCH v4 15/21 bpf: support variable length array in tracing programs

In reply to Yonghong Song

In /proc/net/ipv6_route, we have

  struct fib6_info {
    struct fib6_table *fib6_table;
    ...
    struct fib6_nh fib6_nh[0];
  }
  struct fib6_nh {
    struct fib_nh_common nh_common;
    struct rt6_info **rt6i_pcpu;
    struct rt6_exception_bucket *rt6i_exception_bucket;
  };
  struct fib_nh_common {
    ...
    u8 nhc_gw_family;
    ...

}

The access:
struct fib6_nh *fib6_nh = &rt->fib6_nh;
... fib6_nh->nh_common.nhc_gw_family ...

This patch ensures such an access is handled properly.

---
 kernel/bpf/btf.c | 37 +++++++++++++++++++++++++++++++++++++
 1 file changed, 37 insertions(+)
kernel/bpf/btf.c +37 -0
index c490fbde22d4..dcd233139294 100644
@@ -3833,6 +3833,7 @@ int btf_struct_access(struct bpf_verifier_log *log,
 	const struct btf_type *mtype, *elem_type = NULL;
 	const struct btf_member *member;
 	const char *tname, *mname;
+	u32 vlen;
 
 again:
 	tname = __btf_name_by_offset(btf_vmlinux, t->name_off);
@@ -3841,7 +3842,43 @@ int btf_struct_access(struct bpf_verifier_log *log,
 		return -EINVAL;
 	}
 
+	vlen = btf_type_vlen(t);
 	if (off + size > t->size) {
+		/* If the last element is a variable size array, we may
+		 * need to relax the rule.
+		 */
+		struct btf_array *array_elem;
+
+		if (vlen == 0)
+			goto error;
+
+		member = btf_type_member(t) + vlen - 1;
+		mtype = btf_type_skip_modifiers(btf_vmlinux, member->type,
+						NULL);
+		if (!btf_type_is_array(mtype))
+			goto error;
+
+		array_elem = (struct btf_array *)(mtype + 1);
+		if (array_elem->nelems != 0)
+			goto error;
+
+		moff = btf_member_bit_offset(t, member) / 8;
+		if (off < moff)
+			goto error;
+
+		/* Only allow structure for now, can be relaxed for
+		 * other types later.
+		 */
+		elem_type = btf_type_skip_modifiers(btf_vmlinux,
+						    array_elem->type, NULL);
+		if (!btf_type_is_struct(elem_type))
+			goto error;
+
+		off = (off - moff) % elem_type->size;
+		return btf_struct_access(log, elem_type, off, size, atype,
+					 next_btf_id);
+
+error:
 		bpf_log(log, "access beyond struct %s at off %u size %u\n",
 			tname, off, size);
 		return -EACCES;
Signature · 1 line
2.24.1
#27 lore raw

PATCH v4 16/21 tools/libbpf: add bpf_iter support

In reply to Yonghong Song

Two new libbpf APIs are added to support bpf_iter:
- bpf_program__attach_iter
Given a bpf program and additional parameters, which is none now, returns a bpf_link.
- bpf_iter_create
syscall level API to create a bpf iterator.

The macro BPF_SEQ_PRINTF are also introduced. The format looks like:
BPF_SEQ_PRINTF(seq, "task id %d\n", pid);

This macro can help bpf program writers with nicer bpf_seq_printf syntax similar to the kernel one.

---
 tools/lib/bpf/bpf.c         | 10 +++++++
 tools/lib/bpf/bpf.h         |  2 ++
 tools/lib/bpf/bpf_tracing.h | 16 ++++++++++++
 tools/lib/bpf/libbpf.c      | 52 +++++++++++++++++++++++++++++++++++++
 tools/lib/bpf/libbpf.h      |  9 +++++++
 tools/lib/bpf/libbpf.map    |  2 ++
 6 files changed, 91 insertions(+)
tools/lib/bpf/bpf.c +10 -0
index 43322f0d6c7f..a7329b671c41 100644
@@ -619,6 +619,16 @@ int bpf_link_update(int link_fd, int new_prog_fd,
 	return sys_bpf(BPF_LINK_UPDATE, &attr, sizeof(attr));
 }
 
+int bpf_iter_create(int link_fd)
+{
+	union bpf_attr attr;
+
+	memset(&attr, 0, sizeof(attr));
+	attr.iter_create.link_fd = link_fd;
+
+	return sys_bpf(BPF_ITER_CREATE, &attr, sizeof(attr));
+}
+
 int bpf_prog_query(int target_fd, enum bpf_attach_type type, __u32 query_flags,
 		   __u32 *attach_flags, __u32 *prog_ids, __u32 *prog_cnt)
 {
tools/lib/bpf/bpf.h +2 -0
index 1901b2777854..1b6015b21ba8 100644
@@ -187,6 +187,8 @@ struct bpf_link_update_opts {
 LIBBPF_API int bpf_link_update(int link_fd, int new_prog_fd,
 			       const struct bpf_link_update_opts *opts);
 
+LIBBPF_API int bpf_iter_create(int link_fd);
+
 struct bpf_prog_test_run_attr {
 	int prog_fd;
 	int repeat;
tools/lib/bpf/bpf_tracing.h +16 -0
index f3f3c3fb98cb..cf97d07692b4 100644
@@ -413,4 +413,20 @@ typeof(name(0)) name(struct pt_regs *ctx)				    \
 }									    \
 static __always_inline typeof(name(0)) ____##name(struct pt_regs *ctx, ##args)
 
+/*
+ * BPF_SEQ_PRINTF to wrap bpf_seq_printf to-be-printed values
+ * in a structure.
+ */
+#define BPF_SEQ_PRINTF(seq, fmt, args...)				    \
+	({								    \
+		_Pragma("GCC diagnostic push")				    \
+		_Pragma("GCC diagnostic ignored \"-Wint-conversion\"")	    \
+		static const char ___fmt[] = fmt;			    \
+		unsigned long long ___param[] = { args };		    \
+		_Pragma("GCC diagnostic pop")				    \
+		int ___ret = bpf_seq_printf(seq, ___fmt, sizeof(___fmt),    \
+					    ___param, sizeof(___param));    \
+		___ret;							    \
+	})
+
 #endif
tools/lib/bpf/libbpf.c +52 -0

88 lines not on this page · raw message

tools/lib/bpf/libbpf.h +9 -0
index f1dacecb1619..8ea69558f0a8 100644
@@ -258,6 +258,15 @@ struct bpf_map;
 
 LIBBPF_API struct bpf_link *bpf_map__attach_struct_ops(struct bpf_map *map);
 
+struct bpf_iter_attach_opts {
+	size_t sz; /* size of this struct for forward/backward compatibility */
+};
+#define bpf_iter_attach_opts__last_field sz
+
+LIBBPF_API struct bpf_link *
+bpf_program__attach_iter(struct bpf_program *prog,
+			 const struct bpf_iter_attach_opts *opts);
+
 struct bpf_insn;
 
 /*
tools/lib/bpf/libbpf.map +2 -0
index e03bd4db827e..0133d469d30b 100644
@@ -258,6 +258,8 @@ LIBBPF_0.0.8 {
 LIBBPF_0.0.9 {
 	global:
 		bpf_enable_stats;
+		bpf_iter_create;
 		bpf_link_get_fd_by_id;
 		bpf_link_get_next_id;
+		bpf_program__attach_iter;
 } LIBBPF_0.0.8;
Signature · 1 line
2.24.1
#28 lore raw

Re: [PATCH bpf-next v4 16/21] tools/libbpf: add bpf_iter support

tools/lib/bpf/libbpf.c static int bpf_object__collect_st_ops_relos(struct bpf_object *obj, · line 6901

Quoted 6 lines · Yonghong Song
tools/lib/bpf/libbpf.c +1 -0
@@ -6891,6 +6897,7 @@ static int bpf_object__collect_st_ops_relos(struct bpf_object *obj,
 
 #define BTF_TRACE_PREFIX "btf_trace_"
 #define BTF_LSM_PREFIX "bpf_lsm_"
+#define BTF_ITER_PREFIX "__bpf_iter__"
 #define BTF_MAX_NAME_SIZE 128

In the kernel source the prefix doesn't stand out, but on libbpf side it looks inconsistent. May be drop __ prefix and keep one _ in the suffix?

#29 lore raw

Re: [PATCH bpf-next v4 16/21] tools/libbpf: add bpf_iter support

Quoted 10 lines · Alexei Starovoitov

tools/lib/bpf/libbpf.c static int bpf_object__collect_st_ops_relos(struct bpf_object *obj, · line 6901

Quoted 6 lines · Yonghong Song
tools/lib/bpf/libbpf.c +1 -0
@@ -6891,6 +6897,7 @@ static int bpf_object__collect_st_ops_relos(struct bpf_object *obj,
 
  #define BTF_TRACE_PREFIX "btf_trace_"
  #define BTF_LSM_PREFIX "bpf_lsm_"
+#define BTF_ITER_PREFIX "__bpf_iter__"
  #define BTF_MAX_NAME_SIZE 128

In the kernel source the prefix doesn't stand out, but on libbpf side it looks inconsistent. May be drop __ prefix and keep one _ in the suffix?

Currently, I have context type as
struct bpf_iter__bpf_map
Based on the above proposal, we will have function name as
bpf_iter_bpf_map
It is quite similar to each other. My current usage to have

    __bpf_iter__bpf_map

intends to make func name and struct type name quite different. Or maybe
bpf_iter__bpf_map vs. bpf_iter_bpf_map
just fine as user should not care about func name bpf_iter_bpf_map at all?

#30 lore raw

Re: [PATCH bpf-next v4 16/21] tools/libbpf: add bpf_iter support

Quoted 20 lines · Yonghong Song
Quoted 10 lines · Alexei Starovoitov

tools/lib/bpf/libbpf.c static int bpf_object__collect_st_ops_relos(struct bpf_object *obj, · line 6901

Quoted 6 lines · Yonghong Song
tools/lib/bpf/libbpf.c +1 -0
@@ -6891,6 +6897,7 @@ static int bpf_object__collect_st_ops_relos(struct bpf_object *obj,
 
  #define BTF_TRACE_PREFIX "btf_trace_"
  #define BTF_LSM_PREFIX "bpf_lsm_"
+#define BTF_ITER_PREFIX "__bpf_iter__"
  #define BTF_MAX_NAME_SIZE 128

In the kernel source the prefix doesn't stand out, but on libbpf side it looks inconsistent. May be drop __ prefix and keep one _ in the suffix?

Currently, I have context type as
struct bpf_iter__bpf_map
Based on the above proposal, we will have function name as
bpf_iter_bpf_map
It is quite similar to each other. My current usage to have

Quoted 6 lines
     __bpf_iter__bpf_map

intends to make func name and struct type name quite different. Or maybe
bpf_iter__bpf_map vs. bpf_iter_bpf_map
just fine as user should not care about func name bpf_iter_bpf_map at all?

Type names bpf_iter_bpf_map and bpf_iter_foo don't look unique, but I don't see why they should.
If code really required type name uniqueness __bpf_iter__ prefix wouldn't provide that property anyway.
I think bpf_iter_ falls into the same category of prefixes like those used by lsm, trace, struct_ops. Or I could be missing why iter has to be different.

#31 lore raw

Re: [PATCH bpf-next v4 16/21] tools/libbpf: add bpf_iter support

Quoted 28 lines · Alexei Starovoitov
Quoted 20 lines · Yonghong Song
Quoted 10 lines · Alexei Starovoitov

tools/lib/bpf/libbpf.c static int bpf_object__collect_st_ops_relos(struct bpf_object *obj, · line 6901

Quoted 6 lines · Yonghong Song
tools/lib/bpf/libbpf.c +1 -0
@@ -6891,6 +6897,7 @@ static int bpf_object__collect_st_ops_relos(struct bpf_object *obj,
 
   #define BTF_TRACE_PREFIX "btf_trace_"
   #define BTF_LSM_PREFIX "bpf_lsm_"
+#define BTF_ITER_PREFIX "__bpf_iter__"
   #define BTF_MAX_NAME_SIZE 128

In the kernel source the prefix doesn't stand out, but on libbpf side it looks inconsistent. May be drop __ prefix and keep one _ in the suffix?

Currently, I have context type as
struct bpf_iter__bpf_map
Based on the above proposal, we will have function name as
bpf_iter_bpf_map
It is quite similar to each other. My current usage to have

Quoted 6 lines
      __bpf_iter__bpf_map

intends to make func name and struct type name quite different. Or maybe
bpf_iter__bpf_map vs. bpf_iter_bpf_map
just fine as user should not care about func name bpf_iter_bpf_map at all?

Quoted 7 lines

Type names bpf_iter_bpf_map and bpf_iter_foo don't look unique, but I don't see why they should.
If code really required type name uniqueness __bpf_iter__ prefix wouldn't provide that property anyway.
I think bpf_iter_ falls into the same category of prefixes like those used by lsm, trace, struct_ops. Or I could be missing why iter has to be different.

I will change to bpf_iter_ prefix then. This is hidden from user anyway.

#32 lore raw

PATCH v4 17/21 tools/libpf: add offsetof/container_of macro in bpf_helpers.h

In reply to Yonghong Song

These two helpers will be used later in bpf_iter bpf program bpf_iter_netlink.c. Put them in bpf_helpers.h since they could be useful in other cases.

---
 tools/lib/bpf/bpf_helpers.h | 14 ++++++++++++++
 1 file changed, 14 insertions(+)
tools/lib/bpf/bpf_helpers.h +14 -0
index da00b87aa199..f67dce2af802 100644
@@ -36,6 +36,20 @@
 #define __weak __attribute__((weak))
 #endif
 
+/*
+ * Helper macro to manipulate data structures
+ */
+#ifndef offsetof
+#define offsetof(TYPE, MEMBER)  ((size_t)&((TYPE *)0)->MEMBER)
+#endif
+#ifndef container_of
+#define container_of(ptr, type, member)				\
+	({							\
+		void *__mptr = (void *)(ptr);			\
+		((type *)(__mptr - offsetof(type, member)));	\
+	})
+#endif
+
 /*
  * Helper structure used by eBPF C program
  * to describe BPF map attributes to libbpf loader
Signature · 1 line
2.24.1
#33 lore raw

PATCH v4 18/21 tools/bpftool: add bpf_iter support for bptool

In reply to Yonghong Song

Currently, only one command is supported
bpftool iter pin <bpf_prog.o> <path>

It will pin the trace/iter bpf program in
the object file <bpf_prog.o> to the <path>
where <path> should be on a bpffs mount.

For example,

  $ bpftool iter pin ./bpf_iter_ipv6_route.o \
    /sys/fs/bpf/my_route
User can then do a `cat` to print out the results:
  $ cat /sys/fs/bpf/my_route
    fe800000000000000000000000000000 40 00000000000000000000000000000000 ...
    00000000000000000000000000000000 00 00000000000000000000000000000000 ...
    00000000000000000000000000000001 80 00000000000000000000000000000000 ...
    fe800000000000008c0162fffebdfd57 80 00000000000000000000000000000000 ...
    ff000000000000000000000000000000 08 00000000000000000000000000000000 ...
    00000000000000000000000000000000 00 00000000000000000000000000000000 ...

The implementation for ipv6_route iterator is in one of subsequent patches.

This patch also added BPF_LINK_TYPE_ITER to link query.

In the future, we may add additional parameters to pin command by parameterizing the bpf iterator. For example, a map_id or pid may be added to let bpf program only traverses a single map or task, similar to kernel seq_file single_open().

We may also add introspection command for targets/iterators by leveraging the bpf_iter itself.

---
 .../bpftool/Documentation/bpftool-iter.rst    | 83 +++++++++++++++++
 tools/bpf/bpftool/bash-completion/bpftool     | 13 +++
 tools/bpf/bpftool/iter.c                      | 88 +++++++++++++++++++
 tools/bpf/bpftool/link.c                      |  1 +
 tools/bpf/bpftool/main.c                      |  3 +-
 tools/bpf/bpftool/main.h                      |  1 +
 6 files changed, 188 insertions(+), 1 deletion(-)
 create mode 100644 tools/bpf/bpftool/Documentation/bpftool-iter.rst
 create mode 100644 tools/bpf/bpftool/iter.c
tools/bpf/bpftool/Documentation/bpftool-iter.rst new file +83 -0
new file mode 100644
index 000000000000..13b173d93890
@@ -0,0 +1,83 @@
+============
+bpftool-iter
+============
+-------------------------------------------------------------------------------
+tool to create BPF iterators
+-------------------------------------------------------------------------------
+
+:Manual section: 8
+
+SYNOPSIS
+========
+
+	**bpftool** [*OPTIONS*] **iter** *COMMAND*
+
+	*COMMANDS* := { **pin** | **help** }
+
+ITER COMMANDS
+===================
+
+|	**bpftool** **iter pin** *OBJ* *PATH*
+|	**bpftool** **iter help**
+|
+|	*OBJ* := /a/file/of/bpf_iter_target.o
+
+
+DESCRIPTION
+===========
+	**bpftool iter pin** *OBJ* *PATH*
+		  A bpf iterator combines a kernel iterating of
+		  particular kernel data (e.g., tasks, bpf_maps, etc.)
+		  and a bpf program called for each kernel data object
+		  (e.g., one task, one bpf_map, etc.). User space can
+		  *read* kernel iterator output through *read()* syscall.
+
+		  The *pin* command creates a bpf iterator from *OBJ*,
+		  and pin it to *PATH*. The *PATH* should be located
+		  in *bpffs* mount. It must not contain a dot
+		  character ('.'), which is reserved for future extensions
+		  of *bpffs*.
+
+		  User can then *cat PATH* to see the bpf iterator output.
+
+	**bpftool iter help**
+		  Print short help message.
+
+OPTIONS
+=======
+	-h, --help
+		  Print short generic help message (similar to **bpftool help**).
+
+	-V, --version
+		  Print version number (similar to **bpftool version**).
+
+	-d, --debug
+		  Print all logs available, even debug-level information. This
+		  includes logs from libbpf as well as from the verifier, when
+		  attempting to load programs.
+
+EXAMPLES
+========
+**# bpftool iter pin bpf_iter_netlink.o /sys/fs/bpf/my_netlink**
+
+::
+
+   Create a file-based bpf iterator from bpf_iter_netlink.o and pin it
+   to /sys/fs/bpf/my_netlink
+
+
+SEE ALSO
+========
+	**bpf**\ (2),
+	**bpf-helpers**\ (7),
+	**bpftool**\ (8),
+	**bpftool-prog**\ (8),
+	**bpftool-map**\ (8),
+	**bpftool-link**\ (8),
+	**bpftool-cgroup**\ (8),
+	**bpftool-feature**\ (8),
+	**bpftool-net**\ (8),
+	**bpftool-perf**\ (8),
+	**bpftool-btf**\ (8)
+	**bpftool-gen**\ (8)
+	**bpftool-struct_ops**\ (8)
tools/bpf/bpftool/bash-completion/bpftool +13 -0
index fc989ead7313..9f0f20e73b87 100644
@@ -610,6 +610,19 @@ _bpftool()
                     ;;
             esac
             ;;
+        iter)
+            case $command in
+                pin)
+                    _filedir
+                    return 0
+                    ;;
+                *)
+                    [[ $prev == $object ]] && \
+                        COMPREPLY=( $( compgen -W 'pin help' \
+                            -- "$cur" ) )
+                    ;;
+            esac
+            ;;
         map)
             local MAP_TYPE='id pinned name'
             case $command in
tools/bpf/bpftool/iter.c new file +88 -0

88 lines not on this page · raw message

tools/bpf/bpftool/link.c +1 -0
index adc7dc431ed8..b6a0b35c78ae 100644
@@ -16,6 +16,7 @@ static const char * const link_type_name[] = {
 	[BPF_LINK_TYPE_RAW_TRACEPOINT]		= "raw_tracepoint",
 	[BPF_LINK_TYPE_TRACING]			= "tracing",
 	[BPF_LINK_TYPE_CGROUP]			= "cgroup",
+	[BPF_LINK_TYPE_ITER]			= "iter",
 };
 
 static int link_parse_fd(int *argc, char ***argv)
tools/bpf/bpftool/main.c +2 -1
index 1413a154806e..46bd716a9d86 100644
@@ -59,7 +59,7 @@ static int do_help(int argc, char **argv)
 		"       %s batch file FILE\n"
 		"       %s version\n"
 		"\n"
-		"       OBJECT := { prog | map | link | cgroup | perf | net | feature | btf | gen | struct_ops }\n"
+		"       OBJECT := { prog | map | link | cgroup | perf | net | feature | btf | gen | struct_ops | iter }\n"
 		"       " HELP_SPEC_OPTIONS "\n"
 		"",
 		bin_name, bin_name, bin_name);
@@ -224,6 +224,7 @@ static const struct cmd cmds[] = {
 	{ "btf",	do_btf },
 	{ "gen",	do_gen },
 	{ "struct_ops",	do_struct_ops },
+	{ "iter",	do_iter },
 	{ "version",	do_version },
 	{ 0 }
 };
tools/bpf/bpftool/main.h +1 -0
index 9b1fb81a8331..a41cefabccaf 100644
@@ -199,6 +199,7 @@ int do_feature(int argc, char **argv);
 int do_btf(int argc, char **argv);
 int do_gen(int argc, char **argv);
 int do_struct_ops(int argc, char **argv);
+int do_iter(int argc, char **argv);
 
 int parse_u32_arg(int *argc, char ***argv, __u32 *val, const char *what);
 int prog_parse_fd(int *argc, char ***argv);
Signature · 1 line
2.24.1
#34 lore raw

PATCH v4 19/21 tools/bpf: selftests: add iterator programs for ipv6_route and netlink

In reply to Yonghong Song

Two bpf programs are added in this patch for netlink and ipv6_route target. On my VM, I am able to achieve identical
results compared to /proc/net/netlink and /proc/net/ipv6_route.

  $ cat /proc/net/netlink
  sk               Eth Pid        Groups   Rmem     Wmem     Dump  Locks    Drops    Inode
  000000002c42d58b 0   0          00000000 0        0        0     2        0        7
  00000000a4e8b5e1 0   1          00000551 0        0        0     2        0        18719
  00000000e1b1c195 4   0          00000000 0        0        0     2        0        16422
  000000007e6b29f9 6   0          00000000 0        0        0     2        0        16424
  ....
  00000000159a170d 15  1862       00000002 0        0        0     2        0        1886
  000000009aca4bc9 15  3918224839 00000002 0        0        0     2        0        19076
  00000000d0ab31d2 15  1          00000002 0        0        0     2        0        18683
  000000008398fb08 16  0          00000000 0        0        0     2        0        27
  $ cat /sys/fs/bpf/my_netlink
  sk               Eth Pid        Groups   Rmem     Wmem     Dump  Locks    Drops    Inode
  000000002c42d58b 0   0          00000000 0        0        0     2        0        7
  00000000a4e8b5e1 0   1          00000551 0        0        0     2        0        18719
  00000000e1b1c195 4   0          00000000 0        0        0     2        0        16422
  000000007e6b29f9 6   0          00000000 0        0        0     2        0        16424
  ....
  00000000159a170d 15  1862       00000002 0        0        0     2        0        1886
  000000009aca4bc9 15  3918224839 00000002 0        0        0     2        0        19076
  00000000d0ab31d2 15  1          00000002 0        0        0     2        0        18683
  000000008398fb08 16  0          00000000 0        0        0     2        0        27
  $ cat /proc/net/ipv6_route
  fe800000000000000000000000000000 40 00000000000000000000000000000000 00 00000000000000000000000000000000 00000100 00000001 00000000 00000001     eth0
  00000000000000000000000000000000 00 00000000000000000000000000000000 00 00000000000000000000000000000000 ffffffff 00000001 00000000 00200200       lo
  00000000000000000000000000000001 80 00000000000000000000000000000000 00 00000000000000000000000000000000 00000000 00000003 00000000 80200001       lo
  fe80000000000000c04b03fffe7827ce 80 00000000000000000000000000000000 00 00000000000000000000000000000000 00000000 00000002 00000000 80200001     eth0
  ff000000000000000000000000000000 08 00000000000000000000000000000000 00 00000000000000000000000000000000 00000100 00000003 00000000 00000001     eth0
  00000000000000000000000000000000 00 00000000000000000000000000000000 00 00000000000000000000000000000000 ffffffff 00000001 00000000 00200200       lo
  $ cat /sys/fs/bpf/my_ipv6_route
  fe800000000000000000000000000000 40 00000000000000000000000000000000 00 00000000000000000000000000000000 00000100 00000001 00000000 00000001     eth0
  00000000000000000000000000000000 00 00000000000000000000000000000000 00 00000000000000000000000000000000 ffffffff 00000001 00000000 00200200       lo
  00000000000000000000000000000001 80 00000000000000000000000000000000 00 00000000000000000000000000000000 00000000 00000003 00000000 80200001       lo
  fe80000000000000c04b03fffe7827ce 80 00000000000000000000000000000000 00 00000000000000000000000000000000 00000000 00000002 00000000 80200001     eth0
  ff000000000000000000000000000000 08 00000000000000000000000000000000 00 00000000000000000000000000000000 00000100 00000003 00000000 00000001     eth0
  00000000000000000000000000000000 00 00000000000000000000000000000000 00 00000000000000000000000000000000 ffffffff 00000001 00000000 00200200       lo
---
 .../selftests/bpf/progs/bpf_iter_ipv6_route.c | 62 +++++++++++++++++
 .../selftests/bpf/progs/bpf_iter_netlink.c    | 66 +++++++++++++++++++
 2 files changed, 128 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_ipv6_route.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_netlink.c
tools/testing/selftests/bpf/progs/bpf_iter_ipv6_route.c new file +62 -0
new file mode 100644
index 000000000000..ab9e2650e021
@@ -0,0 +1,62 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2020 Facebook */
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+char _license[] SEC("license") = "GPL";
+
+extern bool CONFIG_IPV6_SUBTREES __kconfig __weak;
+
+#define RTF_GATEWAY		0x0002
+#define IFNAMSIZ		16
+#define fib_nh_gw_family	nh_common.nhc_gw_family
+#define fib_nh_gw6		nh_common.nhc_gw.ipv6
+#define fib_nh_dev		nh_common.nhc_dev
+
+SEC("iter/ipv6_route")
+int dump_ipv6_route(struct bpf_iter__ipv6_route *ctx)
+{
+	struct seq_file *seq = ctx->meta->seq;
+	struct fib6_info *rt = ctx->rt;
+	const struct net_device *dev;
+	struct fib6_nh *fib6_nh;
+	unsigned int flags;
+	struct nexthop *nh;
+
+	if (rt == (void *)0)
+		return 0;
+
+	fib6_nh = &rt->fib6_nh[0];
+	flags = rt->fib6_flags;
+
+	/* FIXME: nexthop_is_multipath is not handled here. */
+	nh = rt->nh;
+	if (rt->nh)
+		fib6_nh = &nh->nh_info->fib6_nh;
+
+	BPF_SEQ_PRINTF(seq, "%pi6 %02x ", &rt->fib6_dst.addr, rt->fib6_dst.plen);
+
+	if (CONFIG_IPV6_SUBTREES)
+		BPF_SEQ_PRINTF(seq, "%pi6 %02x ", &rt->fib6_src.addr,
+			       rt->fib6_src.plen);
+	else
+		BPF_SEQ_PRINTF(seq, "00000000000000000000000000000000 00 ");
+
+	if (fib6_nh->fib_nh_gw_family) {
+		flags |= RTF_GATEWAY;
+		BPF_SEQ_PRINTF(seq, "%pi6 ", &fib6_nh->fib_nh_gw6);
+	} else {
+		BPF_SEQ_PRINTF(seq, "00000000000000000000000000000000 ");
+	}
+
+	dev = fib6_nh->fib_nh_dev;
+	if (dev)
+		BPF_SEQ_PRINTF(seq, "%08x %08x %08x %08x %8s\n", rt->fib6_metric,
+			       rt->fib6_ref.refs.counter, 0, flags, dev->name);
+	else
+		BPF_SEQ_PRINTF(seq, "%08x %08x %08x %08x\n", rt->fib6_metric,
+			       rt->fib6_ref.refs.counter, 0, flags);
+
+	return 0;
+}
tools/testing/selftests/bpf/progs/bpf_iter_netlink.c new file +66 -0

66 lines not on this page · raw message

Signature · 1 line
2.24.1
#35 lore raw

PATCH v4 20/21 tools/bpf: selftests: add iter progs for bpf_map/task/task_file

In reply to Yonghong Song

The implementation is arbitrary, just to show how the bpf programs can be written for bpf_map/task/task_file. They can be costomized for specific needs.

For example, for bpf_map, the iterator prints out:

  $ cat /sys/fs/bpf/my_bpf_map
      id   refcnt  usercnt  locked_vm
       3        2        0         20
       6        2        0         20
       9        2        0         20
      12        2        0         20
      13        2        0         20
      16        2        0         20
      19        2        0         20
      %%% END %%%

For task, the iterator prints out:

  $ cat /sys/fs/bpf/my_task
    tgid      gid
       1        1
       2        2
    ....
    1944     1944
    1948     1948
    1949     1949
    1953     1953
    === END ===

For task/file, the iterator prints out:

  $ cat /sys/fs/bpf/my_task_file
    tgid      gid       fd      file
       1        1        0 ffffffff95c97600
       1        1        1 ffffffff95c97600
       1        1        2 ffffffff95c97600
    ....
    1895     1895      255 ffffffff95c8fe00
    1932     1932        0 ffffffff95c8fe00
    1932     1932        1 ffffffff95c8fe00
    1932     1932        2 ffffffff95c8fe00
    1932     1932        3 ffffffff95c185c0

This is able to print out all open files (fd and file->f_op), so user can compare f_op against a particular kernel file operations to find what it is. For example, from /proc/kallsyms, we can find
ffffffff95c185c0 r eventfd_fops
so we will know tgid 1932 fd 3 is an eventfd file descriptor.

---
 .../selftests/bpf/progs/bpf_iter_bpf_map.c    | 28 +++++++++++++++++++
 .../selftests/bpf/progs/bpf_iter_task.c       | 25 +++++++++++++++++
 .../selftests/bpf/progs/bpf_iter_task_file.c  | 26 +++++++++++++++++
 3 files changed, 79 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_bpf_map.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_task.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_task_file.c
tools/testing/selftests/bpf/progs/bpf_iter_bpf_map.c new file +28 -0
new file mode 100644
index 000000000000..4867cd3445c8
@@ -0,0 +1,28 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2020 Facebook */
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+char _license[] SEC("license") = "GPL";
+
+SEC("iter/bpf_map")
+int dump_bpf_map(struct bpf_iter__bpf_map *ctx)
+{
+	struct seq_file *seq = ctx->meta->seq;
+	__u64 seq_num = ctx->meta->seq_num;
+	struct bpf_map *map = ctx->map;
+
+	if (map == (void *)0) {
+		BPF_SEQ_PRINTF(seq, "      %%%%%% END %%%%%%\n");
+		return 0;
+	}
+
+	if (seq_num == 0)
+		BPF_SEQ_PRINTF(seq, "      id   refcnt  usercnt  locked_vm\n");
+
+	BPF_SEQ_PRINTF(seq, "%8u %8ld %8ld %10lu\n", map->id, map->refcnt.counter,
+		       map->usercnt.counter,
+		       map->memory.user->locked_vm.counter);
+	return 0;
+}
tools/testing/selftests/bpf/progs/bpf_iter_task.c new file +25 -0
new file mode 100644
index 000000000000..90f9011c57ca
@@ -0,0 +1,25 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2020 Facebook */
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+char _license[] SEC("license") = "GPL";
+
+SEC("iter/task")
+int dump_task(struct bpf_iter__task *ctx)
+{
+	struct seq_file *seq = ctx->meta->seq;
+	struct task_struct *task = ctx->task;
+
+	if (task == (void *)0) {
+		BPF_SEQ_PRINTF(seq, "    === END ===\n");
+		return 0;
+	}
+
+	if (ctx->meta->seq_num == 0)
+		BPF_SEQ_PRINTF(seq, "    tgid      gid\n");
+
+	BPF_SEQ_PRINTF(seq, "%8d %8d\n", task->tgid, task->pid);
+	return 0;
+}
tools/testing/selftests/bpf/progs/bpf_iter_task_file.c new file +26 -0
new file mode 100644
index 000000000000..c6ced38f0880
@@ -0,0 +1,26 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2020 Facebook */
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+
+char _license[] SEC("license") = "GPL";
+
+SEC("iter/task_file")
+int dump_task_file(struct bpf_iter__task_file *ctx)
+{
+	struct seq_file *seq = ctx->meta->seq;
+	struct task_struct *task = ctx->task;
+	__u32 fd = ctx->fd;
+	struct file *file = ctx->file;
+
+	if (task == (void *)0 || file == (void *)0)
+		return 0;
+
+	if (ctx->meta->seq_num == 0)
+		BPF_SEQ_PRINTF(seq, "    tgid      gid       fd      file\n");
+
+	BPF_SEQ_PRINTF(seq, "%8d %8d %8d %lx\n", task->tgid, task->pid, fd,
+		       (long)file->f_op);
+	return 0;
+}
Signature · 1 line
2.24.1
#36 lore raw

PATCH v4 21/21 tools/bpf: selftests: add bpf_iter selftests

In reply to Yonghong Song

The added test includes the following subtests:
- test verifier change for btf_id_or_null
- test load/create_iter/read for ipv6_route/netlink/bpf_map/task/task_file
- test anon bpf iterator
- test anon bpf iterator reading one char at a time
- test file bpf iterator
- test overflow (single bpf program output not overflow)
- test overflow (single bpf program output overflows)
- test bpf prog returning 1

The ipv6_route tests the following verifier change
- access fields in the variable length array of the structure.

The netlink load tests the following verifier change
- put a btf_id ptr value in a stack and accessible to tracing/iter programs.

The anon bpf iterator also tests link auto attach through skeleton.

  $ test_progs -n 2
  #2/1 btf_id_or_null:OK
  #2/2 ipv6_route:OK
  #2/3 netlink:OK
  #2/4 bpf_map:OK
  #2/5 task:OK
  #2/6 task_file:OK
  #2/7 anon:OK
  #2/8 anon-read-one-char:OK
  #2/9 file:OK
  #2/10 overflow:OK
  #2/11 overflow-e2big:OK
  #2/12 prog-ret-1:OK
  #2 bpf_iter:OK
  Summary: 1/12 PASSED, 0 SKIPPED, 0 FAILED
---
 .../selftests/bpf/prog_tests/bpf_iter.c       | 409 ++++++++++++++++++
 .../selftests/bpf/progs/bpf_iter_test_kern1.c |   4 +
 .../selftests/bpf/progs/bpf_iter_test_kern2.c |   4 +
 .../selftests/bpf/progs/bpf_iter_test_kern3.c |  18 +
 .../selftests/bpf/progs/bpf_iter_test_kern4.c |  52 +++
 .../bpf/progs/bpf_iter_test_kern_common.h     |  22 +
 6 files changed, 509 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/bpf_iter.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_test_kern1.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_test_kern2.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_test_kern3.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_test_kern4.c
 create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_test_kern_common.h
tools/testing/selftests/bpf/prog_tests/bpf_iter.c new file +409 -0
new file mode 100644
index 000000000000..87c29dde1cf9
@@ -0,0 +1,409 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2020 Facebook */
+#include <test_progs.h>
+#include "bpf_iter_ipv6_route.skel.h"
+#include "bpf_iter_netlink.skel.h"
+#include "bpf_iter_bpf_map.skel.h"
+#include "bpf_iter_task.skel.h"
+#include "bpf_iter_task_file.skel.h"
+#include "bpf_iter_test_kern1.skel.h"
+#include "bpf_iter_test_kern2.skel.h"
+#include "bpf_iter_test_kern3.skel.h"
+#include "bpf_iter_test_kern4.skel.h"
+
+static int duration;
+
+static void test_btf_id_or_null(void)
+{
+	struct bpf_iter_test_kern3 *skel;
+
+	skel = bpf_iter_test_kern3__open_and_load();
+	if (CHECK(skel, "bpf_iter_test_kern3__open_and_load",
+		  "skeleton open_and_load unexpectedly succeeded\n")) {
+		bpf_iter_test_kern3__destroy(skel);
+		return;
+	}
+}
+
+static void do_dummy_read(struct bpf_program *prog)
+{
+	struct bpf_link *link;
+	char buf[16] = {};
+	int iter_fd, len;
+
+	link = bpf_program__attach_iter(prog, NULL);
+	if (CHECK(IS_ERR(link), "attach_iter", "attach_iter failed\n"))
+		return;
+
+	iter_fd = bpf_iter_create(bpf_link__fd(link));
+	if (CHECK(iter_fd < 0, "create_iter", "create_iter failed\n"))
+		goto free_link;
+
+	/* not check contents, but ensure read() ends without error */
+	while ((len = read(iter_fd, buf, sizeof(buf))) > 0)
+		;
+	CHECK(len < 0, "read", "read failed: %s\n", strerror(errno));
+
+	close(iter_fd);
+
+free_link:
+	bpf_link__destroy(link);
+}
+
+static void test_ipv6_route(void)
+{
+	struct bpf_iter_ipv6_route *skel;
+
+	skel = bpf_iter_ipv6_route__open_and_load();
+	if (CHECK(!skel, "bpf_iter_ipv6_route__open_and_load",
+		  "skeleton open_and_load failed\n"))
+		return;
+
+	do_dummy_read(skel->progs.dump_ipv6_route);
+
+	bpf_iter_ipv6_route__destroy(skel);
+}
+
+static void test_netlink(void)
+{
+	struct bpf_iter_netlink *skel;
+
+	skel = bpf_iter_netlink__open_and_load();
+	if (CHECK(!skel, "bpf_iter_netlink__open_and_load",
+		  "skeleton open_and_load failed\n"))
+		return;
+
+	do_dummy_read(skel->progs.dump_netlink);
+
+	bpf_iter_netlink__destroy(skel);
+}
+
+static void test_bpf_map(void)
+{
+	struct bpf_iter_bpf_map *skel;
+
+	skel = bpf_iter_bpf_map__open_and_load();
+	if (CHECK(!skel, "bpf_iter_bpf_map__open_and_load",
+		  "skeleton open_and_load failed\n"))
+		return;
+
+	do_dummy_read(skel->progs.dump_bpf_map);
+
+	bpf_iter_bpf_map__destroy(skel);
+}
+
+static void test_task(void)
+{
+	struct bpf_iter_task *skel;
+
+	skel = bpf_iter_task__open_and_load();
+	if (CHECK(!skel, "bpf_iter_task__open_and_load",
+		  "skeleton open_and_load failed\n"))
+		return;
+
+	do_dummy_read(skel->progs.dump_task);
+
+	bpf_iter_task__destroy(skel);
+}
+
+static void test_task_file(void)
+{
+	struct bpf_iter_task_file *skel;
+
+	skel = bpf_iter_task_file__open_and_load();
+	if (CHECK(!skel, "bpf_iter_task_file__open_and_load",
+		  "skeleton open_and_load failed\n"))
+		return;
+
+	do_dummy_read(skel->progs.dump_task_file);
+
+	bpf_iter_task_file__destroy(skel);
+}
+
+/* The expected string is less than 16 bytes */
+static int do_read_with_fd(int iter_fd, const char *expected,
+			   bool read_one_char)
+{
+	int err = -1, len, read_buf_len, start;
+	char buf[16] = {};
+
+	read_buf_len = read_one_char ? 1 : 16;
+	start = 0;
+	while ((len = read(iter_fd, buf + start, read_buf_len)) > 0) {
+		start += len;
+		if (CHECK(start >= 16, "read", "read len %d\n", len))
+			return -1;
+		read_buf_len = read_one_char ? 1 : 16 - start;
+	}
+	if (CHECK(len < 0, "read", "read failed: %s\n", strerror(errno)))
+		return -1;
+
+	err = strcmp(buf, expected);
+	if (CHECK(err, "read", "incorrect read result: buf %s, expected %s\n",
+		  buf, expected))
+		return -1;
+
+	return 0;
+}
+
+static void test_anon_iter(bool read_one_char)
+{
+	struct bpf_iter_test_kern1 *skel;
+	struct bpf_link *link;
+	int iter_fd, err;
+
+	skel = bpf_iter_test_kern1__open_and_load();
+	if (CHECK(!skel, "bpf_iter_test_kern1__open_and_load",
+		  "skeleton open_and_load failed\n"))
+		return;
+
+	err = bpf_iter_test_kern1__attach(skel);
+	if (CHECK(err, "bpf_iter_test_kern1__attach",
+		  "skeleton attach failed\n")) {
+		goto out;
+	}
+
+	link = skel->links.dump_task;
+	iter_fd = bpf_iter_create(bpf_link__fd(link));
+	if (CHECK(iter_fd < 0, "create_iter", "create_iter failed\n"))
+		goto out;
+
+	do_read_with_fd(iter_fd, "abcd", read_one_char);
+	close(iter_fd);
+
+out:
+	bpf_iter_test_kern1__destroy(skel);
+}
+
+static int do_read(const char *path, const char *expected)
+{
+	int err, iter_fd;
+
+	iter_fd = open(path, O_RDONLY);
+	if (CHECK(iter_fd < 0, "open", "open %s failed: %s\n",
+		  path, strerror(errno)))
+		return -1;
+
+	err = do_read_with_fd(iter_fd, expected, false);
+	close(iter_fd);
+	return err;
+}
+
+static void test_file_iter(void)
+{
+	const char *path = "/sys/fs/bpf/bpf_iter_test1";
+	struct bpf_iter_test_kern1 *skel1;
+	struct bpf_iter_test_kern2 *skel2;
+	struct bpf_link *link;
+	int err;
+
+	skel1 = bpf_iter_test_kern1__open_and_load();
+	if (CHECK(!skel1, "bpf_iter_test_kern1__open_and_load",
+		  "skeleton open_and_load failed\n"))
+		return;
+
+	link = bpf_program__attach_iter(skel1->progs.dump_task, NULL);
+	if (CHECK(IS_ERR(link), "attach_iter", "attach_iter failed\n"))
+		goto out;
+
+	/* unlink this path if it exists. */
+	unlink(path);
+
+	err = bpf_link__pin(link, path);
+	if (CHECK(err, "pin_iter", "pin_iter to %s failed: %d\n", path, err))
+		goto free_link;
+
+	err = do_read(path, "abcd");
+	if (err)
+		goto unlink_path;
+
+	/* file based iterator seems working fine. Let us a link update
+	 * of the underlying link and `cat` the iterator again, its content
+	 * should change.
+	 */
+	skel2 = bpf_iter_test_kern2__open_and_load();
+	if (CHECK(!skel2, "bpf_iter_test_kern2__open_and_load",
+		  "skeleton open_and_load failed\n"))
+		goto unlink_path;
+
+	err = bpf_link__update_program(link, skel2->progs.dump_task);
+	if (CHECK(err, "update_prog", "update_prog failed\n"))
+		goto destroy_skel2;
+
+	do_read(path, "ABCD");
+
+destroy_skel2:
+	bpf_iter_test_kern2__destroy(skel2);
+unlink_path:
+	unlink(path);
+free_link:
+	bpf_link__destroy(link);
+out:
+	bpf_iter_test_kern1__destroy(skel1);
+}
+
+static void test_overflow(bool test_e2big_overflow, bool ret1)
+{
+	__u32 map_info_len, total_read_len, expected_read_len;
+	int err, iter_fd, map1_fd, map2_fd, len;
+	struct bpf_map_info map_info = {};
+	struct bpf_iter_test_kern4 *skel;
+	struct bpf_link *link;
+	__u32 page_size;
+	char *buf;
+
+	skel = bpf_iter_test_kern4__open();
+	if (CHECK(!skel, "bpf_iter_test_kern4__open",
+		  "skeleton open failed\n"))
+		return;
+
+	/* create two maps: bpf program will only do bpf_seq_write
+	 * for these two maps. The goal is one map output almost
+	 * fills seq_file buffer and then the other will trigger
+	 * overflow and needs restart.
+	 */
+	map1_fd = bpf_create_map(BPF_MAP_TYPE_ARRAY, 4, 8, 1, 0);
+	if (CHECK(map1_fd < 0, "bpf_create_map",
+		  "map_creation failed: %s\n", strerror(errno)))
+		goto out;
+	map2_fd = bpf_create_map(BPF_MAP_TYPE_ARRAY, 4, 8, 1, 0);
+	if (CHECK(map2_fd < 0, "bpf_create_map",
+		  "map_creation failed: %s\n", strerror(errno)))
+		goto free_map1;
+
+	/* bpf_seq_printf kernel buffer is one page, so one map
+	 * bpf_seq_write will mostly fill it, and the other map
+	 * will partially fill and then trigger overflow and need
+	 * bpf_seq_read restart.
+	 */
+	page_size = sysconf(_SC_PAGE_SIZE);
+
+	if (test_e2big_overflow) {
+		skel->rodata->print_len = (page_size + 8) / 8;
+		expected_read_len = 2 * (page_size + 8);
+	} else if (!ret1) {
+		skel->rodata->print_len = (page_size - 8) / 8;
+		expected_read_len = 2 * (page_size - 8);
+	} else {
+		skel->rodata->print_len = 1;
+		expected_read_len = 2 * 8;
+	}
+	skel->rodata->ret1 = ret1;
+
+	if (CHECK(bpf_iter_test_kern4__load(skel),
+		  "bpf_iter_test_kern4__load", "skeleton load failed\n"))
+		goto free_map2;
+
+	/* setup filtering map_id in bpf program */
+	map_info_len = sizeof(map_info);
+	err = bpf_obj_get_info_by_fd(map1_fd, &map_info, &map_info_len);
+	if (CHECK(err, "get_map_info", "get map info failed: %s\n",
+		  strerror(errno)))
+		goto free_map2;
+	skel->bss->map1_id = map_info.id;
+
+	err = bpf_obj_get_info_by_fd(map2_fd, &map_info, &map_info_len);
+	if (CHECK(err, "get_map_info", "get map info failed: %s\n",
+		  strerror(errno)))
+		goto free_map2;
+	skel->bss->map2_id = map_info.id;
+
+	link = bpf_program__attach_iter(skel->progs.dump_bpf_map, NULL);
+	if (CHECK(IS_ERR(link), "attach_iter", "attach_iter failed\n"))
+		goto free_map2;
+
+	iter_fd = bpf_iter_create(bpf_link__fd(link));
+	if (CHECK(iter_fd < 0, "create_iter", "create_iter failed\n"))
+		goto free_link;
+
+	buf = malloc(expected_read_len);
+	if (!buf)
+		goto close_iter;
+
+	/* do read */
+	total_read_len = 0;
+	if (test_e2big_overflow) {
+		while ((len = read(iter_fd, buf, expected_read_len)) > 0)
+			total_read_len += len;
+
+		CHECK(len != -1 || errno != E2BIG, "read",
+		      "expected ret -1, errno E2BIG, but get ret %d, error %s\n",
+			  len, strerror(errno));
+		goto free_buf;
+	} else if (!ret1) {
+		while ((len = read(iter_fd, buf, expected_read_len)) > 0)
+			total_read_len += len;
+
+		if (CHECK(len < 0, "read", "read failed: %s\n",
+			  strerror(errno)))
+			goto free_buf;
+	} else {
+		do {
+			len = read(iter_fd, buf, expected_read_len);
+			if (len > 0)
+				total_read_len += len;
+		} while (len > 0 || len == -EAGAIN);
+
+		if (CHECK(len < 0, "read", "read failed: %s\n",
+			  strerror(errno)))
+			goto free_buf;
+	}
+
+	if (CHECK(total_read_len != expected_read_len, "read",
+		  "total len %u, expected len %u\n", total_read_len,
+		  expected_read_len))
+		goto free_buf;
+
+	if (CHECK(skel->bss->map1_accessed != 1, "map1_accessed",
+		  "expected 1 actual %d\n", skel->bss->map1_accessed))
+		goto free_buf;
+
+	if (CHECK(skel->bss->map2_accessed != 2, "map2_accessed",
+		  "expected 2 actual %d\n", skel->bss->map2_accessed))
+		goto free_buf;
+
+	CHECK(skel->bss->map2_seqnum1 != skel->bss->map2_seqnum2,
+	      "map2_seqnum", "two different seqnum %lld %lld\n",
+	      skel->bss->map2_seqnum1, skel->bss->map2_seqnum2);
+
+free_buf:
+	free(buf);
+close_iter:
+	close(iter_fd);
+free_link:
+	bpf_link__destroy(link);
+free_map2:
+	close(map2_fd);
+free_map1:
+	close(map1_fd);
+out:
+	bpf_iter_test_kern4__destroy(skel);
+}
+
+void test_bpf_iter(void)
+{
+	if (test__start_subtest("btf_id_or_null"))
+		test_btf_id_or_null();
+	if (test__start_subtest("ipv6_route"))
+		test_ipv6_route();
+	if (test__start_subtest("netlink"))
+		test_netlink();
+	if (test__start_subtest("bpf_map"))
+		test_bpf_map();
+	if (test__start_subtest("task"))
+		test_task();
+	if (test__start_subtest("task_file"))
+		test_task_file();
+	if (test__start_subtest("anon"))
+		test_anon_iter(false);
+	if (test__start_subtest("anon-read-one-char"))
+		test_anon_iter(true);
+	if (test__start_subtest("file"))
+		test_file_iter();
+	if (test__start_subtest("overflow"))
+		test_overflow(false, false);
+	if (test__start_subtest("overflow-e2big"))
+		test_overflow(true, false);
+	if (test__start_subtest("prog-ret-1"))
+		test_overflow(false, true);
+}
tools/testing/selftests/bpf/progs/bpf_iter_test_kern1.c new file +4 -0
new file mode 100644
index 000000000000..c71a7c283108
@@ -0,0 +1,4 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2020 Facebook */
+#define START_CHAR 'a'
+#include "bpf_iter_test_kern_common.h"
tools/testing/selftests/bpf/progs/bpf_iter_test_kern2.c new file +4 -0
new file mode 100644
index 000000000000..8bdc8dc07444
@@ -0,0 +1,4 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2020 Facebook */
+#define START_CHAR 'A'
+#include "bpf_iter_test_kern_common.h"
tools/testing/selftests/bpf/progs/bpf_iter_test_kern3.c new file +18 -0
new file mode 100644
index 000000000000..636a00fa074d
@@ -0,0 +1,18 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2020 Facebook */
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+
+char _license[] SEC("license") = "GPL";
+
+SEC("iter/task")
+int dump_task(struct bpf_iter__task *ctx)
+{
+	struct seq_file *seq = ctx->meta->seq;
+	struct task_struct *task = ctx->task;
+	int tgid;
+
+	tgid = task->tgid;
+	bpf_seq_write(seq, &tgid, sizeof(tgid));
+	return 0;
+}
tools/testing/selftests/bpf/progs/bpf_iter_test_kern4.c new file +52 -0

52 lines not on this page · raw message

tools/testing/selftests/bpf/progs/bpf_iter_test_kern_common.h new file +22 -0
new file mode 100644
index 000000000000..bdd51cf14b54
@@ -0,0 +1,22 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/* Copyright (c) 2020 Facebook */
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+
+char _license[] SEC("license") = "GPL";
+int count = 0;
+
+SEC("iter/task")
+int dump_task(struct bpf_iter__task *ctx)
+{
+	struct seq_file *seq = ctx->meta->seq;
+	char c;
+
+	if (count < 4) {
+		c = START_CHAR + count;
+		bpf_seq_write(seq, &c, sizeof(c));
+		count++;
+	}
+
+	return 0;
+}
Signature · 1 line
2.24.1
#37 lore raw

Re: [PATCH bpf-next v4 21/21] tools/bpf: selftests: add bpf_iter selftests

Quoted 30 lines · Yonghong Song
tools/testing/selftests/bpf/progs/bpf_iter_test_kern4.c new file +30 -0
+static volatile const __u32 ret1;
+
+SEC("iter/bpf_map")
+int dump_bpf_map(struct bpf_iter__bpf_map *ctx)
+{
+	struct seq_file *seq = ctx->meta->seq;
+	struct bpf_map *map = ctx->map;
+	__u64 seq_num;
+	int i, ret = 0;
+
+	if (map == (void *)0)
+		return 0;
+
+	/* only dump map1_id and map2_id */
+	if (map->id != map1_id && map->id != map2_id)
+		return 0;
+
+	seq_num = ctx->meta->seq_num;
+	if (map->id == map1_id) {
+		map1_seqnum = seq_num;
+		map1_accessed++;
+	}
+
+	if (map->id == map2_id) {
+		if (map2_accessed == 0) {
+			map2_seqnum1 = seq_num;
+			if (ret1)
+				ret = 1;
+		} else {
+			map2_seqnum2 = seq_num;

tools/testing/selftests/bpf/progs/bpf_iter_test_kern4.c line 52

Quoted 10 lines
tools/testing/selftests/bpf/progs/bpf_iter_test_kern4.c new file +10 -0
+		}
+		map2_accessed++;
+	}
+
+	/* fill seq_file buffer */
+	for (i = 0; i < print_len; i++)
+		bpf_seq_write(seq, &seq_num, sizeof(seq_num));
+
+	return ret;
+}

I couldn't find where 'return 1' behavior is documented clearly. I think it's a workaround for overflow.
When bpf prog detects overflow it can request replay of the element? What if it keeps returning 1 ? read() will never finish?

#38 lore raw

Re: [PATCH bpf-next v4 21/21] tools/bpf: selftests: add bpf_iter selftests

Quoted 35 lines · Alexei Starovoitov
Quoted 34 lines · Yonghong Song
tools/testing/selftests/bpf/progs/bpf_iter_test_kern4.c new file +34 -0
+static volatile const __u32 ret1;
+
+SEC("iter/bpf_map")
+int dump_bpf_map(struct bpf_iter__bpf_map *ctx)
+{
+	struct seq_file *seq = ctx->meta->seq;
+	struct bpf_map *map = ctx->map;
+	__u64 seq_num;
+	int i, ret = 0;
+
+	if (map == (void *)0)
+		return 0;
+
+	/* only dump map1_id and map2_id */
+	if (map->id != map1_id && map->id != map2_id)
+		return 0;
+
+	seq_num = ctx->meta->seq_num;
+	if (map->id == map1_id) {
+		map1_seqnum = seq_num;
+		map1_accessed++;
+	}
+
+	if (map->id == map2_id) {
+		if (map2_accessed == 0) {
+			map2_seqnum1 = seq_num;
+			if (ret1)
+				ret = 1;
+		} else {
+			map2_seqnum2 = seq_num;
+		}
+		map2_accessed++;
+	}
+
Quoted 8 lines

tools/testing/selftests/bpf/progs/bpf_iter_test_kern4.c line 52

Quoted 6 lines
tools/testing/selftests/bpf/progs/bpf_iter_test_kern4.c new file +6 -0
+	/* fill seq_file buffer */
+	for (i = 0; i < print_len; i++)
+		bpf_seq_write(seq, &seq_num, sizeof(seq_num));
+
+	return ret;
+}

I couldn't find where 'return 1' behavior is documented clearly.

It is in the commit comments:

commit 15d83c4d7cef5c067a8b075ce59e97df4f60706e

Author: Yonghong Song <[email protected]>
Date:   Sat May 9 10:59:00 2020 -0700

bpf: Allow loading of a bpf_iter program
...
The program return value must be 0 or 1 for now.
0 : successful, except potential seq_file buffer overflow
which is handled by seq_file reader.
1 : request to restart the same object

Internally, bpf program returning 1 will translate show() return -EAGAIN and this error code will return to user space.

I will add some comments in the code to
document this behavior.

Quoted 1 line

I think it's a workaround for overflow.

This can be used for overflow but overflow already been taken care of by bpf_seq_read(). This is mostly used for other use cases:
- currently under RT-linux, bpf_seq_printf() may return -EBUSY. In this case, bpf program itself can request retrying the same object.
- for other conditions where bpf program itself wants to retry the same object. For example, hash table full, the bpf progam can return 1, in which case, user space read() will receive -EAGAIN and may check and make room for hash table and then read() again.

Quoted 1 line

When bpf prog detects overflow it can request replay of the element?

It can. But it can return 0 too since bpf_seq_read() handles this transparently.

Quoted 1 line

What if it keeps returning 1 ? read() will never finish?

The read() will finish and return -EAGAIN to user space. It is up to user space to decide whether to call read() again or not.

#39 lore raw

Re: [PATCH bpf-next v4 21/21] tools/bpf: selftests: add bpf_iter selftests

Quoted 84 lines · Yonghong Song
Quoted 35 lines · Alexei Starovoitov
Quoted 34 lines · Yonghong Song
tools/testing/selftests/bpf/progs/bpf_iter_test_kern4.c new file +34 -0
+static volatile const __u32 ret1;
+
+SEC("iter/bpf_map")
+int dump_bpf_map(struct bpf_iter__bpf_map *ctx)
+{
+    struct seq_file *seq = ctx->meta->seq;
+    struct bpf_map *map = ctx->map;
+    __u64 seq_num;
+    int i, ret = 0;
+
+    if (map == (void *)0)
+            return 0;
+
+    /* only dump map1_id and map2_id */
+    if (map->id != map1_id && map->id != map2_id)
+            return 0;
+
+    seq_num = ctx->meta->seq_num;
+    if (map->id == map1_id) {
+            map1_seqnum = seq_num;
+            map1_accessed++;
+    }
+
+    if (map->id == map2_id) {
+            if (map2_accessed == 0) {
+                    map2_seqnum1 = seq_num;
+                    if (ret1)
+                            ret = 1;
+            } else {
+                    map2_seqnum2 = seq_num;
+            }
+            map2_accessed++;
+    }
+
Quoted 8 lines

tools/testing/selftests/bpf/progs/bpf_iter_test_kern4.c line 52

Quoted 6 lines
tools/testing/selftests/bpf/progs/bpf_iter_test_kern4.c new file +6 -0
+    /* fill seq_file buffer */
+    for (i = 0; i < print_len; i++)
+            bpf_seq_write(seq, &seq_num, sizeof(seq_num));
+
+    return ret;
+}

I couldn't find where 'return 1' behavior is documented clearly.

It is in the commit comments:

commit 15d83c4d7cef5c067a8b075ce59e97df4f60706e

Author: Yonghong Song <[email protected]>
Date:   Sat May 9 10:59:00 2020 -0700

bpf: Allow loading of a bpf_iter program
...
The program return value must be 0 or 1 for now.
0 : successful, except potential seq_file buffer overflow
which is handled by seq_file reader.
1 : request to restart the same object

Internally, bpf program returning 1 will translate show() return -EAGAIN and this error code will return to user space.

I will add some comments in the code to
document this behavior.

Quoted 1 line

I think it's a workaround for overflow.

This can be used for overflow but overflow already been taken care of by bpf_seq_read(). This is mostly used for other use cases:
- currently under RT-linux, bpf_seq_printf() may return -EBUSY. In this case, bpf program itself can request retrying the same object.
- for other conditions where bpf program itself wants to retry the same object. For example, hash table full, the bpf progam can return 1, in which case, user space read() will receive -EAGAIN and may check and make room for hash table and then read() again.

Quoted 1 line

When bpf prog detects overflow it can request replay of the element?

Quoted 8 lines

It can. But it can return 0 too since bpf_seq_read() handles this transparently.

Quoted 1 line

What if it keeps returning 1 ? read() will never finish?

The read() will finish and return -EAGAIN to user space. It is up to user space to decide whether to call read() again or not.

Ahh. Got it. So that EAGAIN returned by bpf_iter_run_prog() propagates by bpf_seq_read() all the way to read() syscall. Now I see it. Thanks for explaining.

#40 lore raw
In reply to Yonghong Song
Quoted 10 lines · Yonghong Song

Changelog:
v3 -> v4:
- in bpf_seq_read(), if start() failed with an error, return that error to user space (Andrii)
- in bpf_seq_printf(), if reading kernel memory failed for %s and %p{i,I}{4,6}, set buffer to empty string or address 0. Documented this behavior in uapi header (Andrii)
- fix a few error handling issues for bpftool (Andrii)
- A few other minor fixes and cosmetic changes.

Looks great overall. Applied.
But few follow ups are necessary.

The main gotcha is that new tests need llvm with the fix
https://reviews.llvm.org/D78466.
I think it was applied to llvm 10 branch already,
but please add selftests/bpf/README.rst and mention
that above llvm commit is necessary to successfully pass the tests. Also mention the verifier error that folks will see when llvm is buggy.

Few other nits I noticed in relevant patches.

#41 lore raw
Quoted 15 lines · Alexei Starovoitov
Quoted 10 lines · Yonghong Song

Changelog:
v3 -> v4:
- in bpf_seq_read(), if start() failed with an error, return that error to user space (Andrii)
- in bpf_seq_printf(), if reading kernel memory failed for %s and %p{i,I}{4,6}, set buffer to empty string or address 0. Documented this behavior in uapi header (Andrii)
- fix a few error handling issues for bpftool (Andrii)
- A few other minor fixes and cosmetic changes.

Looks great overall. Applied.
But few follow ups are necessary.

Quoted 6 lines

The main gotcha is that new tests need llvm with the fix
https://urldefense.proofpoint.com/v2/url?u=https-3A__reviews.llvm.org_D78466&d=DwIBAg&c=5VD0RTtNlTh3ycd41b3MUw&r=DA8e1B5r073vIqRrFz7MRA&m=bCI_s36swhyaNaGGr3YQ0eQH9Lc-hpSGQy3u9fCRL9A&s=FLMd73HYbmZPnZrtCE9ntzg9e9eHOCioQbrt1w4sgWU&e= . I think it was applied to llvm 10 branch already,
but please add selftests/bpf/README.rst and mention
that above llvm commit is necessary to successfully pass the tests.
Also mention the verifier error that folks will see when llvm is buggy.

Okay. Will have a followup patch for this.

Quoted 3 lines

Few other nits I noticed in relevant patches.