[PATCH v3 bpf-next 00/11] bpf: Enable bpf_skc_to_* sock casting helper to networking prog type

[PATCH v3 bpf-next 00/11] bpf: Enable bpf_skc_to_* sock casting helper to networking prog type

[Martin KaFai Lau]

This set allows networking prog type to directly read fields from
the in-kernel socket type, e.g. "struct tcp_sock".

Patch 2 has the details on the use case.

v3:
- ARG_PTR_TO_SOCK_COMMON_OR_NULL was attempted in v2.  The _OR_NULL was
  needed because the PTR_TO_BTF_ID could be NULL but note that a could be NULL
  PTR_TO_BTF_ID is not a scalar NULL to the verifier.  "_OR_NULL" implicitly
  gives an expectation that the helper can take a scalar NULL which does
  not make sense in most (except one) helpers.  Passing scalar NULL
  should be rejected at the verification time.

  Thus, this patch uses ARG_PTR_TO_BTF_ID_SOCK_COMMON to specify that the
  helper can take both the btf-id ptr or the legacy PTR_TO_SOCK_COMMON but
  not scalar NULL.  It requires the func_proto to explicitly specify the
  arg_btf_id such that there is a very clear expectation that the helper
  can handle a NULL PTR_TO_BTF_ID.

v2:
- Add ARG_PTR_TO_SOCK_COMMON_OR_NULL (Lorenz)

Martin KaFai Lau (11):
  bpf: Move the PTR_TO_BTF_ID check to check_reg_type()
  bpf: Enable bpf_skc_to_* sock casting helper to networking prog type
  bpf: Change bpf_sk_release and bpf_sk_*cgroup_id to accept
    ARG_PTR_TO_BTF_ID_SOCK_COMMON
  bpf: Change bpf_sk_storage_*() to accept ARG_PTR_TO_BTF_ID_SOCK_COMMON
  bpf: Change bpf_tcp_*_syncookie to accept
    ARG_PTR_TO_BTF_ID_SOCK_COMMON
  bpf: Change bpf_sk_assign to accept ARG_PTR_TO_BTF_ID_SOCK_COMMON
  bpf: selftest: Add ref_tracking verifier test for bpf_skc casting
  bpf: selftest: Move sock_fields test into test_progs
  bpf: selftest: Adapt sock_fields test to use skel and global variables
  bpf: selftest: Use network_helpers in the sock_fields test
  bpf: selftest: Use bpf_skc_to_tcp_sock() in the sock_fields test

 include/linux/bpf.h                           |   1 +
 include/net/bpf_sk_storage.h                  |   2 -
 include/uapi/linux/bpf.h                      |   1 +
 kernel/bpf/bpf_lsm.c                          |   4 +-
 kernel/bpf/verifier.c                         |  93 ++--
 net/core/bpf_sk_storage.c                     |  31 +-
 net/core/filter.c                             | 112 ++--
 net/ipv4/bpf_tcp_ca.c                         |  23 +-
 tools/include/uapi/linux/bpf.h                |   1 +
 tools/testing/selftests/bpf/Makefile          |   2 +-
 .../selftests/bpf/prog_tests/sock_fields.c    | 361 +++++++++++++
 ..._sock_fields_kern.c => test_sock_fields.c} | 165 +++---
 .../testing/selftests/bpf/test_sock_fields.c  | 482 ------------------
 .../selftests/bpf/verifier/ref_tracking.c     |  47 ++
 14 files changed, 633 insertions(+), 692 deletions(-)
 create mode 100644 tools/testing/selftests/bpf/prog_tests/sock_fields.c
 rename tools/testing/selftests/bpf/progs/{test_sock_fields_kern.c => test_sock_fields.c} (61%)
 delete mode 100644 tools/testing/selftests/bpf/test_sock_fields.c

-- 
2.24.1

[PATCH v3 bpf-next 01/11] bpf: Move the PTR_TO_BTF_ID check to check_reg_type()

[Martin KaFai Lau]

check_reg_type() checks whether a reg can be used as an arg of a
func_proto.  For PTR_TO_BTF_ID, the check is actually not
completely done until the reg->btf_id is pointing to a
kernel struct that is acceptable by the func_proto.

Thus, this patch moves the btf_id check into check_reg_type().
The compatible_reg_types[] usage is localized in check_reg_type() now.

The "if (!btf_id) verbose(...); " is also removed since it won't happen.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
---
 kernel/bpf/verifier.c | 65 +++++++++++++++++++++++--------------------
 1 file changed, 35 insertions(+), 30 deletions(-)

diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 15ab889b0a3f..3ce61c412ea0 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -4028,20 +4028,29 @@ static const struct bpf_reg_types *compatible_reg_types[] = {
 	[__BPF_ARG_TYPE_MAX]		= NULL,
 };
 
-static int check_reg_type(struct bpf_verifier_env *env, u32 regno,
-			  const struct bpf_reg_types *compatible)
+static int check_reg_type(struct bpf_verifier_env *env, u32 arg,
+			  enum bpf_arg_type arg_type,
+			  const struct bpf_func_proto *fn)
 {
+	u32 regno = BPF_REG_1 + arg;
 	struct bpf_reg_state *regs = cur_regs(env), *reg = &regs[regno];
 	enum bpf_reg_type expected, type = reg->type;
+	const struct bpf_reg_types *compatible;
 	int i, j;
 
+	compatible = compatible_reg_types[arg_type];
+	if (!compatible) {
+		verbose(env, "verifier internal error: unsupported arg type %d\n", arg_type);
+		return -EFAULT;
+	}
+
 	for (i = 0; i < ARRAY_SIZE(compatible->types); i++) {
 		expected = compatible->types[i];
 		if (expected == NOT_INIT)
 			break;
 
 		if (type == expected)
-			return 0;
+			goto found;
 	}
 
 	verbose(env, "R%d type=%s expected=", regno, reg_type_str[type]);
@@ -4049,6 +4058,27 @@ static int check_reg_type(struct bpf_verifier_env *env, u32 regno,
 		verbose(env, "%s, ", reg_type_str[compatible->types[j]]);
 	verbose(env, "%s\n", reg_type_str[compatible->types[j]]);
 	return -EACCES;
+
+found:
+	if (type == PTR_TO_BTF_ID) {
+		u32 *expected_btf_id = fn->arg_btf_id[arg];
+
+		if (!btf_struct_ids_match(&env->log, reg->off, reg->btf_id,
+					  *expected_btf_id)) {
+			verbose(env, "R%d is of type %s but %s is expected\n",
+				regno, kernel_type_name(reg->btf_id),
+				kernel_type_name(*expected_btf_id));
+			return -EACCES;
+		}
+
+		if (!tnum_is_const(reg->var_off) || reg->var_off.value) {
+			verbose(env, "R%d is a pointer to in-kernel struct with non-zero offset\n",
+				regno);
+			return -EACCES;
+		}
+	}
+
+	return 0;
 }
 
 static int check_func_arg(struct bpf_verifier_env *env, u32 arg,
@@ -4058,7 +4088,6 @@ static int check_func_arg(struct bpf_verifier_env *env, u32 arg,
 	u32 regno = BPF_REG_1 + arg;
 	struct bpf_reg_state *regs = cur_regs(env), *reg = &regs[regno];
 	enum bpf_arg_type arg_type = fn->arg_type[arg];
-	const struct bpf_reg_types *compatible;
 	enum bpf_reg_type type = reg->type;
 	int err = 0;
 
@@ -4098,35 +4127,11 @@ static int check_func_arg(struct bpf_verifier_env *env, u32 arg,
 		 */
 		goto skip_type_check;
 
-	compatible = compatible_reg_types[arg_type];
-	if (!compatible) {
-		verbose(env, "verifier internal error: unsupported arg type %d\n", arg_type);
-		return -EFAULT;
-	}
-
-	err = check_reg_type(env, regno, compatible);
+	err = check_reg_type(env, arg, arg_type, fn);
 	if (err)
 		return err;
 
-	if (type == PTR_TO_BTF_ID) {
-		const u32 *btf_id = fn->arg_btf_id[arg];
-
-		if (!btf_id) {
-			verbose(env, "verifier internal error: missing BTF ID\n");
-			return -EFAULT;
-		}
-
-		if (!btf_struct_ids_match(&env->log, reg->off, reg->btf_id, *btf_id)) {
-			verbose(env, "R%d is of type %s but %s is expected\n",
-				regno, kernel_type_name(reg->btf_id), kernel_type_name(*btf_id));
-			return -EACCES;
-		}
-		if (!tnum_is_const(reg->var_off) || reg->var_off.value) {
-			verbose(env, "R%d is a pointer to in-kernel struct with non-zero offset\n",
-				regno);
-			return -EACCES;
-		}
-	} else if (type == PTR_TO_CTX) {
+	if (type == PTR_TO_CTX) {
 		err = check_ctx_reg(env, reg, regno);
 		if (err < 0)
 			return err;
-- 
2.24.1

Re: [PATCH v3 bpf-next 01/11] bpf: Move the PTR_TO_BTF_ID check to check_reg_type()

[Lorenz Bauer]

On Tue, 22 Sep 2020 at 08:04, Martin KaFai Lau <kafai@fb.com> wrote:
>
> check_reg_type() checks whether a reg can be used as an arg of a
> func_proto.  For PTR_TO_BTF_ID, the check is actually not
> completely done until the reg->btf_id is pointing to a
> kernel struct that is acceptable by the func_proto.
>
> Thus, this patch moves the btf_id check into check_reg_type().
> The compatible_reg_types[] usage is localized in check_reg_type() now.
>
> The "if (!btf_id) verbose(...); " is also removed since it won't happen.
>
> Signed-off-by: Martin KaFai Lau <kafai@fb.com>
> ---
>  kernel/bpf/verifier.c | 65 +++++++++++++++++++++++--------------------
>  1 file changed, 35 insertions(+), 30 deletions(-)
>
> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index 15ab889b0a3f..3ce61c412ea0 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c
> @@ -4028,20 +4028,29 @@ static const struct bpf_reg_types *compatible_reg_types[] = {
>         [__BPF_ARG_TYPE_MAX]            = NULL,
>  };
>
> -static int check_reg_type(struct bpf_verifier_env *env, u32 regno,
> -                         const struct bpf_reg_types *compatible)
> +static int check_reg_type(struct bpf_verifier_env *env, u32 arg,
> +                         enum bpf_arg_type arg_type,
> +                         const struct bpf_func_proto *fn)

How about (env, regno, arg_type, expected_btf_id) instead? Otherwise
implementing sockmap update from iter with your current approach is
difficult for me. See resolve_map_arg_type, which now needs to resolve
expected_bpf_id.

See below for what I mean:

diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 2468533bc4a1..3a238a295c37 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -3931,7 +3931,8 @@ static int int_ptr_type_to_size(enum bpf_arg_type type)

 static int resolve_map_arg_type(struct bpf_verifier_env *env,
                  const struct bpf_call_arg_meta *meta,
-                 enum bpf_arg_type *arg_type)
+                 enum bpf_arg_type *arg_type,
+                 u32 **expected_btf_id)
 {
     if (!meta->map_ptr) {
         /* kernel subsystem misconfigured verifier */
@@ -3943,7 +3944,8 @@ static int resolve_map_arg_type(struct
bpf_verifier_env *env,
     case BPF_MAP_TYPE_SOCKMAP:
     case BPF_MAP_TYPE_SOCKHASH:
         if (*arg_type == ARG_PTR_TO_MAP_VALUE) {
-            *arg_type = ARG_PTR_TO_SOCKET;
+            *arg_type = ARG_PTR_TO_BTF_ID_SOCK_COMMON;
+            *expected_btf_id = &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON];
         } else {
             verbose(env, "invalid arg_type for sockmap/sockhash\n");
             return -EINVAL;
@@ -4044,11 +4046,9 @@ static const struct bpf_reg_types
*compatible_reg_types[] = {
     [__BPF_ARG_TYPE_MAX]        = NULL,
 };

-static int check_reg_type(struct bpf_verifier_env *env, u32 arg,
-              enum bpf_arg_type arg_type,
-              const struct bpf_func_proto *fn)
+static int check_reg_type(struct bpf_verifier_env *env, u32 regno,
+              enum bpf_arg_type arg_type, u32 *expected_btf_id)
 {
-    u32 regno = BPF_REG_1 + arg;
     struct bpf_reg_state *regs = cur_regs(env), *reg = &regs[regno];
     enum bpf_reg_type expected, type = reg->type;
     const struct bpf_reg_types *compatible;
@@ -4077,8 +4077,6 @@ static int check_reg_type(struct
bpf_verifier_env *env, u32 arg,

 found:
     if (type == PTR_TO_BTF_ID) {
-        u32 *expected_btf_id = fn->arg_btf_id[arg];
-
         if (!btf_struct_ids_match(&env->log, reg->off, reg->btf_id,
                       *expected_btf_id)) {
             verbose(env, "R%d is of type %s but %s is expected\n",
@@ -4105,6 +4103,7 @@ static int check_func_arg(struct
bpf_verifier_env *env, u32 arg,
     struct bpf_reg_state *regs = cur_regs(env), *reg = &regs[regno];
     enum bpf_arg_type arg_type = fn->arg_type[arg];
     enum bpf_reg_type type = reg->type;
+    u32 *expected_btf_id = fn->arg_btf_id[arg];
     int err = 0;

     if (arg_type == ARG_DONTCARE)
@@ -4132,7 +4131,7 @@ static int check_func_arg(struct
bpf_verifier_env *env, u32 arg,
     if (arg_type == ARG_PTR_TO_MAP_VALUE ||
         arg_type == ARG_PTR_TO_UNINIT_MAP_VALUE ||
         arg_type == ARG_PTR_TO_MAP_VALUE_OR_NULL) {
-        err = resolve_map_arg_type(env, meta, &arg_type);
+        err = resolve_map_arg_type(env, meta, &arg_type, &expected_btf_id);
         if (err)
             return err;
     }
@@ -4143,7 +4142,7 @@ static int check_func_arg(struct
bpf_verifier_env *env, u32 arg,
          */
         goto skip_type_check;

-    err = check_reg_type(env, arg, arg_type, fn);
+    err = check_reg_type(env, regno, arg_type, expected_btf_id);
     if (err)
         return err;

-- 
2.25.1



--
Lorenz Bauer  |  Systems Engineer
6th Floor, County Hall/The Riverside Building, SE1 7PB, UK

www.cloudflare.com

Re: [PATCH v3 bpf-next 01/11] bpf: Move the PTR_TO_BTF_ID check to check_reg_type()

[Martin KaFai Lau]

On Tue, Sep 22, 2020 at 10:56:55AM +0100, Lorenz Bauer wrote:
> On Tue, 22 Sep 2020 at 08:04, Martin KaFai Lau <kafai@fb.com> wrote:
> >
> > check_reg_type() checks whether a reg can be used as an arg of a
> > func_proto.  For PTR_TO_BTF_ID, the check is actually not
> > completely done until the reg->btf_id is pointing to a
> > kernel struct that is acceptable by the func_proto.
> >
> > Thus, this patch moves the btf_id check into check_reg_type().
> > The compatible_reg_types[] usage is localized in check_reg_type() now.
> >
> > The "if (!btf_id) verbose(...); " is also removed since it won't happen.
> >
> > Signed-off-by: Martin KaFai Lau <kafai@fb.com>
> > ---
> >  kernel/bpf/verifier.c | 65 +++++++++++++++++++++++--------------------
> >  1 file changed, 35 insertions(+), 30 deletions(-)
> >
> > diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> > index 15ab889b0a3f..3ce61c412ea0 100644
> > --- a/kernel/bpf/verifier.c
> > +++ b/kernel/bpf/verifier.c
> > @@ -4028,20 +4028,29 @@ static const struct bpf_reg_types *compatible_reg_types[] = {
> >         [__BPF_ARG_TYPE_MAX]            = NULL,
> >  };
> >
> > -static int check_reg_type(struct bpf_verifier_env *env, u32 regno,
> > -                         const struct bpf_reg_types *compatible)
> > +static int check_reg_type(struct bpf_verifier_env *env, u32 arg,
> > +                         enum bpf_arg_type arg_type,
> > +                         const struct bpf_func_proto *fn)
Yes. I think that works as good.

An idea for the mid term, I think this map helper's arg override logic
should belong to a new map_ops and this new map_ops can return the whole
"fn" instead of overriding on an arg-by-arg base.

Re: [PATCH v3 bpf-next 01/11] bpf: Move the PTR_TO_BTF_ID check to check_reg_type()

[Lorenz Bauer]

On Tue, 22 Sep 2020 at 19:38, Martin KaFai Lau <kafai@fb.com> wrote:
>
> On Tue, Sep 22, 2020 at 10:56:55AM +0100, Lorenz Bauer wrote:
> > On Tue, 22 Sep 2020 at 08:04, Martin KaFai Lau <kafai@fb.com> wrote:
> > >
> > > check_reg_type() checks whether a reg can be used as an arg of a
> > > func_proto.  For PTR_TO_BTF_ID, the check is actually not
> > > completely done until the reg->btf_id is pointing to a
> > > kernel struct that is acceptable by the func_proto.
> > >
> > > Thus, this patch moves the btf_id check into check_reg_type().
> > > The compatible_reg_types[] usage is localized in check_reg_type() now.
> > >
> > > The "if (!btf_id) verbose(...); " is also removed since it won't happen.
> > >
> > > Signed-off-by: Martin KaFai Lau <kafai@fb.com>
> > > ---
> > >  kernel/bpf/verifier.c | 65 +++++++++++++++++++++++--------------------
> > >  1 file changed, 35 insertions(+), 30 deletions(-)
> > >
> > > diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> > > index 15ab889b0a3f..3ce61c412ea0 100644
> > > --- a/kernel/bpf/verifier.c
> > > +++ b/kernel/bpf/verifier.c
> > > @@ -4028,20 +4028,29 @@ static const struct bpf_reg_types *compatible_reg_types[] = {
> > >         [__BPF_ARG_TYPE_MAX]            = NULL,
> > >  };
> > >
> > > -static int check_reg_type(struct bpf_verifier_env *env, u32 regno,
> > > -                         const struct bpf_reg_types *compatible)
> > > +static int check_reg_type(struct bpf_verifier_env *env, u32 arg,
> > > +                         enum bpf_arg_type arg_type,
> > > +                         const struct bpf_func_proto *fn)
> Yes. I think that works as good.
>
> An idea for the mid term, I think this map helper's arg override logic
> should belong to a new map_ops and this new map_ops can return the whole
> "fn" instead of overriding on an arg-by-arg base.

Yeah, agreed. I've had a similar idea, but no time to implement it yet.

-- 
Lorenz Bauer  |  Systems Engineer
6th Floor, County Hall/The Riverside Building, SE1 7PB, UK

www.cloudflare.com

Re: [PATCH v3 bpf-next 01/11] bpf: Move the PTR_TO_BTF_ID check to check_reg_type()

[Martin KaFai Lau]

On Tue, Sep 22, 2020 at 10:56:55AM +0100, Lorenz Bauer wrote:
> On Tue, 22 Sep 2020 at 08:04, Martin KaFai Lau <kafai@fb.com> wrote:
> >
> > check_reg_type() checks whether a reg can be used as an arg of a
> > func_proto.  For PTR_TO_BTF_ID, the check is actually not
> > completely done until the reg->btf_id is pointing to a
> > kernel struct that is acceptable by the func_proto.
> >
> > Thus, this patch moves the btf_id check into check_reg_type().
> > The compatible_reg_types[] usage is localized in check_reg_type() now.
> >
> > The "if (!btf_id) verbose(...); " is also removed since it won't happen.
> >
> > Signed-off-by: Martin KaFai Lau <kafai@fb.com>
> > ---
> >  kernel/bpf/verifier.c | 65 +++++++++++++++++++++++--------------------
> >  1 file changed, 35 insertions(+), 30 deletions(-)
> >
> > diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> > index 15ab889b0a3f..3ce61c412ea0 100644
> > --- a/kernel/bpf/verifier.c
> > +++ b/kernel/bpf/verifier.c
> > @@ -4028,20 +4028,29 @@ static const struct bpf_reg_types *compatible_reg_types[] = {
> >         [__BPF_ARG_TYPE_MAX]            = NULL,
> >  };
> >
> > -static int check_reg_type(struct bpf_verifier_env *env, u32 regno,
> > -                         const struct bpf_reg_types *compatible)
> > +static int check_reg_type(struct bpf_verifier_env *env, u32 arg,
> > +                         enum bpf_arg_type arg_type,
> > +                         const struct bpf_func_proto *fn)
> 
> How about (env, regno, arg_type, expected_btf_id) instead? Otherwise
Last reply cut off in the wrong line.  I will make this change in v4.

[PATCH v3 bpf-next 02/11] bpf: Enable bpf_skc_to_* sock casting helper to networking prog type

[Martin KaFai Lau]

There is a constant need to add more fields into the bpf_tcp_sock
for the bpf programs running at tc, sock_ops...etc.

A current workaround could be to use bpf_probe_read_kernel().  However,
other than making another helper call for reading each field and missing
CO-RE, it is also not as intuitive to use as directly reading
"tp->lsndtime" for example.  While already having perfmon cap to do
bpf_probe_read_kernel(), it will be much easier if the bpf prog can
directly read from the tcp_sock.

This patch tries to do that by using the existing casting-helpers
bpf_skc_to_*() whose func_proto returns a btf_id.  For example, the
func_proto of bpf_skc_to_tcp_sock returns the btf_id of the
kernel "struct tcp_sock".

These helpers are also added to is_ptr_cast_function().
It ensures the returning reg (BPF_REF_0) will also carries the ref_obj_id.
That will keep the ref-tracking works properly.

The bpf_skc_to_* helpers are made available to most of the bpf prog
types in filter.c. They are limited by perfmon cap.

This patch adds a ARG_PTR_TO_BTF_ID_SOCK_COMMON.  The helper accepting
this arg can accept a btf-id-ptr (PTR_TO_BTF_ID + &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON])
or a legacy-ctx-convert-skc-ptr (PTR_TO_SOCK_COMMON).  The bpf_skc_to_*()
helpers are changed to take ARG_PTR_TO_BTF_ID_SOCK_COMMON such that
they will accept pointer obtained from skb->sk.

PTR_TO_*_OR_NULL is not accepted as an ARG_PTR_TO_BTF_ID_SOCK_COMMON
at verification time.  All PTR_TO_*_OR_NULL reg has to do a NULL check
first before passing into the helper or else the bpf prog will be
rejected by the verifier.

[ ARG_PTR_TO_SOCK_COMMON_OR_NULL was attempted earlier.  The _OR_NULL was
  needed because the PTR_TO_BTF_ID could be NULL but note that a could be NULL
  PTR_TO_BTF_ID is not a scalar NULL to the verifier.  "_OR_NULL" implicitly
  gives an expectation that the helper can take a scalar NULL which does
  not make sense in most (except one) helpers.  Passing scalar NULL
  should be rejected at the verification time.

  Thus, this patch uses ARG_PTR_TO_BTF_ID_SOCK_COMMON to specify that the
  helper can take both the btf-id ptr or the legacy PTR_TO_SOCK_COMMON but
  not scalar NULL.  It requires the func_proto to explicitly specify the
  arg_btf_id such that there is a very clear expectation that the helper
  can handle a NULL PTR_TO_BTF_ID. ]

[ ARG_PTR_TO_BTF_ID_SOCK_COMMON will be used to replace
  ARG_PTR_TO_SOCK* of other existing helpers later such that
  those existing helpers can take the PTR_TO_BTF_ID returned by
  the bpf_skc_to_*() helpers.

  The only special case is bpf_sk_lookup_assign() which can accept a
  scalar NULL ptr.  It has to be handled specially in another follow
  up patch (e.g. by renaming ARG_PTR_TO_SOCKET_OR_NULL to
  ARG_PTR_TO_BTF_ID_SOCK_COMMON_OR_NULL). ]

[ When converting the older helpers that take ARG_PTR_TO_SOCK* in
  the later patch, if the kernel does not support BTF,
  ARG_PTR_TO_BTF_ID_SOCK_COMMON will behave like ARG_PTR_TO_SOCK_COMMON
  because no reg->type could have PTR_TO_BTF_ID in this case.

  It is not a concern for the newer-btf-only helper like the bpf_skc_to_*()
  here though because these helpers must require BTF vmlinux to begin
  with. ]

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
---
 include/linux/bpf.h   |  1 +
 kernel/bpf/verifier.c | 28 ++++++++++++++++---
 net/core/filter.c     | 64 +++++++++++++++++++++++++++++++++----------
 3 files changed, 74 insertions(+), 19 deletions(-)

diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index fc5c901c7542..d0937f1d2980 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -292,6 +292,7 @@ enum bpf_arg_type {
 	ARG_PTR_TO_ALLOC_MEM,	/* pointer to dynamically allocated memory */
 	ARG_PTR_TO_ALLOC_MEM_OR_NULL,	/* pointer to dynamically allocated memory or NULL */
 	ARG_CONST_ALLOC_SIZE_OR_ZERO,	/* number of allocated bytes requested */
+	ARG_PTR_TO_BTF_ID_SOCK_COMMON,	/* pointer to in-kernel sock_common or bpf-mirrored bpf_sock */
 	__BPF_ARG_TYPE_MAX,
 };
 
diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 3ce61c412ea0..2468533bc4a1 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -486,7 +486,12 @@ static bool is_acquire_function(enum bpf_func_id func_id,
 static bool is_ptr_cast_function(enum bpf_func_id func_id)
 {
 	return func_id == BPF_FUNC_tcp_sock ||
-		func_id == BPF_FUNC_sk_fullsock;
+		func_id == BPF_FUNC_sk_fullsock ||
+		func_id == BPF_FUNC_skc_to_tcp_sock ||
+		func_id == BPF_FUNC_skc_to_tcp6_sock ||
+		func_id == BPF_FUNC_skc_to_udp6_sock ||
+		func_id == BPF_FUNC_skc_to_tcp_timewait_sock ||
+		func_id == BPF_FUNC_skc_to_tcp_request_sock;
 }
 
 /* string representation of 'enum bpf_reg_type' */
@@ -3973,6 +3978,16 @@ static const struct bpf_reg_types sock_types = {
 	},
 };
 
+static const struct bpf_reg_types btf_id_sock_common_types = {
+	.types = {
+		PTR_TO_SOCK_COMMON,
+		PTR_TO_SOCKET,
+		PTR_TO_TCP_SOCK,
+		PTR_TO_XDP_SOCK,
+		PTR_TO_BTF_ID,
+	},
+};
+
 static const struct bpf_reg_types mem_types = {
 	.types = {
 		PTR_TO_STACK,
@@ -4014,6 +4029,7 @@ static const struct bpf_reg_types *compatible_reg_types[] = {
 	[ARG_PTR_TO_CTX]		= &context_types,
 	[ARG_PTR_TO_CTX_OR_NULL]	= &context_types,
 	[ARG_PTR_TO_SOCK_COMMON]	= &sock_types,
+	[ARG_PTR_TO_BTF_ID_SOCK_COMMON]	= &btf_id_sock_common_types,
 	[ARG_PTR_TO_SOCKET]		= &fullsock_types,
 	[ARG_PTR_TO_SOCKET_OR_NULL]	= &fullsock_types,
 	[ARG_PTR_TO_BTF_ID]		= &btf_ptr_types,
@@ -4579,9 +4595,13 @@ static bool check_btf_id_ok(const struct bpf_func_proto *fn)
 {
 	int i;
 
-	for (i = 0; i < ARRAY_SIZE(fn->arg_type); i++)
-		if (fn->arg_type[i] == ARG_PTR_TO_BTF_ID && !fn->arg_btf_id[i])
-			return false;
+	for (i = 0; i < ARRAY_SIZE(fn->arg_type); i++) {
+		if (fn->arg_type[i] == ARG_PTR_TO_BTF_ID ||
+		    fn->arg_type[i] == ARG_PTR_TO_BTF_ID_SOCK_COMMON) {
+			if (!fn->arg_btf_id[i])
+				return false;
+		}
+	}
 
 	return true;
 }
diff --git a/net/core/filter.c b/net/core/filter.c
index 6014e5f40c58..54b338de4bb8 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -77,6 +77,9 @@
 #include <net/transp_v6.h>
 #include <linux/btf_ids.h>
 
+static const struct bpf_func_proto *
+bpf_sk_base_func_proto(enum bpf_func_id func_id);
+
 int copy_bpf_fprog_from_user(struct sock_fprog *dst, sockptr_t src, int len)
 {
 	if (in_compat_syscall()) {
@@ -6619,7 +6622,7 @@ sock_addr_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 			return NULL;
 		}
 	default:
-		return bpf_base_func_proto(func_id);
+		return bpf_sk_base_func_proto(func_id);
 	}
 }
 
@@ -6638,7 +6641,7 @@ sk_filter_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 	case BPF_FUNC_perf_event_output:
 		return &bpf_skb_event_output_proto;
 	default:
-		return bpf_base_func_proto(func_id);
+		return bpf_sk_base_func_proto(func_id);
 	}
 }
 
@@ -6799,7 +6802,7 @@ tc_cls_act_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 		return &bpf_sk_assign_proto;
 #endif
 	default:
-		return bpf_base_func_proto(func_id);
+		return bpf_sk_base_func_proto(func_id);
 	}
 }
 
@@ -6840,7 +6843,7 @@ xdp_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 		return &bpf_tcp_gen_syncookie_proto;
 #endif
 	default:
-		return bpf_base_func_proto(func_id);
+		return bpf_sk_base_func_proto(func_id);
 	}
 }
 
@@ -6882,7 +6885,7 @@ sock_ops_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 		return &bpf_tcp_sock_proto;
 #endif /* CONFIG_INET */
 	default:
-		return bpf_base_func_proto(func_id);
+		return bpf_sk_base_func_proto(func_id);
 	}
 }
 
@@ -6928,7 +6931,7 @@ sk_msg_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 		return &bpf_get_cgroup_classid_curr_proto;
 #endif
 	default:
-		return bpf_base_func_proto(func_id);
+		return bpf_sk_base_func_proto(func_id);
 	}
 }
 
@@ -6970,7 +6973,7 @@ sk_skb_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 		return &bpf_skc_lookup_tcp_proto;
 #endif
 	default:
-		return bpf_base_func_proto(func_id);
+		return bpf_sk_base_func_proto(func_id);
 	}
 }
 
@@ -6981,7 +6984,7 @@ flow_dissector_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 	case BPF_FUNC_skb_load_bytes:
 		return &bpf_flow_dissector_load_bytes_proto;
 	default:
-		return bpf_base_func_proto(func_id);
+		return bpf_sk_base_func_proto(func_id);
 	}
 }
 
@@ -7008,7 +7011,7 @@ lwt_out_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 	case BPF_FUNC_skb_under_cgroup:
 		return &bpf_skb_under_cgroup_proto;
 	default:
-		return bpf_base_func_proto(func_id);
+		return bpf_sk_base_func_proto(func_id);
 	}
 }
 
@@ -9745,7 +9748,7 @@ sk_lookup_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 	case BPF_FUNC_sk_release:
 		return &bpf_sk_release_proto;
 	default:
-		return bpf_base_func_proto(func_id);
+		return bpf_sk_base_func_proto(func_id);
 	}
 }
 
@@ -9912,7 +9915,7 @@ const struct bpf_func_proto bpf_skc_to_tcp6_sock_proto = {
 	.func			= bpf_skc_to_tcp6_sock,
 	.gpl_only		= false,
 	.ret_type		= RET_PTR_TO_BTF_ID_OR_NULL,
-	.arg1_type		= ARG_PTR_TO_BTF_ID,
+	.arg1_type		= ARG_PTR_TO_BTF_ID_SOCK_COMMON,
 	.arg1_btf_id		= &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON],
 	.ret_btf_id		= &btf_sock_ids[BTF_SOCK_TYPE_TCP6],
 };
@@ -9929,7 +9932,7 @@ const struct bpf_func_proto bpf_skc_to_tcp_sock_proto = {
 	.func			= bpf_skc_to_tcp_sock,
 	.gpl_only		= false,
 	.ret_type		= RET_PTR_TO_BTF_ID_OR_NULL,
-	.arg1_type		= ARG_PTR_TO_BTF_ID,
+	.arg1_type		= ARG_PTR_TO_BTF_ID_SOCK_COMMON,
 	.arg1_btf_id		= &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON],
 	.ret_btf_id		= &btf_sock_ids[BTF_SOCK_TYPE_TCP],
 };
@@ -9953,7 +9956,7 @@ const struct bpf_func_proto bpf_skc_to_tcp_timewait_sock_proto = {
 	.func			= bpf_skc_to_tcp_timewait_sock,
 	.gpl_only		= false,
 	.ret_type		= RET_PTR_TO_BTF_ID_OR_NULL,
-	.arg1_type		= ARG_PTR_TO_BTF_ID,
+	.arg1_type		= ARG_PTR_TO_BTF_ID_SOCK_COMMON,
 	.arg1_btf_id		= &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON],
 	.ret_btf_id		= &btf_sock_ids[BTF_SOCK_TYPE_TCP_TW],
 };
@@ -9977,7 +9980,7 @@ const struct bpf_func_proto bpf_skc_to_tcp_request_sock_proto = {
 	.func			= bpf_skc_to_tcp_request_sock,
 	.gpl_only		= false,
 	.ret_type		= RET_PTR_TO_BTF_ID_OR_NULL,
-	.arg1_type		= ARG_PTR_TO_BTF_ID,
+	.arg1_type		= ARG_PTR_TO_BTF_ID_SOCK_COMMON,
 	.arg1_btf_id		= &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON],
 	.ret_btf_id		= &btf_sock_ids[BTF_SOCK_TYPE_TCP_REQ],
 };
@@ -9999,7 +10002,38 @@ const struct bpf_func_proto bpf_skc_to_udp6_sock_proto = {
 	.func			= bpf_skc_to_udp6_sock,
 	.gpl_only		= false,
 	.ret_type		= RET_PTR_TO_BTF_ID_OR_NULL,
-	.arg1_type		= ARG_PTR_TO_BTF_ID,
+	.arg1_type		= ARG_PTR_TO_BTF_ID_SOCK_COMMON,
 	.arg1_btf_id		= &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON],
 	.ret_btf_id		= &btf_sock_ids[BTF_SOCK_TYPE_UDP6],
 };
+
+static const struct bpf_func_proto *
+bpf_sk_base_func_proto(enum bpf_func_id func_id)
+{
+	const struct bpf_func_proto *func;
+
+	switch (func_id) {
+	case BPF_FUNC_skc_to_tcp6_sock:
+		func = &bpf_skc_to_tcp6_sock_proto;
+		break;
+	case BPF_FUNC_skc_to_tcp_sock:
+		func = &bpf_skc_to_tcp_sock_proto;
+		break;
+	case BPF_FUNC_skc_to_tcp_timewait_sock:
+		func = &bpf_skc_to_tcp_timewait_sock_proto;
+		break;
+	case BPF_FUNC_skc_to_tcp_request_sock:
+		func = &bpf_skc_to_tcp_request_sock_proto;
+		break;
+	case BPF_FUNC_skc_to_udp6_sock:
+		func = &bpf_skc_to_udp6_sock_proto;
+		break;
+	default:
+		return bpf_base_func_proto(func_id);
+	}
+
+	if (!perfmon_capable())
+		return NULL;
+
+	return func;
+}
-- 
2.24.1

Re: [PATCH v3 bpf-next 02/11] bpf: Enable bpf_skc_to_* sock casting helper to networking prog type

[Lorenz Bauer]

On Tue, 22 Sep 2020 at 08:04, Martin KaFai Lau <kafai@fb.com> wrote:
>
> There is a constant need to add more fields into the bpf_tcp_sock
> for the bpf programs running at tc, sock_ops...etc.
>
> A current workaround could be to use bpf_probe_read_kernel().  However,
> other than making another helper call for reading each field and missing
> CO-RE, it is also not as intuitive to use as directly reading
> "tp->lsndtime" for example.  While already having perfmon cap to do
> bpf_probe_read_kernel(), it will be much easier if the bpf prog can
> directly read from the tcp_sock.
>
> This patch tries to do that by using the existing casting-helpers
> bpf_skc_to_*() whose func_proto returns a btf_id.  For example, the
> func_proto of bpf_skc_to_tcp_sock returns the btf_id of the
> kernel "struct tcp_sock".
>
> These helpers are also added to is_ptr_cast_function().
> It ensures the returning reg (BPF_REF_0) will also carries the ref_obj_id.
> That will keep the ref-tracking works properly.
>
> The bpf_skc_to_* helpers are made available to most of the bpf prog
> types in filter.c. They are limited by perfmon cap.
>
> This patch adds a ARG_PTR_TO_BTF_ID_SOCK_COMMON.  The helper accepting
> this arg can accept a btf-id-ptr (PTR_TO_BTF_ID + &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON])
> or a legacy-ctx-convert-skc-ptr (PTR_TO_SOCK_COMMON).  The bpf_skc_to_*()
> helpers are changed to take ARG_PTR_TO_BTF_ID_SOCK_COMMON such that
> they will accept pointer obtained from skb->sk.
>
> PTR_TO_*_OR_NULL is not accepted as an ARG_PTR_TO_BTF_ID_SOCK_COMMON
> at verification time.  All PTR_TO_*_OR_NULL reg has to do a NULL check
> first before passing into the helper or else the bpf prog will be
> rejected by the verifier.
>
> [ ARG_PTR_TO_SOCK_COMMON_OR_NULL was attempted earlier.  The _OR_NULL was
>   needed because the PTR_TO_BTF_ID could be NULL but note that a could be NULL
>   PTR_TO_BTF_ID is not a scalar NULL to the verifier.  "_OR_NULL" implicitly
>   gives an expectation that the helper can take a scalar NULL which does
>   not make sense in most (except one) helpers.  Passing scalar NULL
>   should be rejected at the verification time.

What is the benefit of requiring a !sk check from the user if all of
the helpers know how to deal with a NULL pointer?

>
>   Thus, this patch uses ARG_PTR_TO_BTF_ID_SOCK_COMMON to specify that the
>   helper can take both the btf-id ptr or the legacy PTR_TO_SOCK_COMMON but
>   not scalar NULL.  It requires the func_proto to explicitly specify the
>   arg_btf_id such that there is a very clear expectation that the helper
>   can handle a NULL PTR_TO_BTF_ID. ]

I think ARG_PTR_TO_BTF_ID_SOCK_COMMON is actually a misnomer, since
nothing enforces that arg_btf_id is actually an ID for sock common.
This is where ARG_PTR_TO_SOCK_COMMON_OR_NULL is much easier to
understand, even though it's more permissive than it has to be. It
communicates very clearly what values the argument can take.

If you're set on ARG_PTR_TO_BTF_ID_SOCK_COMMON I'd suggest forcing the
btf_id in struct bpf_reg_types. This avoids the weird case where the
btf_id doesn't actually point at sock_common, and it also makes my
life easier for sockmap update from iter, as mentioned in the other
email.

--
Lorenz Bauer  |  Systems Engineer
6th Floor, County Hall/The Riverside Building, SE1 7PB, UK

www.cloudflare.com

Re: [PATCH v3 bpf-next 02/11] bpf: Enable bpf_skc_to_* sock casting helper to networking prog type

[Martin KaFai Lau]

On Tue, Sep 22, 2020 at 10:46:41AM +0100, Lorenz Bauer wrote:
> On Tue, 22 Sep 2020 at 08:04, Martin KaFai Lau <kafai@fb.com> wrote:
> >
> > There is a constant need to add more fields into the bpf_tcp_sock
> > for the bpf programs running at tc, sock_ops...etc.
> >
> > A current workaround could be to use bpf_probe_read_kernel().  However,
> > other than making another helper call for reading each field and missing
> > CO-RE, it is also not as intuitive to use as directly reading
> > "tp->lsndtime" for example.  While already having perfmon cap to do
> > bpf_probe_read_kernel(), it will be much easier if the bpf prog can
> > directly read from the tcp_sock.
> >
> > This patch tries to do that by using the existing casting-helpers
> > bpf_skc_to_*() whose func_proto returns a btf_id.  For example, the
> > func_proto of bpf_skc_to_tcp_sock returns the btf_id of the
> > kernel "struct tcp_sock".
> >
> > These helpers are also added to is_ptr_cast_function().
> > It ensures the returning reg (BPF_REF_0) will also carries the ref_obj_id.
> > That will keep the ref-tracking works properly.
> >
> > The bpf_skc_to_* helpers are made available to most of the bpf prog
> > types in filter.c. They are limited by perfmon cap.
> >
> > This patch adds a ARG_PTR_TO_BTF_ID_SOCK_COMMON.  The helper accepting
> > this arg can accept a btf-id-ptr (PTR_TO_BTF_ID + &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON])
> > or a legacy-ctx-convert-skc-ptr (PTR_TO_SOCK_COMMON).  The bpf_skc_to_*()
> > helpers are changed to take ARG_PTR_TO_BTF_ID_SOCK_COMMON such that
> > they will accept pointer obtained from skb->sk.
> >
> > PTR_TO_*_OR_NULL is not accepted as an ARG_PTR_TO_BTF_ID_SOCK_COMMON
> > at verification time.  All PTR_TO_*_OR_NULL reg has to do a NULL check
> > first before passing into the helper or else the bpf prog will be
> > rejected by the verifier.
> >
> > [ ARG_PTR_TO_SOCK_COMMON_OR_NULL was attempted earlier.  The _OR_NULL was
> >   needed because the PTR_TO_BTF_ID could be NULL but note that a could be NULL
> >   PTR_TO_BTF_ID is not a scalar NULL to the verifier.  "_OR_NULL" implicitly
> >   gives an expectation that the helper can take a scalar NULL which does
> >   not make sense in most (except one) helpers.  Passing scalar NULL
> >   should be rejected at the verification time.
> 
> What is the benefit of requiring a !sk check from the user if all of
> the helpers know how to deal with a NULL pointer?
I don't see a reason why the verifier should not reject an incorrect
program at load time if it can.

> 
> >
> >   Thus, this patch uses ARG_PTR_TO_BTF_ID_SOCK_COMMON to specify that the
> >   helper can take both the btf-id ptr or the legacy PTR_TO_SOCK_COMMON but
> >   not scalar NULL.  It requires the func_proto to explicitly specify the
> >   arg_btf_id such that there is a very clear expectation that the helper
> >   can handle a NULL PTR_TO_BTF_ID. ]
> 
> I think ARG_PTR_TO_BTF_ID_SOCK_COMMON is actually a misnomer, since
> nothing enforces that arg_btf_id is actually an ID for sock common.
> This is where ARG_PTR_TO_SOCK_COMMON_OR_NULL is much easier to
> understand, even though it's more permissive than it has to be. It
> communicates very clearly what values the argument can take.
_OR_NULL is incorrect which implies a scalar NULL as mentioned in
this commit message.  From verifier pov, _OR_NULL can take
a scalar NULL.

> 
> If you're set on ARG_PTR_TO_BTF_ID_SOCK_COMMON I'd suggest forcing the
> btf_id in struct bpf_reg_types. This avoids the weird case where the
> btf_id doesn't actually point at sock_common, and it also makes my
I have considered the bpf_reg_types option.  I prefer all
arg info (arg_type and arg_btf_id) stay in the same one
place (i.e. func_proto) as much as possible for now
instead of introducing another place to specify/override it
which then depends on a particular arg_type that some arg_type may be
in func_proto while some may be in other places.

The arg_btf_id can be checked in check_btf_id_ok() if it would be a
big concern that it might slip through the review but I think the
chance is pretty low.

Re: [PATCH v3 bpf-next 02/11] bpf: Enable bpf_skc_to_* sock casting helper to networking prog type

[Lorenz Bauer]

On Tue, 22 Sep 2020 at 19:26, Martin KaFai Lau <kafai@fb.com> wrote:
>
> On Tue, Sep 22, 2020 at 10:46:41AM +0100, Lorenz Bauer wrote:
> > On Tue, 22 Sep 2020 at 08:04, Martin KaFai Lau <kafai@fb.com> wrote:
> > >
> > > There is a constant need to add more fields into the bpf_tcp_sock
> > > for the bpf programs running at tc, sock_ops...etc.
> > >
> > > A current workaround could be to use bpf_probe_read_kernel().  However,
> > > other than making another helper call for reading each field and missing
> > > CO-RE, it is also not as intuitive to use as directly reading
> > > "tp->lsndtime" for example.  While already having perfmon cap to do
> > > bpf_probe_read_kernel(), it will be much easier if the bpf prog can
> > > directly read from the tcp_sock.
> > >
> > > This patch tries to do that by using the existing casting-helpers
> > > bpf_skc_to_*() whose func_proto returns a btf_id.  For example, the
> > > func_proto of bpf_skc_to_tcp_sock returns the btf_id of the
> > > kernel "struct tcp_sock".
> > >
> > > These helpers are also added to is_ptr_cast_function().
> > > It ensures the returning reg (BPF_REF_0) will also carries the ref_obj_id.
> > > That will keep the ref-tracking works properly.
> > >
> > > The bpf_skc_to_* helpers are made available to most of the bpf prog
> > > types in filter.c. They are limited by perfmon cap.
> > >
> > > This patch adds a ARG_PTR_TO_BTF_ID_SOCK_COMMON.  The helper accepting
> > > this arg can accept a btf-id-ptr (PTR_TO_BTF_ID + &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON])
> > > or a legacy-ctx-convert-skc-ptr (PTR_TO_SOCK_COMMON).  The bpf_skc_to_*()
> > > helpers are changed to take ARG_PTR_TO_BTF_ID_SOCK_COMMON such that
> > > they will accept pointer obtained from skb->sk.
> > >
> > > PTR_TO_*_OR_NULL is not accepted as an ARG_PTR_TO_BTF_ID_SOCK_COMMON
> > > at verification time.  All PTR_TO_*_OR_NULL reg has to do a NULL check
> > > first before passing into the helper or else the bpf prog will be
> > > rejected by the verifier.
> > >
> > > [ ARG_PTR_TO_SOCK_COMMON_OR_NULL was attempted earlier.  The _OR_NULL was
> > >   needed because the PTR_TO_BTF_ID could be NULL but note that a could be NULL
> > >   PTR_TO_BTF_ID is not a scalar NULL to the verifier.  "_OR_NULL" implicitly
> > >   gives an expectation that the helper can take a scalar NULL which does
> > >   not make sense in most (except one) helpers.  Passing scalar NULL
> > >   should be rejected at the verification time.
> >
> > What is the benefit of requiring a !sk check from the user if all of
> > the helpers know how to deal with a NULL pointer?
> I don't see a reason why the verifier should not reject an incorrect
> program at load time if it can.
>
> >
> > >
> > >   Thus, this patch uses ARG_PTR_TO_BTF_ID_SOCK_COMMON to specify that the
> > >   helper can take both the btf-id ptr or the legacy PTR_TO_SOCK_COMMON but
> > >   not scalar NULL.  It requires the func_proto to explicitly specify the
> > >   arg_btf_id such that there is a very clear expectation that the helper
> > >   can handle a NULL PTR_TO_BTF_ID. ]
> >
> > I think ARG_PTR_TO_BTF_ID_SOCK_COMMON is actually a misnomer, since
> > nothing enforces that arg_btf_id is actually an ID for sock common.
> > This is where ARG_PTR_TO_SOCK_COMMON_OR_NULL is much easier to
> > understand, even though it's more permissive than it has to be. It
> > communicates very clearly what values the argument can take.
> _OR_NULL is incorrect which implies a scalar NULL as mentioned in
> this commit message.  From verifier pov, _OR_NULL can take
> a scalar NULL.

Yes, I know. I'm saying that the distinction between scalar NULL and
runtime NULL only makes sense after you understand how BTF pointers
are implemented. It only clicked for me after I read the support code
in the JIT that Yonghong pointed out. Should everybody that writes a
helper need to read the JIT? In my opinion we shouldn't. I guess I
don't even care about the verifier rejecting scalar NULL or not, I'd
just like the types to have a name that conveys their NULLness.

>
> >
> > If you're set on ARG_PTR_TO_BTF_ID_SOCK_COMMON I'd suggest forcing the
> > btf_id in struct bpf_reg_types. This avoids the weird case where the
> > btf_id doesn't actually point at sock_common, and it also makes my
> I have considered the bpf_reg_types option.  I prefer all
> arg info (arg_type and arg_btf_id) stay in the same one
> place (i.e. func_proto) as much as possible for now
> instead of introducing another place to specify/override it
> which then depends on a particular arg_type that some arg_type may be
> in func_proto while some may be in other places.

In my opinion that ship sailed when we started aliasing arg_type to
multiple reg_type, but OK.

>
> The arg_btf_id can be checked in check_btf_id_ok() if it would be a
> big concern that it might slip through the review but I think the
> chance is pretty low.

Why increase the burden on human reviewers? Why add code to check an
invariant that we could get rid of in the first place?

-- 
Lorenz Bauer  |  Systems Engineer
6th Floor, County Hall/The Riverside Building, SE1 7PB, UK

www.cloudflare.com

Re: [PATCH v3 bpf-next 02/11] bpf: Enable bpf_skc_to_* sock casting helper to networking prog type

[Martin KaFai Lau]

On Wed, Sep 23, 2020 at 10:27:27AM +0100, Lorenz Bauer wrote:
> On Tue, 22 Sep 2020 at 19:26, Martin KaFai Lau <kafai@fb.com> wrote:
> >
> > On Tue, Sep 22, 2020 at 10:46:41AM +0100, Lorenz Bauer wrote:
> > > On Tue, 22 Sep 2020 at 08:04, Martin KaFai Lau <kafai@fb.com> wrote:
> > > >
> > > > There is a constant need to add more fields into the bpf_tcp_sock
> > > > for the bpf programs running at tc, sock_ops...etc.
> > > >
> > > > A current workaround could be to use bpf_probe_read_kernel().  However,
> > > > other than making another helper call for reading each field and missing
> > > > CO-RE, it is also not as intuitive to use as directly reading
> > > > "tp->lsndtime" for example.  While already having perfmon cap to do
> > > > bpf_probe_read_kernel(), it will be much easier if the bpf prog can
> > > > directly read from the tcp_sock.
> > > >
> > > > This patch tries to do that by using the existing casting-helpers
> > > > bpf_skc_to_*() whose func_proto returns a btf_id.  For example, the
> > > > func_proto of bpf_skc_to_tcp_sock returns the btf_id of the
> > > > kernel "struct tcp_sock".
> > > >
> > > > These helpers are also added to is_ptr_cast_function().
> > > > It ensures the returning reg (BPF_REF_0) will also carries the ref_obj_id.
> > > > That will keep the ref-tracking works properly.
> > > >
> > > > The bpf_skc_to_* helpers are made available to most of the bpf prog
> > > > types in filter.c. They are limited by perfmon cap.
> > > >
> > > > This patch adds a ARG_PTR_TO_BTF_ID_SOCK_COMMON.  The helper accepting
> > > > this arg can accept a btf-id-ptr (PTR_TO_BTF_ID + &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON])
> > > > or a legacy-ctx-convert-skc-ptr (PTR_TO_SOCK_COMMON).  The bpf_skc_to_*()
> > > > helpers are changed to take ARG_PTR_TO_BTF_ID_SOCK_COMMON such that
> > > > they will accept pointer obtained from skb->sk.
> > > >
> > > > PTR_TO_*_OR_NULL is not accepted as an ARG_PTR_TO_BTF_ID_SOCK_COMMON
> > > > at verification time.  All PTR_TO_*_OR_NULL reg has to do a NULL check
> > > > first before passing into the helper or else the bpf prog will be
> > > > rejected by the verifier.
> > > >
> > > > [ ARG_PTR_TO_SOCK_COMMON_OR_NULL was attempted earlier.  The _OR_NULL was
> > > >   needed because the PTR_TO_BTF_ID could be NULL but note that a could be NULL
> > > >   PTR_TO_BTF_ID is not a scalar NULL to the verifier.  "_OR_NULL" implicitly
> > > >   gives an expectation that the helper can take a scalar NULL which does
> > > >   not make sense in most (except one) helpers.  Passing scalar NULL
> > > >   should be rejected at the verification time.
> > >
> > > What is the benefit of requiring a !sk check from the user if all of
> > > the helpers know how to deal with a NULL pointer?
> > I don't see a reason why the verifier should not reject an incorrect
> > program at load time if it can.
> >
> > >
> > > >
> > > >   Thus, this patch uses ARG_PTR_TO_BTF_ID_SOCK_COMMON to specify that the
> > > >   helper can take both the btf-id ptr or the legacy PTR_TO_SOCK_COMMON but
> > > >   not scalar NULL.  It requires the func_proto to explicitly specify the
> > > >   arg_btf_id such that there is a very clear expectation that the helper
> > > >   can handle a NULL PTR_TO_BTF_ID. ]
> > >
> > > I think ARG_PTR_TO_BTF_ID_SOCK_COMMON is actually a misnomer, since
> > > nothing enforces that arg_btf_id is actually an ID for sock common.
> > > This is where ARG_PTR_TO_SOCK_COMMON_OR_NULL is much easier to
> > > understand, even though it's more permissive than it has to be. It
> > > communicates very clearly what values the argument can take.
> > _OR_NULL is incorrect which implies a scalar NULL as mentioned in
> > this commit message.  From verifier pov, _OR_NULL can take
> > a scalar NULL.
> 
> Yes, I know. I'm saying that the distinction between scalar NULL and
> runtime NULL only makes sense after you understand how BTF pointers
> are implemented. It only clicked for me after I read the support code
> in the JIT that Yonghong pointed out. Should everybody that writes a
> helper need to read the JIT? In my opinion we shouldn't. I guess I
> don't even care about the verifier rejecting scalar NULL or not, I'd
> just like the types to have a name that conveys their NULLness.
It is not only about verifier and/or JIT, not sure why it is related to
JIT also.

For some helpers, explicitly passing NULL may make sense.
e.g. bpf_sk_assign(ctx, NULL, 0) makes sense.

For most helpers, the bpf prog is wrong for sure, for example
in sockmap, what does bpf_map_update_elem(sock_map, key, NULL, 0)
mean?  I would expect a delete from the sock_map if the verifier
accepted it.

> 
> >
> > >
> > > If you're set on ARG_PTR_TO_BTF_ID_SOCK_COMMON I'd suggest forcing the
> > > btf_id in struct bpf_reg_types. This avoids the weird case where the
> > > btf_id doesn't actually point at sock_common, and it also makes my
> > I have considered the bpf_reg_types option.  I prefer all
> > arg info (arg_type and arg_btf_id) stay in the same one
> > place (i.e. func_proto) as much as possible for now
> > instead of introducing another place to specify/override it
> > which then depends on a particular arg_type that some arg_type may be
> > in func_proto while some may be in other places.
> 
> In my opinion that ship sailed when we started aliasing arg_type to
> multiple reg_type, but OK.
> 
> >
> > The arg_btf_id can be checked in check_btf_id_ok() if it would be a
> > big concern that it might slip through the review but I think the
> > chance is pretty low.
> 
> Why increase the burden on human reviewers? Why add code to check an
> invariant that we could get rid of in the first place?
Lets take the scalar NULL example that requires to read multiple
pieces of codes in different places (verifier, JIT...etc.).
As you also mentioned, yes, it may be easy for a few people.
However, for most others, having some obvious things in the same place is
easier to review.

I think we have to agree we disagree on this one implementation details
which I think it has been over-thought (and time also).

If you insist that should go into bpf_reg_types (i.e. compatible->btf_id),
I can do that in v4 and then add another check in another place to
ensure "!compatible->btf_id" as in v2.

Re: [PATCH v3 bpf-next 02/11] bpf: Enable bpf_skc_to_* sock casting helper to networking prog type

[Lorenz Bauer]

On Wed, 23 Sep 2020 at 18:06, Martin KaFai Lau <kafai@fb.com> wrote:
>
> On Wed, Sep 23, 2020 at 10:27:27AM +0100, Lorenz Bauer wrote:
> > On Tue, 22 Sep 2020 at 19:26, Martin KaFai Lau <kafai@fb.com> wrote:
> > >
> > > On Tue, Sep 22, 2020 at 10:46:41AM +0100, Lorenz Bauer wrote:
> > > > On Tue, 22 Sep 2020 at 08:04, Martin KaFai Lau <kafai@fb.com> wrote:
> > > > >
> > > > > There is a constant need to add more fields into the bpf_tcp_sock
> > > > > for the bpf programs running at tc, sock_ops...etc.
> > > > >
> > > > > A current workaround could be to use bpf_probe_read_kernel().  However,
> > > > > other than making another helper call for reading each field and missing
> > > > > CO-RE, it is also not as intuitive to use as directly reading
> > > > > "tp->lsndtime" for example.  While already having perfmon cap to do
> > > > > bpf_probe_read_kernel(), it will be much easier if the bpf prog can
> > > > > directly read from the tcp_sock.
> > > > >
> > > > > This patch tries to do that by using the existing casting-helpers
> > > > > bpf_skc_to_*() whose func_proto returns a btf_id.  For example, the
> > > > > func_proto of bpf_skc_to_tcp_sock returns the btf_id of the
> > > > > kernel "struct tcp_sock".
> > > > >
> > > > > These helpers are also added to is_ptr_cast_function().
> > > > > It ensures the returning reg (BPF_REF_0) will also carries the ref_obj_id.
> > > > > That will keep the ref-tracking works properly.
> > > > >
> > > > > The bpf_skc_to_* helpers are made available to most of the bpf prog
> > > > > types in filter.c. They are limited by perfmon cap.
> > > > >
> > > > > This patch adds a ARG_PTR_TO_BTF_ID_SOCK_COMMON.  The helper accepting
> > > > > this arg can accept a btf-id-ptr (PTR_TO_BTF_ID + &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON])
> > > > > or a legacy-ctx-convert-skc-ptr (PTR_TO_SOCK_COMMON).  The bpf_skc_to_*()
> > > > > helpers are changed to take ARG_PTR_TO_BTF_ID_SOCK_COMMON such that
> > > > > they will accept pointer obtained from skb->sk.
> > > > >
> > > > > PTR_TO_*_OR_NULL is not accepted as an ARG_PTR_TO_BTF_ID_SOCK_COMMON
> > > > > at verification time.  All PTR_TO_*_OR_NULL reg has to do a NULL check
> > > > > first before passing into the helper or else the bpf prog will be
> > > > > rejected by the verifier.
> > > > >
> > > > > [ ARG_PTR_TO_SOCK_COMMON_OR_NULL was attempted earlier.  The _OR_NULL was
> > > > >   needed because the PTR_TO_BTF_ID could be NULL but note that a could be NULL
> > > > >   PTR_TO_BTF_ID is not a scalar NULL to the verifier.  "_OR_NULL" implicitly
> > > > >   gives an expectation that the helper can take a scalar NULL which does
> > > > >   not make sense in most (except one) helpers.  Passing scalar NULL
> > > > >   should be rejected at the verification time.
> > > >
> > > > What is the benefit of requiring a !sk check from the user if all of
> > > > the helpers know how to deal with a NULL pointer?
> > > I don't see a reason why the verifier should not reject an incorrect
> > > program at load time if it can.
> > >
> > > >
> > > > >
> > > > >   Thus, this patch uses ARG_PTR_TO_BTF_ID_SOCK_COMMON to specify that the
> > > > >   helper can take both the btf-id ptr or the legacy PTR_TO_SOCK_COMMON but
> > > > >   not scalar NULL.  It requires the func_proto to explicitly specify the
> > > > >   arg_btf_id such that there is a very clear expectation that the helper
> > > > >   can handle a NULL PTR_TO_BTF_ID. ]
> > > >
> > > > I think ARG_PTR_TO_BTF_ID_SOCK_COMMON is actually a misnomer, since
> > > > nothing enforces that arg_btf_id is actually an ID for sock common.
> > > > This is where ARG_PTR_TO_SOCK_COMMON_OR_NULL is much easier to
> > > > understand, even though it's more permissive than it has to be. It
> > > > communicates very clearly what values the argument can take.
> > > _OR_NULL is incorrect which implies a scalar NULL as mentioned in
> > > this commit message.  From verifier pov, _OR_NULL can take
> > > a scalar NULL.
> >
> > Yes, I know. I'm saying that the distinction between scalar NULL and
> > runtime NULL only makes sense after you understand how BTF pointers
> > are implemented. It only clicked for me after I read the support code
> > in the JIT that Yonghong pointed out. Should everybody that writes a
> > helper need to read the JIT? In my opinion we shouldn't. I guess I
> > don't even care about the verifier rejecting scalar NULL or not, I'd
> > just like the types to have a name that conveys their NULLness.
> It is not only about verifier and/or JIT, not sure why it is related to
> JIT also.
>
> For some helpers, explicitly passing NULL may make sense.
> e.g. bpf_sk_assign(ctx, NULL, 0) makes sense.
>
> For most helpers, the bpf prog is wrong for sure, for example
> in sockmap, what does bpf_map_update_elem(sock_map, key, NULL, 0)
> mean?  I would expect a delete from the sock_map if the verifier
> accepted it.
>
> >
> > >
> > > >
> > > > If you're set on ARG_PTR_TO_BTF_ID_SOCK_COMMON I'd suggest forcing the
> > > > btf_id in struct bpf_reg_types. This avoids the weird case where the
> > > > btf_id doesn't actually point at sock_common, and it also makes my
> > > I have considered the bpf_reg_types option.  I prefer all
> > > arg info (arg_type and arg_btf_id) stay in the same one
> > > place (i.e. func_proto) as much as possible for now
> > > instead of introducing another place to specify/override it
> > > which then depends on a particular arg_type that some arg_type may be
> > > in func_proto while some may be in other places.
> >
> > In my opinion that ship sailed when we started aliasing arg_type to
> > multiple reg_type, but OK.
> >
> > >
> > > The arg_btf_id can be checked in check_btf_id_ok() if it would be a
> > > big concern that it might slip through the review but I think the
> > > chance is pretty low.
> >
> > Why increase the burden on human reviewers? Why add code to check an
> > invariant that we could get rid of in the first place?
> Lets take the scalar NULL example that requires to read multiple
> pieces of codes in different places (verifier, JIT...etc.).
> As you also mentioned, yes, it may be easy for a few people.
> However, for most others, having some obvious things in the same place is
> easier to review.
>
> I think we have to agree we disagree on this one implementation details
> which I think it has been over-thought (and time also).
>
> If you insist that should go into bpf_reg_types (i.e. compatible->btf_id),
> I can do that in v4 and then add another check in another place to
> ensure "!compatible->btf_id" as in v2.

No, I don't insist. I was hoping I could convince you, but alas :)

-- 
Lorenz Bauer  |  Systems Engineer
6th Floor, County Hall/The Riverside Building, SE1 7PB, UK

www.cloudflare.com

Re: [PATCH v3 bpf-next 02/11] bpf: Enable bpf_skc_to_* sock casting helper to networking prog type

[Martin KaFai Lau]

On Thu, Sep 24, 2020 at 09:38:09AM +0100, Lorenz Bauer wrote:
> On Wed, 23 Sep 2020 at 18:06, Martin KaFai Lau <kafai@fb.com> wrote:
> >
> > On Wed, Sep 23, 2020 at 10:27:27AM +0100, Lorenz Bauer wrote:
> > > On Tue, 22 Sep 2020 at 19:26, Martin KaFai Lau <kafai@fb.com> wrote:
> > > >
> > > > On Tue, Sep 22, 2020 at 10:46:41AM +0100, Lorenz Bauer wrote:
> > > > > On Tue, 22 Sep 2020 at 08:04, Martin KaFai Lau <kafai@fb.com> wrote:
> > > > > >
> > > > > > There is a constant need to add more fields into the bpf_tcp_sock
> > > > > > for the bpf programs running at tc, sock_ops...etc.
> > > > > >
> > > > > > A current workaround could be to use bpf_probe_read_kernel().  However,
> > > > > > other than making another helper call for reading each field and missing
> > > > > > CO-RE, it is also not as intuitive to use as directly reading
> > > > > > "tp->lsndtime" for example.  While already having perfmon cap to do
> > > > > > bpf_probe_read_kernel(), it will be much easier if the bpf prog can
> > > > > > directly read from the tcp_sock.
> > > > > >
> > > > > > This patch tries to do that by using the existing casting-helpers
> > > > > > bpf_skc_to_*() whose func_proto returns a btf_id.  For example, the
> > > > > > func_proto of bpf_skc_to_tcp_sock returns the btf_id of the
> > > > > > kernel "struct tcp_sock".
> > > > > >
> > > > > > These helpers are also added to is_ptr_cast_function().
> > > > > > It ensures the returning reg (BPF_REF_0) will also carries the ref_obj_id.
> > > > > > That will keep the ref-tracking works properly.
> > > > > >
> > > > > > The bpf_skc_to_* helpers are made available to most of the bpf prog
> > > > > > types in filter.c. They are limited by perfmon cap.
> > > > > >
> > > > > > This patch adds a ARG_PTR_TO_BTF_ID_SOCK_COMMON.  The helper accepting
> > > > > > this arg can accept a btf-id-ptr (PTR_TO_BTF_ID + &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON])
> > > > > > or a legacy-ctx-convert-skc-ptr (PTR_TO_SOCK_COMMON).  The bpf_skc_to_*()
> > > > > > helpers are changed to take ARG_PTR_TO_BTF_ID_SOCK_COMMON such that
> > > > > > they will accept pointer obtained from skb->sk.
> > > > > >
> > > > > > PTR_TO_*_OR_NULL is not accepted as an ARG_PTR_TO_BTF_ID_SOCK_COMMON
> > > > > > at verification time.  All PTR_TO_*_OR_NULL reg has to do a NULL check
> > > > > > first before passing into the helper or else the bpf prog will be
> > > > > > rejected by the verifier.
> > > > > >
> > > > > > [ ARG_PTR_TO_SOCK_COMMON_OR_NULL was attempted earlier.  The _OR_NULL was
> > > > > >   needed because the PTR_TO_BTF_ID could be NULL but note that a could be NULL
> > > > > >   PTR_TO_BTF_ID is not a scalar NULL to the verifier.  "_OR_NULL" implicitly
> > > > > >   gives an expectation that the helper can take a scalar NULL which does
> > > > > >   not make sense in most (except one) helpers.  Passing scalar NULL
> > > > > >   should be rejected at the verification time.
> > > > >
> > > > > What is the benefit of requiring a !sk check from the user if all of
> > > > > the helpers know how to deal with a NULL pointer?
> > > > I don't see a reason why the verifier should not reject an incorrect
> > > > program at load time if it can.
> > > >
> > > > >
> > > > > >
> > > > > >   Thus, this patch uses ARG_PTR_TO_BTF_ID_SOCK_COMMON to specify that the
> > > > > >   helper can take both the btf-id ptr or the legacy PTR_TO_SOCK_COMMON but
> > > > > >   not scalar NULL.  It requires the func_proto to explicitly specify the
> > > > > >   arg_btf_id such that there is a very clear expectation that the helper
> > > > > >   can handle a NULL PTR_TO_BTF_ID. ]
> > > > >
> > > > > I think ARG_PTR_TO_BTF_ID_SOCK_COMMON is actually a misnomer, since
> > > > > nothing enforces that arg_btf_id is actually an ID for sock common.
> > > > > This is where ARG_PTR_TO_SOCK_COMMON_OR_NULL is much easier to
> > > > > understand, even though it's more permissive than it has to be. It
> > > > > communicates very clearly what values the argument can take.
> > > > _OR_NULL is incorrect which implies a scalar NULL as mentioned in
> > > > this commit message.  From verifier pov, _OR_NULL can take
> > > > a scalar NULL.
> > >
> > > Yes, I know. I'm saying that the distinction between scalar NULL and
> > > runtime NULL only makes sense after you understand how BTF pointers
> > > are implemented. It only clicked for me after I read the support code
> > > in the JIT that Yonghong pointed out. Should everybody that writes a
> > > helper need to read the JIT? In my opinion we shouldn't. I guess I
> > > don't even care about the verifier rejecting scalar NULL or not, I'd
> > > just like the types to have a name that conveys their NULLness.
> > It is not only about verifier and/or JIT, not sure why it is related to
> > JIT also.
> >
> > For some helpers, explicitly passing NULL may make sense.
> > e.g. bpf_sk_assign(ctx, NULL, 0) makes sense.
> >
> > For most helpers, the bpf prog is wrong for sure, for example
> > in sockmap, what does bpf_map_update_elem(sock_map, key, NULL, 0)
> > mean?  I would expect a delete from the sock_map if the verifier
> > accepted it.
> >
> > >
> > > >
> > > > >
> > > > > If you're set on ARG_PTR_TO_BTF_ID_SOCK_COMMON I'd suggest forcing the
> > > > > btf_id in struct bpf_reg_types. This avoids the weird case where the
> > > > > btf_id doesn't actually point at sock_common, and it also makes my
> > > > I have considered the bpf_reg_types option.  I prefer all
> > > > arg info (arg_type and arg_btf_id) stay in the same one
> > > > place (i.e. func_proto) as much as possible for now
> > > > instead of introducing another place to specify/override it
> > > > which then depends on a particular arg_type that some arg_type may be
> > > > in func_proto while some may be in other places.
> > >
> > > In my opinion that ship sailed when we started aliasing arg_type to
> > > multiple reg_type, but OK.
> > >
> > > >
> > > > The arg_btf_id can be checked in check_btf_id_ok() if it would be a
> > > > big concern that it might slip through the review but I think the
> > > > chance is pretty low.
> > >
> > > Why increase the burden on human reviewers? Why add code to check an
> > > invariant that we could get rid of in the first place?
> > Lets take the scalar NULL example that requires to read multiple
> > pieces of codes in different places (verifier, JIT...etc.).
> > As you also mentioned, yes, it may be easy for a few people.
> > However, for most others, having some obvious things in the same place is
> > easier to review.
> >
> > I think we have to agree we disagree on this one implementation details
> > which I think it has been over-thought (and time also).
> >
> > If you insist that should go into bpf_reg_types (i.e. compatible->btf_id),
> > I can do that in v4 and then add another check in another place to
> > ensure "!compatible->btf_id" as in v2.
> 
> No, I don't insist. I was hoping I could convince you, but alas :)

Not very convinced but I don't feel very strongly on this ;)
so I have posted v4 with btf_id in struct bpf_reg_types to continue
the review of the set.  Thanks.

[PATCH v3 bpf-next 03/11] bpf: Change bpf_sk_release and bpf_sk_*cgroup_id to accept ARG_PTR_TO_BTF_ID_SOCK_COMMON

[Martin KaFai Lau]

The previous patch allows the networking bpf prog to use the
bpf_skc_to_*() helpers to get a PTR_TO_BTF_ID socket pointer,
e.g. "struct tcp_sock *".  It allows the bpf prog to read all the
fields of the tcp_sock.

This patch changes the bpf_sk_release() and bpf_sk_*cgroup_id()
to take ARG_PTR_TO_BTF_ID_SOCK_COMMON such that they will
work with the pointer returned by the bpf_skc_to_*() helpers
also.  For example, the following will work:

	sk = bpf_skc_lookup_tcp(skb, tuple, tuplen, BPF_F_CURRENT_NETNS, 0);
	if (!sk)
		return;
	tp = bpf_skc_to_tcp_sock(sk);
	if (!tp) {
		bpf_sk_release(sk);
		return;
	}
	lsndtime = tp->lsndtime;
	/* Pass tp to bpf_sk_release() will also work */
	bpf_sk_release(tp);

Since PTR_TO_BTF_ID could be NULL, the helper taking
ARG_PTR_TO_BTF_ID_SOCK_COMMON has to check for NULL at runtime.

A btf_id of "struct sock" may not always mean a fullsock.  Regardless
the helper's running context may get a non-fullsock or not,
considering fullsock check/handling is pretty cheap, it is better to
keep the same verifier expectation on helper that takes ARG_PTR_TO_BTF_ID*
will be able to handle the minisock situation.  In the bpf_sk_*cgroup_id()
case,  it will try to get a fullsock by using sk_to_full_sk() as its
skb variant bpf_sk"b"_*cgroup_id() has already been doing.

bpf_sk_release can already handle minisock, so nothing special has to
be done.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
---
 net/core/filter.c | 33 +++++++++++++++++----------------
 1 file changed, 17 insertions(+), 16 deletions(-)

diff --git a/net/core/filter.c b/net/core/filter.c
index 54b338de4bb8..532a85894ce0 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -4088,18 +4088,17 @@ static inline u64 __bpf_sk_cgroup_id(struct sock *sk)
 {
 	struct cgroup *cgrp;
 
+	sk = sk_to_full_sk(sk);
+	if (!sk || !sk_fullsock(sk))
+		return 0;
+
 	cgrp = sock_cgroup_ptr(&sk->sk_cgrp_data);
 	return cgroup_id(cgrp);
 }
 
 BPF_CALL_1(bpf_skb_cgroup_id, const struct sk_buff *, skb)
 {
-	struct sock *sk = skb_to_full_sk(skb);
-
-	if (!sk || !sk_fullsock(sk))
-		return 0;
-
-	return __bpf_sk_cgroup_id(sk);
+	return __bpf_sk_cgroup_id(skb->sk);
 }
 
 static const struct bpf_func_proto bpf_skb_cgroup_id_proto = {
@@ -4115,6 +4114,10 @@ static inline u64 __bpf_sk_ancestor_cgroup_id(struct sock *sk,
 	struct cgroup *ancestor;
 	struct cgroup *cgrp;
 
+	sk = sk_to_full_sk(sk);
+	if (!sk || !sk_fullsock(sk))
+		return 0;
+
 	cgrp = sock_cgroup_ptr(&sk->sk_cgrp_data);
 	ancestor = cgroup_ancestor(cgrp, ancestor_level);
 	if (!ancestor)
@@ -4126,12 +4129,7 @@ static inline u64 __bpf_sk_ancestor_cgroup_id(struct sock *sk,
 BPF_CALL_2(bpf_skb_ancestor_cgroup_id, const struct sk_buff *, skb, int,
 	   ancestor_level)
 {
-	struct sock *sk = skb_to_full_sk(skb);
-
-	if (!sk || !sk_fullsock(sk))
-		return 0;
-
-	return __bpf_sk_ancestor_cgroup_id(sk, ancestor_level);
+	return __bpf_sk_ancestor_cgroup_id(skb->sk, ancestor_level);
 }
 
 static const struct bpf_func_proto bpf_skb_ancestor_cgroup_id_proto = {
@@ -4151,7 +4149,8 @@ static const struct bpf_func_proto bpf_sk_cgroup_id_proto = {
 	.func           = bpf_sk_cgroup_id,
 	.gpl_only       = false,
 	.ret_type       = RET_INTEGER,
-	.arg1_type      = ARG_PTR_TO_SOCKET,
+	.arg1_type      = ARG_PTR_TO_BTF_ID_SOCK_COMMON,
+	.arg1_btf_id	= &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON],
 };
 
 BPF_CALL_2(bpf_sk_ancestor_cgroup_id, struct sock *, sk, int, ancestor_level)
@@ -4163,7 +4162,8 @@ static const struct bpf_func_proto bpf_sk_ancestor_cgroup_id_proto = {
 	.func           = bpf_sk_ancestor_cgroup_id,
 	.gpl_only       = false,
 	.ret_type       = RET_INTEGER,
-	.arg1_type      = ARG_PTR_TO_SOCKET,
+	.arg1_type      = ARG_PTR_TO_BTF_ID_SOCK_COMMON,
+	.arg1_btf_id	= &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON],
 	.arg2_type      = ARG_ANYTHING,
 };
 #endif
@@ -5696,7 +5696,7 @@ static const struct bpf_func_proto bpf_sk_lookup_udp_proto = {
 
 BPF_CALL_1(bpf_sk_release, struct sock *, sk)
 {
-	if (sk_is_refcounted(sk))
+	if (sk && sk_is_refcounted(sk))
 		sock_gen_put(sk);
 	return 0;
 }
@@ -5705,7 +5705,8 @@ static const struct bpf_func_proto bpf_sk_release_proto = {
 	.func		= bpf_sk_release,
 	.gpl_only	= false,
 	.ret_type	= RET_INTEGER,
-	.arg1_type	= ARG_PTR_TO_SOCK_COMMON,
+	.arg1_type	= ARG_PTR_TO_BTF_ID_SOCK_COMMON,
+	.arg1_btf_id	= &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON],
 };
 
 BPF_CALL_5(bpf_xdp_sk_lookup_udp, struct xdp_buff *, ctx,
-- 
2.24.1

[PATCH v3 bpf-next 04/11] bpf: Change bpf_sk_storage_*() to accept ARG_PTR_TO_BTF_ID_SOCK_COMMON

[Martin KaFai Lau]

This patch changes the bpf_sk_storage_*() to take
ARG_PTR_TO_BTF_ID_SOCK_COMMON such that they will work with the pointer
returned by the bpf_skc_to_*() helpers also.

A micro benchmark has been done on a "cgroup_skb/egress" bpf program
which does a bpf_sk_storage_get().  It was driven by netperf doing
a 4096 connected UDP_STREAM test with 64bytes packet.
The stats from "kernel.bpf_stats_enabled" shows no meaningful difference.

The sk_storage_get_btf_proto, sk_storage_delete_btf_proto,
btf_sk_storage_get_proto, and btf_sk_storage_delete_proto are
no longer needed, so they are removed.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
---
 include/net/bpf_sk_storage.h   |  2 --
 include/uapi/linux/bpf.h       |  1 +
 kernel/bpf/bpf_lsm.c           |  4 ++--
 net/core/bpf_sk_storage.c      | 31 ++++++++-----------------------
 net/ipv4/bpf_tcp_ca.c          | 23 ++---------------------
 tools/include/uapi/linux/bpf.h |  1 +
 6 files changed, 14 insertions(+), 48 deletions(-)

diff --git a/include/net/bpf_sk_storage.h b/include/net/bpf_sk_storage.h
index 119f4c9c3a9c..3c516dd07caf 100644
--- a/include/net/bpf_sk_storage.h
+++ b/include/net/bpf_sk_storage.h
@@ -20,8 +20,6 @@ void bpf_sk_storage_free(struct sock *sk);
 
 extern const struct bpf_func_proto bpf_sk_storage_get_proto;
 extern const struct bpf_func_proto bpf_sk_storage_delete_proto;
-extern const struct bpf_func_proto sk_storage_get_btf_proto;
-extern const struct bpf_func_proto sk_storage_delete_btf_proto;
 
 struct bpf_local_storage_elem;
 struct bpf_sk_storage_diag;
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index a22812561064..1634ca9b752a 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -2861,6 +2861,7 @@ union bpf_attr {
  *		0 on success.
  *
  *		**-ENOENT** if the bpf-local-storage cannot be found.
+ *		**-EINVAL** if sk is not a fullsock (e.g. a request_sock).
  *
  * long bpf_send_signal(u32 sig)
  *	Description
diff --git a/kernel/bpf/bpf_lsm.c b/kernel/bpf/bpf_lsm.c
index 9cd1428c7199..78ea8a7bd27f 100644
--- a/kernel/bpf/bpf_lsm.c
+++ b/kernel/bpf/bpf_lsm.c
@@ -56,9 +56,9 @@ bpf_lsm_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 	case BPF_FUNC_inode_storage_delete:
 		return &bpf_inode_storage_delete_proto;
 	case BPF_FUNC_sk_storage_get:
-		return &sk_storage_get_btf_proto;
+		return &bpf_sk_storage_get_proto;
 	case BPF_FUNC_sk_storage_delete:
-		return &sk_storage_delete_btf_proto;
+		return &bpf_sk_storage_delete_proto;
 	default:
 		return tracing_prog_func_proto(func_id, prog);
 	}
diff --git a/net/core/bpf_sk_storage.c b/net/core/bpf_sk_storage.c
index 838efc682cff..aa39e9676050 100644
--- a/net/core/bpf_sk_storage.c
+++ b/net/core/bpf_sk_storage.c
@@ -269,7 +269,7 @@ BPF_CALL_4(bpf_sk_storage_get, struct bpf_map *, map, struct sock *, sk,
 {
 	struct bpf_local_storage_data *sdata;
 
-	if (flags > BPF_SK_STORAGE_GET_F_CREATE)
+	if (!sk || !sk_fullsock(sk) || flags > BPF_SK_STORAGE_GET_F_CREATE)
 		return (unsigned long)NULL;
 
 	sdata = sk_storage_lookup(sk, map, true);
@@ -299,6 +299,9 @@ BPF_CALL_4(bpf_sk_storage_get, struct bpf_map *, map, struct sock *, sk,
 
 BPF_CALL_2(bpf_sk_storage_delete, struct bpf_map *, map, struct sock *, sk)
 {
+	if (!sk || !sk_fullsock(sk))
+		return -EINVAL;
+
 	if (refcount_inc_not_zero(&sk->sk_refcnt)) {
 		int err;
 
@@ -355,7 +358,8 @@ const struct bpf_func_proto bpf_sk_storage_get_proto = {
 	.gpl_only	= false,
 	.ret_type	= RET_PTR_TO_MAP_VALUE_OR_NULL,
 	.arg1_type	= ARG_CONST_MAP_PTR,
-	.arg2_type	= ARG_PTR_TO_SOCKET,
+	.arg2_type	= ARG_PTR_TO_BTF_ID_SOCK_COMMON,
+	.arg2_btf_id	= &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON],
 	.arg3_type	= ARG_PTR_TO_MAP_VALUE_OR_NULL,
 	.arg4_type	= ARG_ANYTHING,
 };
@@ -375,27 +379,8 @@ const struct bpf_func_proto bpf_sk_storage_delete_proto = {
 	.gpl_only	= false,
 	.ret_type	= RET_INTEGER,
 	.arg1_type	= ARG_CONST_MAP_PTR,
-	.arg2_type	= ARG_PTR_TO_SOCKET,
-};
-
-const struct bpf_func_proto sk_storage_get_btf_proto = {
-	.func		= bpf_sk_storage_get,
-	.gpl_only	= false,
-	.ret_type	= RET_PTR_TO_MAP_VALUE_OR_NULL,
-	.arg1_type	= ARG_CONST_MAP_PTR,
-	.arg2_type	= ARG_PTR_TO_BTF_ID,
-	.arg2_btf_id	= &btf_sock_ids[BTF_SOCK_TYPE_SOCK],
-	.arg3_type	= ARG_PTR_TO_MAP_VALUE_OR_NULL,
-	.arg4_type	= ARG_ANYTHING,
-};
-
-const struct bpf_func_proto sk_storage_delete_btf_proto = {
-	.func		= bpf_sk_storage_delete,
-	.gpl_only	= false,
-	.ret_type	= RET_INTEGER,
-	.arg1_type	= ARG_CONST_MAP_PTR,
-	.arg2_type	= ARG_PTR_TO_BTF_ID,
-	.arg2_btf_id	= &btf_sock_ids[BTF_SOCK_TYPE_SOCK],
+	.arg2_type	= ARG_PTR_TO_BTF_ID_SOCK_COMMON,
+	.arg2_btf_id	= &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON],
 };
 
 struct bpf_sk_storage_diag {
diff --git a/net/ipv4/bpf_tcp_ca.c b/net/ipv4/bpf_tcp_ca.c
index 74a2ef598c31..618954f82764 100644
--- a/net/ipv4/bpf_tcp_ca.c
+++ b/net/ipv4/bpf_tcp_ca.c
@@ -28,22 +28,6 @@ static u32 unsupported_ops[] = {
 static const struct btf_type *tcp_sock_type;
 static u32 tcp_sock_id, sock_id;
 
-static struct bpf_func_proto btf_sk_storage_get_proto __read_mostly;
-static struct bpf_func_proto btf_sk_storage_delete_proto __read_mostly;
-
-static void convert_sk_func_proto(struct bpf_func_proto *to, const struct bpf_func_proto *from)
-{
-	int i;
-
-	*to = *from;
-	for (i = 0; i < ARRAY_SIZE(to->arg_type); i++) {
-		if (to->arg_type[i] == ARG_PTR_TO_SOCKET) {
-			to->arg_type[i] = ARG_PTR_TO_BTF_ID;
-			to->arg_btf_id[i] = &tcp_sock_id;
-		}
-	}
-}
-
 static int bpf_tcp_ca_init(struct btf *btf)
 {
 	s32 type_id;
@@ -59,9 +43,6 @@ static int bpf_tcp_ca_init(struct btf *btf)
 	tcp_sock_id = type_id;
 	tcp_sock_type = btf_type_by_id(btf, tcp_sock_id);
 
-	convert_sk_func_proto(&btf_sk_storage_get_proto, &bpf_sk_storage_get_proto);
-	convert_sk_func_proto(&btf_sk_storage_delete_proto, &bpf_sk_storage_delete_proto);
-
 	return 0;
 }
 
@@ -188,9 +169,9 @@ bpf_tcp_ca_get_func_proto(enum bpf_func_id func_id,
 	case BPF_FUNC_tcp_send_ack:
 		return &bpf_tcp_send_ack_proto;
 	case BPF_FUNC_sk_storage_get:
-		return &btf_sk_storage_get_proto;
+		return &bpf_sk_storage_get_proto;
 	case BPF_FUNC_sk_storage_delete:
-		return &btf_sk_storage_delete_proto;
+		return &bpf_sk_storage_delete_proto;
 	default:
 		return bpf_base_func_proto(func_id);
 	}
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index a22812561064..1634ca9b752a 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -2861,6 +2861,7 @@ union bpf_attr {
  *		0 on success.
  *
  *		**-ENOENT** if the bpf-local-storage cannot be found.
+ *		**-EINVAL** if sk is not a fullsock (e.g. a request_sock).
  *
  * long bpf_send_signal(u32 sig)
  *	Description
-- 
2.24.1

Re: [PATCH v3 bpf-next 04/11] bpf: Change bpf_sk_storage_*() to accept ARG_PTR_TO_BTF_ID_SOCK_COMMON

[Lorenz Bauer]

On Tue, 22 Sep 2020 at 08:04, Martin KaFai Lau <kafai@fb.com> wrote:
>
> This patch changes the bpf_sk_storage_*() to take
> ARG_PTR_TO_BTF_ID_SOCK_COMMON such that they will work with the pointer
> returned by the bpf_skc_to_*() helpers also.
>
> A micro benchmark has been done on a "cgroup_skb/egress" bpf program
> which does a bpf_sk_storage_get().  It was driven by netperf doing
> a 4096 connected UDP_STREAM test with 64bytes packet.
> The stats from "kernel.bpf_stats_enabled" shows no meaningful difference.
>
> The sk_storage_get_btf_proto, sk_storage_delete_btf_proto,
> btf_sk_storage_get_proto, and btf_sk_storage_delete_proto are
> no longer needed, so they are removed.
>
> Signed-off-by: Martin KaFai Lau <kafai@fb.com>

I like that we can get rid of the init code.

Acked-by: Lorenz Bauer <lmb@cloudflare.com>

--
Lorenz Bauer  |  Systems Engineer
6th Floor, County Hall/The Riverside Building, SE1 7PB, UK

www.cloudflare.com

[PATCH v3 bpf-next 05/11] bpf: Change bpf_tcp_*_syncookie to accept ARG_PTR_TO_BTF_ID_SOCK_COMMON

[Martin KaFai Lau]

This patch changes the bpf_tcp_*_syncookie() to take
ARG_PTR_TO_BTF_ID_SOCK_COMMON such that they will work with the pointer
returned by the bpf_skc_to_*() helpers also.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
---
 net/core/filter.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/net/core/filter.c b/net/core/filter.c
index 532a85894ce0..6ab12d8cdd85 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -6088,7 +6088,7 @@ BPF_CALL_5(bpf_tcp_check_syncookie, struct sock *, sk, void *, iph, u32, iph_len
 	u32 cookie;
 	int ret;
 
-	if (unlikely(th_len < sizeof(*th)))
+	if (unlikely(!sk || th_len < sizeof(*th)))
 		return -EINVAL;
 
 	/* sk_listener() allows TCP_NEW_SYN_RECV, which makes no sense here. */
@@ -6141,7 +6141,8 @@ static const struct bpf_func_proto bpf_tcp_check_syncookie_proto = {
 	.gpl_only	= true,
 	.pkt_access	= true,
 	.ret_type	= RET_INTEGER,
-	.arg1_type	= ARG_PTR_TO_SOCK_COMMON,
+	.arg1_type	= ARG_PTR_TO_BTF_ID_SOCK_COMMON,
+	.arg1_btf_id	= &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON],
 	.arg2_type	= ARG_PTR_TO_MEM,
 	.arg3_type	= ARG_CONST_SIZE,
 	.arg4_type	= ARG_PTR_TO_MEM,
@@ -6155,7 +6156,7 @@ BPF_CALL_5(bpf_tcp_gen_syncookie, struct sock *, sk, void *, iph, u32, iph_len,
 	u32 cookie;
 	u16 mss;
 
-	if (unlikely(th_len < sizeof(*th) || th_len != th->doff * 4))
+	if (unlikely(!sk || th_len < sizeof(*th) || th_len != th->doff * 4))
 		return -EINVAL;
 
 	if (sk->sk_protocol != IPPROTO_TCP || sk->sk_state != TCP_LISTEN)
@@ -6210,7 +6211,8 @@ static const struct bpf_func_proto bpf_tcp_gen_syncookie_proto = {
 	.gpl_only	= true, /* __cookie_v*_init_sequence() is GPL */
 	.pkt_access	= true,
 	.ret_type	= RET_INTEGER,
-	.arg1_type	= ARG_PTR_TO_SOCK_COMMON,
+	.arg1_type	= ARG_PTR_TO_BTF_ID_SOCK_COMMON,
+	.arg1_btf_id	= &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON],
 	.arg2_type	= ARG_PTR_TO_MEM,
 	.arg3_type	= ARG_CONST_SIZE,
 	.arg4_type	= ARG_PTR_TO_MEM,
-- 
2.24.1

Re: [PATCH v3 bpf-next 05/11] bpf: Change bpf_tcp_*_syncookie to accept ARG_PTR_TO_BTF_ID_SOCK_COMMON

[Lorenz Bauer]

On Tue, 22 Sep 2020 at 08:04, Martin KaFai Lau <kafai@fb.com> wrote:
>
> This patch changes the bpf_tcp_*_syncookie() to take
> ARG_PTR_TO_BTF_ID_SOCK_COMMON such that they will work with the pointer
> returned by the bpf_skc_to_*() helpers also.
>
> Signed-off-by: Martin KaFai Lau <kafai@fb.com>

Acked-by: Lorenz Bauer <lmb@cloudflare.com>

-- 
Lorenz Bauer  |  Systems Engineer
6th Floor, County Hall/The Riverside Building, SE1 7PB, UK

www.cloudflare.com

[PATCH v3 bpf-next 06/11] bpf: Change bpf_sk_assign to accept ARG_PTR_TO_BTF_ID_SOCK_COMMON

[Martin KaFai Lau]

This patch changes the bpf_sk_assign() to take
ARG_PTR_TO_BTF_ID_SOCK_COMMON such that they will work with the pointer
returned by the bpf_skc_to_*() helpers also.

The bpf_sk_lookup_assign() is taking ARG_PTR_TO_SOCKET_"OR_NULL".  Meaning
it specifically takes a scalar NULL.  ARG_PTR_TO_BTF_ID_SOCK_COMMON
does not allow a scalar NULL, so another ARG type is required
for this purpose and another folllow-up patch can be used if
there is such need.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
---
 net/core/filter.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/net/core/filter.c b/net/core/filter.c
index 6ab12d8cdd85..063aba8a81e6 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -6221,7 +6221,7 @@ static const struct bpf_func_proto bpf_tcp_gen_syncookie_proto = {
 
 BPF_CALL_3(bpf_sk_assign, struct sk_buff *, skb, struct sock *, sk, u64, flags)
 {
-	if (flags != 0)
+	if (!sk || flags != 0)
 		return -EINVAL;
 	if (!skb_at_tc_ingress(skb))
 		return -EOPNOTSUPP;
@@ -6245,7 +6245,8 @@ static const struct bpf_func_proto bpf_sk_assign_proto = {
 	.gpl_only	= false,
 	.ret_type	= RET_INTEGER,
 	.arg1_type      = ARG_PTR_TO_CTX,
-	.arg2_type      = ARG_PTR_TO_SOCK_COMMON,
+	.arg2_type      = ARG_PTR_TO_BTF_ID_SOCK_COMMON,
+	.arg2_btf_id	= &btf_sock_ids[BTF_SOCK_TYPE_SOCK_COMMON],
 	.arg3_type	= ARG_ANYTHING,
 };
 
-- 
2.24.1

Re: [PATCH v3 bpf-next 06/11] bpf: Change bpf_sk_assign to accept ARG_PTR_TO_BTF_ID_SOCK_COMMON

[Lorenz Bauer]

On Tue, 22 Sep 2020 at 08:04, Martin KaFai Lau <kafai@fb.com> wrote:
>
> This patch changes the bpf_sk_assign() to take
> ARG_PTR_TO_BTF_ID_SOCK_COMMON such that they will work with the pointer
> returned by the bpf_skc_to_*() helpers also.
>
> The bpf_sk_lookup_assign() is taking ARG_PTR_TO_SOCKET_"OR_NULL".  Meaning
> it specifically takes a scalar NULL.  ARG_PTR_TO_BTF_ID_SOCK_COMMON
> does not allow a scalar NULL, so another ARG type is required
> for this purpose and another folllow-up patch can be used if
> there is such need.
>
> Signed-off-by: Martin KaFai Lau <kafai@fb.com>

Acked-by: Lorenz Bauer <lmb@cloudflare.com>

-- 
Lorenz Bauer  |  Systems Engineer
6th Floor, County Hall/The Riverside Building, SE1 7PB, UK

www.cloudflare.com

[PATCH v3 bpf-next 07/11] bpf: selftest: Add ref_tracking verifier test for bpf_skc casting

[Martin KaFai Lau]

The patch tests for:
1. bpf_sk_release() can be called on a tcp_sock btf_id ptr.

2. Ensure the tcp_sock btf_id pointer cannot be used
   after bpf_sk_release().

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
---
 .../selftests/bpf/verifier/ref_tracking.c     | 47 +++++++++++++++++++
 1 file changed, 47 insertions(+)

diff --git a/tools/testing/selftests/bpf/verifier/ref_tracking.c b/tools/testing/selftests/bpf/verifier/ref_tracking.c
index 056e0273bf12..006b5bd99c08 100644
--- a/tools/testing/selftests/bpf/verifier/ref_tracking.c
+++ b/tools/testing/selftests/bpf/verifier/ref_tracking.c
@@ -854,3 +854,50 @@
 	.errstr = "Unreleased reference",
 	.result = REJECT,
 },
+{
+	"reference tracking: bpf_sk_release(btf_tcp_sock)",
+	.insns = {
+	BPF_SK_LOOKUP(sk_lookup_tcp),
+	BPF_JMP_IMM(BPF_JNE, BPF_REG_0, 0, 1),
+	BPF_EXIT_INSN(),
+	BPF_MOV64_REG(BPF_REG_6, BPF_REG_0),
+	BPF_MOV64_REG(BPF_REG_1, BPF_REG_0),
+	BPF_EMIT_CALL(BPF_FUNC_skc_to_tcp_sock),
+	BPF_JMP_IMM(BPF_JNE, BPF_REG_0, 0, 3),
+	BPF_MOV64_REG(BPF_REG_1, BPF_REG_6),
+	BPF_EMIT_CALL(BPF_FUNC_sk_release),
+	BPF_EXIT_INSN(),
+	BPF_MOV64_REG(BPF_REG_1, BPF_REG_0),
+	BPF_EMIT_CALL(BPF_FUNC_sk_release),
+	BPF_EXIT_INSN(),
+	},
+	.prog_type = BPF_PROG_TYPE_SCHED_CLS,
+	.result = ACCEPT,
+	.result_unpriv = REJECT,
+	.errstr_unpriv = "unknown func",
+},
+{
+	"reference tracking: use ptr from bpf_skc_to_tcp_sock() after release",
+	.insns = {
+	BPF_SK_LOOKUP(sk_lookup_tcp),
+	BPF_JMP_IMM(BPF_JNE, BPF_REG_0, 0, 1),
+	BPF_EXIT_INSN(),
+	BPF_MOV64_REG(BPF_REG_6, BPF_REG_0),
+	BPF_MOV64_REG(BPF_REG_1, BPF_REG_0),
+	BPF_EMIT_CALL(BPF_FUNC_skc_to_tcp_sock),
+	BPF_JMP_IMM(BPF_JNE, BPF_REG_0, 0, 3),
+	BPF_MOV64_REG(BPF_REG_1, BPF_REG_6),
+	BPF_EMIT_CALL(BPF_FUNC_sk_release),
+	BPF_EXIT_INSN(),
+	BPF_MOV64_REG(BPF_REG_7, BPF_REG_0),
+	BPF_MOV64_REG(BPF_REG_1, BPF_REG_6),
+	BPF_EMIT_CALL(BPF_FUNC_sk_release),
+	BPF_LDX_MEM(BPF_B, BPF_REG_0, BPF_REG_7, 0),
+	BPF_EXIT_INSN(),
+	},
+	.prog_type = BPF_PROG_TYPE_SCHED_CLS,
+	.result = REJECT,
+	.errstr = "invalid mem access",
+	.result_unpriv = REJECT,
+	.errstr_unpriv = "unknown func",
+},
-- 
2.24.1

[PATCH v3 bpf-next 08/11] bpf: selftest: Move sock_fields test into test_progs

[Martin KaFai Lau]

This is a mechanical change to
1. move test_sock_fields.c to prog_tests/sock_fields.c
2. rename progs/test_sock_fields_kern.c to progs/test_sock_fields.c

Minimal change is made to the code itself.  Next patch will make
changes to use new ways of writing test, e.g. use skel and global
variables.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
---
 tools/testing/selftests/bpf/Makefile                        | 2 +-
 .../bpf/{test_sock_fields.c => prog_tests/sock_fields.c}    | 6 ++----
 .../progs/{test_sock_fields_kern.c => test_sock_fields.c}   | 0
 3 files changed, 3 insertions(+), 5 deletions(-)
 rename tools/testing/selftests/bpf/{test_sock_fields.c => prog_tests/sock_fields.c} (99%)
 rename tools/testing/selftests/bpf/progs/{test_sock_fields_kern.c => test_sock_fields.c} (100%)

diff --git a/tools/testing/selftests/bpf/Makefile b/tools/testing/selftests/bpf/Makefile
index 59a5fa5fe837..bdbeafec371b 100644
--- a/tools/testing/selftests/bpf/Makefile
+++ b/tools/testing/selftests/bpf/Makefile
@@ -35,7 +35,7 @@ TEST_GEN_PROGS = test_verifier test_tag test_maps test_lru_map test_lpm_map test
 	test_verifier_log test_dev_cgroup test_tcpbpf_user \
 	test_sock test_sockmap get_cgroup_id_user test_socket_cookie \
 	test_cgroup_storage \
-	test_netcnt test_tcpnotify_user test_sock_fields test_sysctl \
+	test_netcnt test_tcpnotify_user test_sysctl \
 	test_progs-no_alu32 \
 	test_current_pid_tgid_new_ns
 
diff --git a/tools/testing/selftests/bpf/test_sock_fields.c b/tools/testing/selftests/bpf/prog_tests/sock_fields.c
similarity index 99%
rename from tools/testing/selftests/bpf/test_sock_fields.c
rename to tools/testing/selftests/bpf/prog_tests/sock_fields.c
index 6c9f269c396d..1138223780fc 100644
--- a/tools/testing/selftests/bpf/test_sock_fields.c
+++ b/tools/testing/selftests/bpf/prog_tests/sock_fields.c
@@ -409,10 +409,10 @@ static void test(void)
 	check_result();
 }
 
-int main(int argc, char **argv)
+void test_sock_fields(void)
 {
 	struct bpf_prog_load_attr attr = {
-		.file = "test_sock_fields_kern.o",
+		.file = "test_sock_fields.o",
 		.prog_type = BPF_PROG_TYPE_CGROUP_SKB,
 		.prog_flags = BPF_F_TEST_RND_HI32,
 	};
@@ -477,6 +477,4 @@ int main(int argc, char **argv)
 	cleanup_cgroup_environment();
 
 	printf("PASS\n");
-
-	return 0;
 }
diff --git a/tools/testing/selftests/bpf/progs/test_sock_fields_kern.c b/tools/testing/selftests/bpf/progs/test_sock_fields.c
similarity index 100%
rename from tools/testing/selftests/bpf/progs/test_sock_fields_kern.c
rename to tools/testing/selftests/bpf/progs/test_sock_fields.c
-- 
2.24.1

Re: [PATCH v3 bpf-next 08/11] bpf: selftest: Move sock_fields test into test_progs

[Andrii Nakryiko]

On Tue, Sep 22, 2020 at 12:40 AM Martin KaFai Lau <kafai@fb.com> wrote:
>
> This is a mechanical change to
> 1. move test_sock_fields.c to prog_tests/sock_fields.c
> 2. rename progs/test_sock_fields_kern.c to progs/test_sock_fields.c
>
> Minimal change is made to the code itself.  Next patch will make
> changes to use new ways of writing test, e.g. use skel and global
> variables.
>
> Signed-off-by: Martin KaFai Lau <kafai@fb.com>
> ---
>  tools/testing/selftests/bpf/Makefile                        | 2 +-
>  .../bpf/{test_sock_fields.c => prog_tests/sock_fields.c}    | 6 ++----
>  .../progs/{test_sock_fields_kern.c => test_sock_fields.c}   | 0

Remove it from .gitignore as well?

>  3 files changed, 3 insertions(+), 5 deletions(-)
>  rename tools/testing/selftests/bpf/{test_sock_fields.c => prog_tests/sock_fields.c} (99%)
>  rename tools/testing/selftests/bpf/progs/{test_sock_fields_kern.c => test_sock_fields.c} (100%)
>

[...]

[PATCH v3 bpf-next 09/11] bpf: selftest: Adapt sock_fields test to use skel and global variables

[Martin KaFai Lau]

skel is used.

Global variables are used to store the result from bpf prog.
addr_map, sock_result_map, and tcp_sock_result_map are gone.
Instead, global variables listen_tp, srv_sa6, cli_tp,, srv_tp,
listen_sk, srv_sk, and cli_sk are added.
Because of that, bpf_addr_array_idx and bpf_result_array_idx are also
no longer needed.

CHECK() macro from test_progs.h is reused and bail as soon as
a CHECK failure.

shutdown() is used to ensure the previous data-ack is received.
The bytes_acked, bytes_received, and the pkt_out_cnt checks are
using "<" to accommodate the final ack may not have been received/sent.
It is enough since it is not the focus of this test.

The sk local storage is all initialized to 0xeB9F now, so the
check_sk_pkt_out_cnt() always checks with the 0xeB9F base.  It is to
keep things simple.

The next patch will reuse helpers from network_helpers.h to simplify
things further.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
---
 .../selftests/bpf/prog_tests/sock_fields.c    | 385 ++++++++----------
 .../selftests/bpf/progs/test_sock_fields.c    | 154 +++----
 2 files changed, 232 insertions(+), 307 deletions(-)

diff --git a/tools/testing/selftests/bpf/prog_tests/sock_fields.c b/tools/testing/selftests/bpf/prog_tests/sock_fields.c
index 1138223780fc..cb6d41bc79a4 100644
--- a/tools/testing/selftests/bpf/prog_tests/sock_fields.c
+++ b/tools/testing/selftests/bpf/prog_tests/sock_fields.c
@@ -14,20 +14,9 @@
 #include <bpf/libbpf.h>
 
 #include "cgroup_helpers.h"
+#include "test_progs.h"
 #include "bpf_rlimit.h"
-
-enum bpf_addr_array_idx {
-	ADDR_SRV_IDX,
-	ADDR_CLI_IDX,
-	__NR_BPF_ADDR_ARRAY_IDX,
-};
-
-enum bpf_result_array_idx {
-	EGRESS_SRV_IDX,
-	EGRESS_CLI_IDX,
-	INGRESS_LISTEN_IDX,
-	__NR_BPF_RESULT_ARRAY_IDX,
-};
+#include "test_sock_fields.skel.h"
 
 enum bpf_linum_array_idx {
 	EGRESS_LINUM_IDX,
@@ -40,34 +29,16 @@ struct bpf_spinlock_cnt {
 	__u32 cnt;
 };
 
-#define CHECK(condition, tag, format...) ({				\
-	int __ret = !!(condition);					\
-	if (__ret) {							\
-		printf("%s(%d):FAIL:%s ", __func__, __LINE__, tag);	\
-		printf(format);						\
-		printf("\n");						\
-		exit(-1);						\
-	}								\
-})
-
 #define TEST_CGROUP "/test-bpf-sock-fields"
 #define DATA "Hello BPF!"
 #define DATA_LEN sizeof(DATA)
 
 static struct sockaddr_in6 srv_sa6, cli_sa6;
 static int sk_pkt_out_cnt10_fd;
+struct test_sock_fields *skel;
 static int sk_pkt_out_cnt_fd;
 static int linum_map_fd;
-static int addr_map_fd;
-static int tp_map_fd;
-static int sk_map_fd;
-
-static __u32 addr_srv_idx = ADDR_SRV_IDX;
-static __u32 addr_cli_idx = ADDR_CLI_IDX;
-
-static __u32 egress_srv_idx = EGRESS_SRV_IDX;
-static __u32 egress_cli_idx = EGRESS_CLI_IDX;
-static __u32 ingress_listen_idx = INGRESS_LISTEN_IDX;
+static __u32 duration;
 
 static __u32 egress_linum_idx = EGRESS_LINUM_IDX;
 static __u32 ingress_linum_idx = INGRESS_LINUM_IDX;
@@ -79,7 +50,7 @@ static void init_loopback6(struct sockaddr_in6 *sa6)
 	sa6->sin6_addr = in6addr_loopback;
 }
 
-static void print_sk(const struct bpf_sock *sk)
+static void print_sk(const struct bpf_sock *sk, const char *prefix)
 {
 	char src_ip4[24], dst_ip4[24];
 	char src_ip6[64], dst_ip6[64];
@@ -89,9 +60,10 @@ static void print_sk(const struct bpf_sock *sk)
 	inet_ntop(AF_INET, &sk->dst_ip4, dst_ip4, sizeof(dst_ip4));
 	inet_ntop(AF_INET6, &sk->dst_ip6, dst_ip6, sizeof(dst_ip6));
 
-	printf("state:%u bound_dev_if:%u family:%u type:%u protocol:%u mark:%u priority:%u "
+	printf("%s: state:%u bound_dev_if:%u family:%u type:%u protocol:%u mark:%u priority:%u "
 	       "src_ip4:%x(%s) src_ip6:%x:%x:%x:%x(%s) src_port:%u "
 	       "dst_ip4:%x(%s) dst_ip6:%x:%x:%x:%x(%s) dst_port:%u\n",
+	       prefix,
 	       sk->state, sk->bound_dev_if, sk->family, sk->type, sk->protocol,
 	       sk->mark, sk->priority,
 	       sk->src_ip4, src_ip4,
@@ -102,14 +74,15 @@ static void print_sk(const struct bpf_sock *sk)
 	       dst_ip6, ntohs(sk->dst_port));
 }
 
-static void print_tp(const struct bpf_tcp_sock *tp)
+static void print_tp(const struct bpf_tcp_sock *tp, const char *prefix)
 {
-	printf("snd_cwnd:%u srtt_us:%u rtt_min:%u snd_ssthresh:%u rcv_nxt:%u "
+	printf("%s: snd_cwnd:%u srtt_us:%u rtt_min:%u snd_ssthresh:%u rcv_nxt:%u "
 	       "snd_nxt:%u snd:una:%u mss_cache:%u ecn_flags:%u "
 	       "rate_delivered:%u rate_interval_us:%u packets_out:%u "
 	       "retrans_out:%u total_retrans:%u segs_in:%u data_segs_in:%u "
 	       "segs_out:%u data_segs_out:%u lost_out:%u sacked_out:%u "
 	       "bytes_received:%llu bytes_acked:%llu\n",
+	       prefix,
 	       tp->snd_cwnd, tp->srtt_us, tp->rtt_min, tp->snd_ssthresh,
 	       tp->rcv_nxt, tp->snd_nxt, tp->snd_una, tp->mss_cache,
 	       tp->ecn_flags, tp->rate_delivered, tp->rate_interval_us,
@@ -129,57 +102,26 @@ static void check_result(void)
 	err = bpf_map_lookup_elem(linum_map_fd, &egress_linum_idx,
 				  &egress_linum);
 	CHECK(err == -1, "bpf_map_lookup_elem(linum_map_fd)",
-	      "err:%d errno:%d", err, errno);
+	      "err:%d errno:%d\n", err, errno);
 
 	err = bpf_map_lookup_elem(linum_map_fd, &ingress_linum_idx,
 				  &ingress_linum);
 	CHECK(err == -1, "bpf_map_lookup_elem(linum_map_fd)",
-	      "err:%d errno:%d", err, errno);
-
-	err = bpf_map_lookup_elem(sk_map_fd, &egress_srv_idx, &srv_sk);
-	CHECK(err == -1, "bpf_map_lookup_elem(sk_map_fd, &egress_srv_idx)",
-	      "err:%d errno:%d", err, errno);
-	err = bpf_map_lookup_elem(tp_map_fd, &egress_srv_idx, &srv_tp);
-	CHECK(err == -1, "bpf_map_lookup_elem(tp_map_fd, &egress_srv_idx)",
-	      "err:%d errno:%d", err, errno);
-
-	err = bpf_map_lookup_elem(sk_map_fd, &egress_cli_idx, &cli_sk);
-	CHECK(err == -1, "bpf_map_lookup_elem(sk_map_fd, &egress_cli_idx)",
-	      "err:%d errno:%d", err, errno);
-	err = bpf_map_lookup_elem(tp_map_fd, &egress_cli_idx, &cli_tp);
-	CHECK(err == -1, "bpf_map_lookup_elem(tp_map_fd, &egress_cli_idx)",
-	      "err:%d errno:%d", err, errno);
-
-	err = bpf_map_lookup_elem(sk_map_fd, &ingress_listen_idx, &listen_sk);
-	CHECK(err == -1, "bpf_map_lookup_elem(sk_map_fd, &ingress_listen_idx)",
-	      "err:%d errno:%d", err, errno);
-	err = bpf_map_lookup_elem(tp_map_fd, &ingress_listen_idx, &listen_tp);
-	CHECK(err == -1, "bpf_map_lookup_elem(tp_map_fd, &ingress_listen_idx)",
-	      "err:%d errno:%d", err, errno);
-
-	printf("listen_sk: ");
-	print_sk(&listen_sk);
-	printf("\n");
-
-	printf("srv_sk: ");
-	print_sk(&srv_sk);
-	printf("\n");
-
-	printf("cli_sk: ");
-	print_sk(&cli_sk);
-	printf("\n");
-
-	printf("listen_tp: ");
-	print_tp(&listen_tp);
-	printf("\n");
-
-	printf("srv_tp: ");
-	print_tp(&srv_tp);
-	printf("\n");
-
-	printf("cli_tp: ");
-	print_tp(&cli_tp);
-	printf("\n");
+	      "err:%d errno:%d\n", err, errno);
+
+	memcpy(&srv_sk, &skel->bss->srv_sk, sizeof(srv_sk));
+	memcpy(&srv_tp, &skel->bss->srv_tp, sizeof(srv_tp));
+	memcpy(&cli_sk, &skel->bss->cli_sk, sizeof(cli_sk));
+	memcpy(&cli_tp, &skel->bss->cli_tp, sizeof(cli_tp));
+	memcpy(&listen_sk, &skel->bss->listen_sk, sizeof(listen_sk));
+	memcpy(&listen_tp, &skel->bss->listen_tp, sizeof(listen_tp));
+
+	print_sk(&listen_sk, "listen_sk");
+	print_sk(&srv_sk, "srv_sk");
+	print_sk(&cli_sk, "cli_sk");
+	print_tp(&listen_tp, "listen_tp");
+	print_tp(&srv_tp, "srv_tp");
+	print_tp(&cli_tp, "cli_tp");
 
 	CHECK(listen_sk.state != 10 ||
 	      listen_sk.family != AF_INET6 ||
@@ -190,8 +132,8 @@ static void check_result(void)
 	      listen_sk.dst_ip6[2] || listen_sk.dst_ip6[3] ||
 	      listen_sk.src_port != ntohs(srv_sa6.sin6_port) ||
 	      listen_sk.dst_port,
-	      "Unexpected listen_sk",
-	      "Check listen_sk output. ingress_linum:%u",
+	      "listen_sk",
+	      "Unexpected. Check listen_sk output. ingress_linum:%u\n",
 	      ingress_linum);
 
 	CHECK(srv_sk.state == 10 ||
@@ -204,7 +146,7 @@ static void check_result(void)
 		     sizeof(srv_sk.dst_ip6)) ||
 	      srv_sk.src_port != ntohs(srv_sa6.sin6_port) ||
 	      srv_sk.dst_port != cli_sa6.sin6_port,
-	      "Unexpected srv_sk", "Check srv_sk output. egress_linum:%u",
+	      "srv_sk", "Unexpected. Check srv_sk output. egress_linum:%u\n",
 	      egress_linum);
 
 	CHECK(cli_sk.state == 10 ||
@@ -217,30 +159,31 @@ static void check_result(void)
 		     sizeof(cli_sk.dst_ip6)) ||
 	      cli_sk.src_port != ntohs(cli_sa6.sin6_port) ||
 	      cli_sk.dst_port != srv_sa6.sin6_port,
-	      "Unexpected cli_sk", "Check cli_sk output. egress_linum:%u",
+	      "cli_sk", "Unexpected. Check cli_sk output. egress_linum:%u\n",
 	      egress_linum);
 
 	CHECK(listen_tp.data_segs_out ||
 	      listen_tp.data_segs_in ||
 	      listen_tp.total_retrans ||
 	      listen_tp.bytes_acked,
-	      "Unexpected listen_tp", "Check listen_tp output. ingress_linum:%u",
+	      "listen_tp",
+	      "Unexpected. Check listen_tp output. ingress_linum:%u\n",
 	      ingress_linum);
 
 	CHECK(srv_tp.data_segs_out != 2 ||
 	      srv_tp.data_segs_in ||
 	      srv_tp.snd_cwnd != 10 ||
 	      srv_tp.total_retrans ||
-	      srv_tp.bytes_acked != 2 * DATA_LEN,
-	      "Unexpected srv_tp", "Check srv_tp output. egress_linum:%u",
+	      srv_tp.bytes_acked < 2 * DATA_LEN,
+	      "srv_tp", "Unexpected. Check srv_tp output. egress_linum:%u\n",
 	      egress_linum);
 
 	CHECK(cli_tp.data_segs_out ||
 	      cli_tp.data_segs_in != 2 ||
 	      cli_tp.snd_cwnd != 10 ||
 	      cli_tp.total_retrans ||
-	      cli_tp.bytes_received != 2 * DATA_LEN,
-	      "Unexpected cli_tp", "Check cli_tp output. egress_linum:%u",
+	      cli_tp.bytes_received < 2 * DATA_LEN,
+	      "cli_tp", "Unexpected. Check cli_tp output. egress_linum:%u\n",
 	      egress_linum);
 }
 
@@ -257,15 +200,14 @@ static void check_sk_pkt_out_cnt(int accept_fd, int cli_fd)
 					  &pkt_out_cnt10);
 
 	/* The bpf prog only counts for fullsock and
-	 * passive conneciton did not become fullsock until 3WHS
-	 * had been finished.
-	 * The bpf prog only counted two data packet out but we
-	 * specially init accept_fd's pkt_out_cnt by 2 in
-	 * init_sk_storage().  Hence, 4 here.
+	 * passive connection did not become fullsock until 3WHS
+	 * had been finished, so the bpf prog only counted two data
+	 * packet out.
 	 */
-	CHECK(err || pkt_out_cnt.cnt != 4 || pkt_out_cnt10.cnt != 40,
+	CHECK(err || pkt_out_cnt.cnt < 0xeB9F + 2 ||
+	      pkt_out_cnt10.cnt < 0xeB9F + 20,
 	      "bpf_map_lookup_elem(sk_pkt_out_cnt, &accept_fd)",
-	      "err:%d errno:%d pkt_out_cnt:%u pkt_out_cnt10:%u",
+	      "err:%d errno:%d pkt_out_cnt:%u pkt_out_cnt10:%u\n",
 	      err, errno, pkt_out_cnt.cnt, pkt_out_cnt10.cnt);
 
 	pkt_out_cnt.cnt = ~0;
@@ -280,14 +222,14 @@ static void check_sk_pkt_out_cnt(int accept_fd, int cli_fd)
 	 *
 	 * The bpf_prog initialized it to 0xeB9F.
 	 */
-	CHECK(err || pkt_out_cnt.cnt != 0xeB9F + 4 ||
-	      pkt_out_cnt10.cnt != 0xeB9F + 40,
+	CHECK(err || pkt_out_cnt.cnt < 0xeB9F + 4 ||
+	      pkt_out_cnt10.cnt < 0xeB9F + 40,
 	      "bpf_map_lookup_elem(sk_pkt_out_cnt, &cli_fd)",
-	      "err:%d errno:%d pkt_out_cnt:%u pkt_out_cnt10:%u",
+	      "err:%d errno:%d pkt_out_cnt:%u pkt_out_cnt10:%u\n",
 	      err, errno, pkt_out_cnt.cnt, pkt_out_cnt10.cnt);
 }
 
-static void init_sk_storage(int sk_fd, __u32 pkt_out_cnt)
+static int init_sk_storage(int sk_fd, __u32 pkt_out_cnt)
 {
 	struct bpf_spinlock_cnt scnt = {};
 	int err;
@@ -295,186 +237,191 @@ static void init_sk_storage(int sk_fd, __u32 pkt_out_cnt)
 	scnt.cnt = pkt_out_cnt;
 	err = bpf_map_update_elem(sk_pkt_out_cnt_fd, &sk_fd, &scnt,
 				  BPF_NOEXIST);
-	CHECK(err, "bpf_map_update_elem(sk_pkt_out_cnt_fd)",
-	      "err:%d errno:%d", err, errno);
+	if (CHECK(err, "bpf_map_update_elem(sk_pkt_out_cnt_fd)",
+		  "err:%d errno:%d\n", err, errno))
+		return err;
 
 	scnt.cnt *= 10;
 	err = bpf_map_update_elem(sk_pkt_out_cnt10_fd, &sk_fd, &scnt,
 				  BPF_NOEXIST);
-	CHECK(err, "bpf_map_update_elem(sk_pkt_out_cnt10_fd)",
-	      "err:%d errno:%d", err, errno);
+	if (CHECK(err, "bpf_map_update_elem(sk_pkt_out_cnt10_fd)",
+		  "err:%d errno:%d\n", err, errno))
+		return err;
+
+	return 0;
 }
 
-static void test(void)
+static void do_test(void)
 {
-	int listen_fd, cli_fd, accept_fd, epfd, err;
+	int listen_fd = -1, cli_fd = -1, accept_fd = -1, epfd, err, i;
 	struct epoll_event ev;
 	socklen_t addrlen;
-	int i;
+	char buf[DATA_LEN];
 
 	addrlen = sizeof(struct sockaddr_in6);
 	ev.events = EPOLLIN;
 
 	epfd = epoll_create(1);
-	CHECK(epfd == -1, "epoll_create()", "epfd:%d errno:%d", epfd, errno);
+	if (CHECK(epfd == -1, "epoll_create()", "epfd:%d errno:%d\n",
+		  epfd, errno))
+		return;
 
 	/* Prepare listen_fd */
 	listen_fd = socket(AF_INET6, SOCK_STREAM | SOCK_NONBLOCK, 0);
-	CHECK(listen_fd == -1, "socket()", "listen_fd:%d errno:%d",
-	      listen_fd, errno);
+	if (CHECK(listen_fd == -1, "socket()", "listen_fd:%d errno:%d\n",
+		  listen_fd, errno))
+		goto done;
 
 	init_loopback6(&srv_sa6);
 	err = bind(listen_fd, (struct sockaddr *)&srv_sa6, sizeof(srv_sa6));
-	CHECK(err, "bind(listen_fd)", "err:%d errno:%d", err, errno);
+	if (CHECK(err, "bind(listen_fd)", "err:%d errno:%d\n", err, errno))
+		goto done;
 
 	err = getsockname(listen_fd, (struct sockaddr *)&srv_sa6, &addrlen);
-	CHECK(err, "getsockname(listen_fd)", "err:%d errno:%d", err, errno);
+	if (CHECK(err, "getsockname(listen_fd)", "err:%d errno:%d\n", err,
+		  errno))
+		goto done;
+	memcpy(&skel->bss->srv_sa6, &srv_sa6, sizeof(srv_sa6));
 
 	err = listen(listen_fd, 1);
-	CHECK(err, "listen(listen_fd)", "err:%d errno:%d", err, errno);
+	if (CHECK(err, "listen(listen_fd)", "err:%d errno:%d\n", err, errno))
+		goto done;
 
 	/* Prepare cli_fd */
 	cli_fd = socket(AF_INET6, SOCK_STREAM | SOCK_NONBLOCK, 0);
-	CHECK(cli_fd == -1, "socket()", "cli_fd:%d errno:%d", cli_fd, errno);
+	if (CHECK(cli_fd == -1, "socket()", "cli_fd:%d errno:%d\n", cli_fd,
+		  errno))
+		goto done;
 
 	init_loopback6(&cli_sa6);
 	err = bind(cli_fd, (struct sockaddr *)&cli_sa6, sizeof(cli_sa6));
-	CHECK(err, "bind(cli_fd)", "err:%d errno:%d", err, errno);
+	if (CHECK(err, "bind(cli_fd)", "err:%d errno:%d\n", err, errno))
+		goto done;
 
 	err = getsockname(cli_fd, (struct sockaddr *)&cli_sa6, &addrlen);
-	CHECK(err, "getsockname(cli_fd)", "err:%d errno:%d",
-	      err, errno);
-
-	/* Update addr_map with srv_sa6 and cli_sa6 */
-	err = bpf_map_update_elem(addr_map_fd, &addr_srv_idx, &srv_sa6, 0);
-	CHECK(err, "map_update", "err:%d errno:%d", err, errno);
-
-	err = bpf_map_update_elem(addr_map_fd, &addr_cli_idx, &cli_sa6, 0);
-	CHECK(err, "map_update", "err:%d errno:%d", err, errno);
+	if (CHECK(err, "getsockname(cli_fd)", "err:%d errno:%d\n",
+		  err, errno))
+		goto done;
 
 	/* Connect from cli_sa6 to srv_sa6 */
 	err = connect(cli_fd, (struct sockaddr *)&srv_sa6, addrlen);
 	printf("srv_sa6.sin6_port:%u cli_sa6.sin6_port:%u\n\n",
 	       ntohs(srv_sa6.sin6_port), ntohs(cli_sa6.sin6_port));
-	CHECK(err && errno != EINPROGRESS,
-	      "connect(cli_fd)", "err:%d errno:%d", err, errno);
+	if (CHECK(err && errno != EINPROGRESS,
+		  "connect(cli_fd)", "err:%d errno:%d\n", err, errno))
+		goto done;
 
 	ev.data.fd = listen_fd;
 	err = epoll_ctl(epfd, EPOLL_CTL_ADD, listen_fd, &ev);
-	CHECK(err, "epoll_ctl(EPOLL_CTL_ADD, listen_fd)", "err:%d errno:%d",
-	      err, errno);
+	if (CHECK(err, "epoll_ctl(EPOLL_CTL_ADD, listen_fd)",
+		  "err:%d errno:%d\n", err, errno))
+		goto done;
 
 	/* Accept the connection */
 	/* Have some timeout in accept(listen_fd). Just in case. */
 	err = epoll_wait(epfd, &ev, 1, 1000);
-	CHECK(err != 1 || ev.data.fd != listen_fd,
-	      "epoll_wait(listen_fd)",
-	      "err:%d errno:%d ev.data.fd:%d listen_fd:%d",
-	      err, errno, ev.data.fd, listen_fd);
+	if (CHECK(err != 1 || ev.data.fd != listen_fd,
+		  "epoll_wait(listen_fd)",
+		  "err:%d errno:%d ev.data.fd:%d listen_fd:%d\n",
+		  err, errno, ev.data.fd, listen_fd))
+		goto done;
 
 	accept_fd = accept(listen_fd, NULL, NULL);
-	CHECK(accept_fd == -1, "accept(listen_fd)", "accept_fd:%d errno:%d",
-	      accept_fd, errno);
-	close(listen_fd);
+	if (CHECK(accept_fd == -1, "accept(listen_fd)",
+		  "accept_fd:%d errno:%d\n",
+		  accept_fd, errno))
+		goto done;
 
 	ev.data.fd = cli_fd;
 	err = epoll_ctl(epfd, EPOLL_CTL_ADD, cli_fd, &ev);
-	CHECK(err, "epoll_ctl(EPOLL_CTL_ADD, cli_fd)", "err:%d errno:%d",
-	      err, errno);
+	if (CHECK(err, "epoll_ctl(EPOLL_CTL_ADD, cli_fd)",
+		  "err:%d errno:%d\n", err, errno))
+		goto done;
 
-	init_sk_storage(accept_fd, 2);
+	if (init_sk_storage(accept_fd, 0xeB9F))
+		goto done;
 
 	for (i = 0; i < 2; i++) {
-		/* Send some data from accept_fd to cli_fd */
-		err = send(accept_fd, DATA, DATA_LEN, 0);
-		CHECK(err != DATA_LEN, "send(accept_fd)", "err:%d errno:%d",
-		      err, errno);
+		/* Send some data from accept_fd to cli_fd.
+		 * MSG_EOR to stop kernel from coalescing two pkts.
+		 */
+		err = send(accept_fd, DATA, DATA_LEN, MSG_EOR);
+		if (CHECK(err != DATA_LEN, "send(accept_fd)",
+			  "err:%d errno:%d\n", err, errno))
+			goto done;
 
 		/* Have some timeout in recv(cli_fd). Just in case. */
 		err = epoll_wait(epfd, &ev, 1, 1000);
-		CHECK(err != 1 || ev.data.fd != cli_fd,
-		      "epoll_wait(cli_fd)", "err:%d errno:%d ev.data.fd:%d cli_fd:%d",
-		      err, errno, ev.data.fd, cli_fd);
-
-		err = recv(cli_fd, NULL, 0, MSG_TRUNC);
-		CHECK(err, "recv(cli_fd)", "err:%d errno:%d", err, errno);
+		if (CHECK(err != 1 || ev.data.fd != cli_fd,
+			  "epoll_wait(cli_fd)",
+			  "err:%d errno:%d ev.data.fd:%d cli_fd:%d\n",
+			  err, errno, ev.data.fd, cli_fd))
+			goto done;
+
+		err = recv(cli_fd, buf, DATA_LEN, 0);
+		if (CHECK(err != DATA_LEN, "recv(cli_fd)", "err:%d errno:%d\n",
+			  err, errno))
+			goto done;
 	}
 
+	shutdown(cli_fd, SHUT_WR);
+	err = recv(accept_fd, buf, 1, 0);
+	if (CHECK(err, "recv(accept_fd) for fin", "err:%d errno:%d\n",
+		  err, errno))
+		goto done;
+	shutdown(accept_fd, SHUT_WR);
+	err = recv(cli_fd, buf, 1, 0);
+	if (CHECK(err, "recv(cli_fd) for fin", "err:%d errno:%d\n",
+		  err, errno))
+		goto done;
 	check_sk_pkt_out_cnt(accept_fd, cli_fd);
+	check_result();
 
+done:
+	if (accept_fd != -1)
+		close(accept_fd);
+	if (cli_fd != -1)
+		close(cli_fd);
+	if (listen_fd != -1)
+		close(listen_fd);
 	close(epfd);
-	close(accept_fd);
-	close(cli_fd);
-
-	check_result();
 }
 
 void test_sock_fields(void)
 {
-	struct bpf_prog_load_attr attr = {
-		.file = "test_sock_fields.o",
-		.prog_type = BPF_PROG_TYPE_CGROUP_SKB,
-		.prog_flags = BPF_F_TEST_RND_HI32,
-	};
-	int cgroup_fd, egress_fd, ingress_fd, err;
-	struct bpf_program *ingress_prog;
-	struct bpf_object *obj;
-	struct bpf_map *map;
+	struct bpf_link *egress_link = NULL, *ingress_link = NULL;
+	int cgroup_fd;
 
 	/* Create a cgroup, get fd, and join it */
-	cgroup_fd = cgroup_setup_and_join(TEST_CGROUP);
-	CHECK(cgroup_fd < 0, "cgroup_setup_and_join()",
-	      "cgroup_fd:%d errno:%d", cgroup_fd, errno);
-	atexit(cleanup_cgroup_environment);
-
-	err = bpf_prog_load_xattr(&attr, &obj, &egress_fd);
-	CHECK(err, "bpf_prog_load_xattr()", "err:%d", err);
-
-	ingress_prog = bpf_object__find_program_by_title(obj,
-							 "cgroup_skb/ingress");
-	CHECK(!ingress_prog,
-	      "bpf_object__find_program_by_title(cgroup_skb/ingress)",
-	      "not found");
-	ingress_fd = bpf_program__fd(ingress_prog);
-
-	err = bpf_prog_attach(egress_fd, cgroup_fd, BPF_CGROUP_INET_EGRESS, 0);
-	CHECK(err == -1, "bpf_prog_attach(CPF_CGROUP_INET_EGRESS)",
-	      "err:%d errno%d", err, errno);
-
-	err = bpf_prog_attach(ingress_fd, cgroup_fd,
-			      BPF_CGROUP_INET_INGRESS, 0);
-	CHECK(err == -1, "bpf_prog_attach(CPF_CGROUP_INET_INGRESS)",
-	      "err:%d errno%d", err, errno);
+	cgroup_fd = test__join_cgroup(TEST_CGROUP);
+	if (CHECK_FAIL(cgroup_fd < 0))
+		return;
+
+	skel = test_sock_fields__open_and_load();
+	if (CHECK(!skel, "test_sock_fields__open_and_load", "failed\n"))
+		goto done;
+
+	egress_link = bpf_program__attach_cgroup(skel->progs.egress_read_sock_fields,
+						 cgroup_fd);
+	if (CHECK(IS_ERR(egress_link), "attach_cgroup(egress)", "err:%ld\n",
+		  PTR_ERR(egress_link)))
+		goto done;
+
+	ingress_link = bpf_program__attach_cgroup(skel->progs.ingress_read_sock_fields,
+						  cgroup_fd);
+	if (CHECK(IS_ERR(ingress_link), "attach_cgroup(ingress)", "err:%ld\n",
+		  PTR_ERR(ingress_link)))
+		goto done;
+
+	linum_map_fd = bpf_map__fd(skel->maps.linum_map);
+	sk_pkt_out_cnt_fd = bpf_map__fd(skel->maps.sk_pkt_out_cnt);
+	sk_pkt_out_cnt10_fd = bpf_map__fd(skel->maps.sk_pkt_out_cnt10);
+
+	do_test();
+
+done:
+	bpf_link__destroy(egress_link);
+	bpf_link__destroy(ingress_link);
+	test_sock_fields__destroy(skel);
 	close(cgroup_fd);
-
-	map = bpf_object__find_map_by_name(obj, "addr_map");
-	CHECK(!map, "cannot find addr_map", "(null)");
-	addr_map_fd = bpf_map__fd(map);
-
-	map = bpf_object__find_map_by_name(obj, "sock_result_map");
-	CHECK(!map, "cannot find sock_result_map", "(null)");
-	sk_map_fd = bpf_map__fd(map);
-
-	map = bpf_object__find_map_by_name(obj, "tcp_sock_result_map");
-	CHECK(!map, "cannot find tcp_sock_result_map", "(null)");
-	tp_map_fd = bpf_map__fd(map);
-
-	map = bpf_object__find_map_by_name(obj, "linum_map");
-	CHECK(!map, "cannot find linum_map", "(null)");
-	linum_map_fd = bpf_map__fd(map);
-
-	map = bpf_object__find_map_by_name(obj, "sk_pkt_out_cnt");
-	CHECK(!map, "cannot find sk_pkt_out_cnt", "(null)");
-	sk_pkt_out_cnt_fd = bpf_map__fd(map);
-
-	map = bpf_object__find_map_by_name(obj, "sk_pkt_out_cnt10");
-	CHECK(!map, "cannot find sk_pkt_out_cnt10", "(null)");
-	sk_pkt_out_cnt10_fd = bpf_map__fd(map);
-
-	test();
-
-	bpf_object__close(obj);
-	cleanup_cgroup_environment();
-
-	printf("PASS\n");
 }
diff --git a/tools/testing/selftests/bpf/progs/test_sock_fields.c b/tools/testing/selftests/bpf/progs/test_sock_fields.c
index 9bcaa37f476a..370e33a858db 100644
--- a/tools/testing/selftests/bpf/progs/test_sock_fields.c
+++ b/tools/testing/selftests/bpf/progs/test_sock_fields.c
@@ -8,46 +8,12 @@
 #include <bpf/bpf_helpers.h>
 #include <bpf/bpf_endian.h>
 
-enum bpf_addr_array_idx {
-	ADDR_SRV_IDX,
-	ADDR_CLI_IDX,
-	__NR_BPF_ADDR_ARRAY_IDX,
-};
-
-enum bpf_result_array_idx {
-	EGRESS_SRV_IDX,
-	EGRESS_CLI_IDX,
-	INGRESS_LISTEN_IDX,
-	__NR_BPF_RESULT_ARRAY_IDX,
-};
-
 enum bpf_linum_array_idx {
 	EGRESS_LINUM_IDX,
 	INGRESS_LINUM_IDX,
 	__NR_BPF_LINUM_ARRAY_IDX,
 };
 
-struct {
-	__uint(type, BPF_MAP_TYPE_ARRAY);
-	__uint(max_entries, __NR_BPF_ADDR_ARRAY_IDX);
-	__type(key, __u32);
-	__type(value, struct sockaddr_in6);
-} addr_map SEC(".maps");
-
-struct {
-	__uint(type, BPF_MAP_TYPE_ARRAY);
-	__uint(max_entries, __NR_BPF_RESULT_ARRAY_IDX);
-	__type(key, __u32);
-	__type(value, struct bpf_sock);
-} sock_result_map SEC(".maps");
-
-struct {
-	__uint(type, BPF_MAP_TYPE_ARRAY);
-	__uint(max_entries, __NR_BPF_RESULT_ARRAY_IDX);
-	__type(key, __u32);
-	__type(value, struct bpf_tcp_sock);
-} tcp_sock_result_map SEC(".maps");
-
 struct {
 	__uint(type, BPF_MAP_TYPE_ARRAY);
 	__uint(max_entries, __NR_BPF_LINUM_ARRAY_IDX);
@@ -74,6 +40,14 @@ struct {
 	__type(value, struct bpf_spinlock_cnt);
 } sk_pkt_out_cnt10 SEC(".maps");
 
+struct bpf_tcp_sock listen_tp = {};
+struct sockaddr_in6 srv_sa6 = {};
+struct bpf_tcp_sock cli_tp = {};
+struct bpf_tcp_sock srv_tp = {};
+struct bpf_sock listen_sk = {};
+struct bpf_sock srv_sk = {};
+struct bpf_sock cli_sk = {};
+
 static bool is_loopback6(__u32 *a6)
 {
 	return !a6[0] && !a6[1] && !a6[2] && a6[3] == bpf_htonl(1);
@@ -130,19 +104,20 @@ static void tpcpy(struct bpf_tcp_sock *dst,
 	dst->bytes_acked = src->bytes_acked;
 }
 
-#define RETURN {						\
+/* Always return CG_OK so that no pkt will be filtered out */
+#define CG_OK 1
+
+#define RET_LOG() ({						\
 	linum = __LINE__;					\
-	bpf_map_update_elem(&linum_map, &linum_idx, &linum, 0);	\
-	return 1;						\
-}
+	bpf_map_update_elem(&linum_map, &linum_idx, &linum, BPF_NOEXIST);	\
+	return CG_OK;						\
+})
 
 SEC("cgroup_skb/egress")
 int egress_read_sock_fields(struct __sk_buff *skb)
 {
 	struct bpf_spinlock_cnt cli_cnt_init = { .lock = 0, .cnt = 0xeB9F };
-	__u32 srv_idx = ADDR_SRV_IDX, cli_idx = ADDR_CLI_IDX, result_idx;
 	struct bpf_spinlock_cnt *pkt_out_cnt, *pkt_out_cnt10;
-	struct sockaddr_in6 *srv_sa6, *cli_sa6;
 	struct bpf_tcp_sock *tp, *tp_ret;
 	struct bpf_sock *sk, *sk_ret;
 	__u32 linum, linum_idx;
@@ -150,39 +125,46 @@ int egress_read_sock_fields(struct __sk_buff *skb)
 	linum_idx = EGRESS_LINUM_IDX;
 
 	sk = skb->sk;
-	if (!sk || sk->state == 10)
-		RETURN;
+	if (!sk)
+		RET_LOG();
 
+	/* Not the testing egress traffic or
+	 * TCP_LISTEN (10) socket will be copied at the ingress side.
+	 */
+	if (sk->family != AF_INET6 || !is_loopback6(sk->src_ip6) ||
+	    sk->state == 10)
+		return CG_OK;
+
+	if (sk->src_port == bpf_ntohs(srv_sa6.sin6_port)) {
+		/* Server socket */
+		sk_ret = &srv_sk;
+		tp_ret = &srv_tp;
+	} else if (sk->dst_port == srv_sa6.sin6_port) {
+		/* Client socket */
+		sk_ret = &cli_sk;
+		tp_ret = &cli_tp;
+	} else {
+		/* Not the testing egress traffic */
+		return CG_OK;
+	}
+
+	/* It must be a fullsock for cgroup_skb/egress prog */
 	sk = bpf_sk_fullsock(sk);
-	if (!sk || sk->family != AF_INET6 || sk->protocol != IPPROTO_TCP ||
-	    !is_loopback6(sk->src_ip6))
-		RETURN;
+	if (!sk)
+		RET_LOG();
+
+	/* Not the testing egress traffic */
+	if (sk->protocol != IPPROTO_TCP)
+		return CG_OK;
 
 	tp = bpf_tcp_sock(sk);
 	if (!tp)
-		RETURN;
-
-	srv_sa6 = bpf_map_lookup_elem(&addr_map, &srv_idx);
-	cli_sa6 = bpf_map_lookup_elem(&addr_map, &cli_idx);
-	if (!srv_sa6 || !cli_sa6)
-		RETURN;
-
-	if (sk->src_port == bpf_ntohs(srv_sa6->sin6_port))
-		result_idx = EGRESS_SRV_IDX;
-	else if (sk->src_port == bpf_ntohs(cli_sa6->sin6_port))
-		result_idx = EGRESS_CLI_IDX;
-	else
-		RETURN;
-
-	sk_ret = bpf_map_lookup_elem(&sock_result_map, &result_idx);
-	tp_ret = bpf_map_lookup_elem(&tcp_sock_result_map, &result_idx);
-	if (!sk_ret || !tp_ret)
-		RETURN;
+		RET_LOG();
 
 	skcpy(sk_ret, sk);
 	tpcpy(tp_ret, tp);
 
-	if (result_idx == EGRESS_SRV_IDX) {
+	if (sk_ret == &srv_sk) {
 		/* The userspace has created it for srv sk */
 		pkt_out_cnt = bpf_sk_storage_get(&sk_pkt_out_cnt, sk, 0, 0);
 		pkt_out_cnt10 = bpf_sk_storage_get(&sk_pkt_out_cnt10, sk,
@@ -197,7 +179,7 @@ int egress_read_sock_fields(struct __sk_buff *skb)
 	}
 
 	if (!pkt_out_cnt || !pkt_out_cnt10)
-		RETURN;
+		RET_LOG();
 
 	/* Even both cnt and cnt10 have lock defined in their BTF,
 	 * intentionally one cnt takes lock while one does not
@@ -208,48 +190,44 @@ int egress_read_sock_fields(struct __sk_buff *skb)
 	pkt_out_cnt10->cnt += 10;
 	bpf_spin_unlock(&pkt_out_cnt10->lock);
 
-	RETURN;
+	return CG_OK;
 }
 
 SEC("cgroup_skb/ingress")
 int ingress_read_sock_fields(struct __sk_buff *skb)
 {
-	__u32 srv_idx = ADDR_SRV_IDX, result_idx = INGRESS_LISTEN_IDX;
-	struct bpf_tcp_sock *tp, *tp_ret;
-	struct bpf_sock *sk, *sk_ret;
-	struct sockaddr_in6 *srv_sa6;
+	struct bpf_tcp_sock *tp;
 	__u32 linum, linum_idx;
+	struct bpf_sock *sk;
 
 	linum_idx = INGRESS_LINUM_IDX;
 
 	sk = skb->sk;
-	if (!sk || sk->family != AF_INET6 || !is_loopback6(sk->src_ip6))
-		RETURN;
+	if (!sk)
+		RET_LOG();
 
-	srv_sa6 = bpf_map_lookup_elem(&addr_map, &srv_idx);
-	if (!srv_sa6 || sk->src_port != bpf_ntohs(srv_sa6->sin6_port))
-		RETURN;
+	/* Not the testing ingress traffic to the server */
+	if (sk->family != AF_INET6 || !is_loopback6(sk->src_ip6) ||
+	    sk->src_port != bpf_ntohs(srv_sa6.sin6_port))
+		return CG_OK;
 
-	if (sk->state != 10 && sk->state != 12)
-		RETURN;
+	/* Only interested in TCP_LISTEN */
+	if (sk->state != 10)
+		return CG_OK;
 
-	sk = bpf_get_listener_sock(sk);
+	/* It must be a fullsock for cgroup_skb/ingress prog */
+	sk = bpf_sk_fullsock(sk);
 	if (!sk)
-		RETURN;
+		RET_LOG();
 
 	tp = bpf_tcp_sock(sk);
 	if (!tp)
-		RETURN;
-
-	sk_ret = bpf_map_lookup_elem(&sock_result_map, &result_idx);
-	tp_ret = bpf_map_lookup_elem(&tcp_sock_result_map, &result_idx);
-	if (!sk_ret || !tp_ret)
-		RETURN;
+		RET_LOG();
 
-	skcpy(sk_ret, sk);
-	tpcpy(tp_ret, tp);
+	skcpy(&listen_sk, sk);
+	tpcpy(&listen_tp, tp);
 
-	RETURN;
+	return CG_OK;
 }
 
 char _license[] SEC("license") = "GPL";
-- 
2.24.1

[PATCH v3 bpf-next 10/11] bpf: selftest: Use network_helpers in the sock_fields test

[Martin KaFai Lau]

This patch uses start_server() and connect_to_fd() from network_helpers.h
to remove the network testing boiler plate codes.  epoll is no longer
needed also since the timeout has already been taken care of also.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
---
 .../selftests/bpf/prog_tests/sock_fields.c    | 86 ++-----------------
 1 file changed, 9 insertions(+), 77 deletions(-)

diff --git a/tools/testing/selftests/bpf/prog_tests/sock_fields.c b/tools/testing/selftests/bpf/prog_tests/sock_fields.c
index cb6d41bc79a4..23d14e2d0d28 100644
--- a/tools/testing/selftests/bpf/prog_tests/sock_fields.c
+++ b/tools/testing/selftests/bpf/prog_tests/sock_fields.c
@@ -12,7 +12,9 @@
 
 #include <bpf/bpf.h>
 #include <bpf/libbpf.h>
+#include <linux/compiler.h>
 
+#include "network_helpers.h"
 #include "cgroup_helpers.h"
 #include "test_progs.h"
 #include "bpf_rlimit.h"
@@ -43,13 +45,6 @@ static __u32 duration;
 static __u32 egress_linum_idx = EGRESS_LINUM_IDX;
 static __u32 ingress_linum_idx = INGRESS_LINUM_IDX;
 
-static void init_loopback6(struct sockaddr_in6 *sa6)
-{
-	memset(sa6, 0, sizeof(*sa6));
-	sa6->sin6_family = AF_INET6;
-	sa6->sin6_addr = in6addr_loopback;
-}
-
 static void print_sk(const struct bpf_sock *sk, const char *prefix)
 {
 	char src_ip4[24], dst_ip4[24];
@@ -253,28 +248,14 @@ static int init_sk_storage(int sk_fd, __u32 pkt_out_cnt)
 
 static void do_test(void)
 {
-	int listen_fd = -1, cli_fd = -1, accept_fd = -1, epfd, err, i;
-	struct epoll_event ev;
-	socklen_t addrlen;
+	int listen_fd = -1, cli_fd = -1, accept_fd = -1, err, i;
+	socklen_t addrlen = sizeof(struct sockaddr_in6);
 	char buf[DATA_LEN];
 
-	addrlen = sizeof(struct sockaddr_in6);
-	ev.events = EPOLLIN;
-
-	epfd = epoll_create(1);
-	if (CHECK(epfd == -1, "epoll_create()", "epfd:%d errno:%d\n",
-		  epfd, errno))
-		return;
-
 	/* Prepare listen_fd */
-	listen_fd = socket(AF_INET6, SOCK_STREAM | SOCK_NONBLOCK, 0);
-	if (CHECK(listen_fd == -1, "socket()", "listen_fd:%d errno:%d\n",
-		  listen_fd, errno))
-		goto done;
-
-	init_loopback6(&srv_sa6);
-	err = bind(listen_fd, (struct sockaddr *)&srv_sa6, sizeof(srv_sa6));
-	if (CHECK(err, "bind(listen_fd)", "err:%d errno:%d\n", err, errno))
+	listen_fd = start_server(AF_INET6, SOCK_STREAM, "::1", 0, 0);
+	/* start_server() has logged the error details */
+	if (CHECK_FAIL(listen_fd == -1))
 		goto done;
 
 	err = getsockname(listen_fd, (struct sockaddr *)&srv_sa6, &addrlen);
@@ -283,19 +264,8 @@ static void do_test(void)
 		goto done;
 	memcpy(&skel->bss->srv_sa6, &srv_sa6, sizeof(srv_sa6));
 
-	err = listen(listen_fd, 1);
-	if (CHECK(err, "listen(listen_fd)", "err:%d errno:%d\n", err, errno))
-		goto done;
-
-	/* Prepare cli_fd */
-	cli_fd = socket(AF_INET6, SOCK_STREAM | SOCK_NONBLOCK, 0);
-	if (CHECK(cli_fd == -1, "socket()", "cli_fd:%d errno:%d\n", cli_fd,
-		  errno))
-		goto done;
-
-	init_loopback6(&cli_sa6);
-	err = bind(cli_fd, (struct sockaddr *)&cli_sa6, sizeof(cli_sa6));
-	if (CHECK(err, "bind(cli_fd)", "err:%d errno:%d\n", err, errno))
+	cli_fd = connect_to_fd(listen_fd, 0);
+	if (CHECK_FAIL(cli_fd == -1))
 		goto done;
 
 	err = getsockname(cli_fd, (struct sockaddr *)&cli_sa6, &addrlen);
@@ -303,41 +273,12 @@ static void do_test(void)
 		  err, errno))
 		goto done;
 
-	/* Connect from cli_sa6 to srv_sa6 */
-	err = connect(cli_fd, (struct sockaddr *)&srv_sa6, addrlen);
-	printf("srv_sa6.sin6_port:%u cli_sa6.sin6_port:%u\n\n",
-	       ntohs(srv_sa6.sin6_port), ntohs(cli_sa6.sin6_port));
-	if (CHECK(err && errno != EINPROGRESS,
-		  "connect(cli_fd)", "err:%d errno:%d\n", err, errno))
-		goto done;
-
-	ev.data.fd = listen_fd;
-	err = epoll_ctl(epfd, EPOLL_CTL_ADD, listen_fd, &ev);
-	if (CHECK(err, "epoll_ctl(EPOLL_CTL_ADD, listen_fd)",
-		  "err:%d errno:%d\n", err, errno))
-		goto done;
-
-	/* Accept the connection */
-	/* Have some timeout in accept(listen_fd). Just in case. */
-	err = epoll_wait(epfd, &ev, 1, 1000);
-	if (CHECK(err != 1 || ev.data.fd != listen_fd,
-		  "epoll_wait(listen_fd)",
-		  "err:%d errno:%d ev.data.fd:%d listen_fd:%d\n",
-		  err, errno, ev.data.fd, listen_fd))
-		goto done;
-
 	accept_fd = accept(listen_fd, NULL, NULL);
 	if (CHECK(accept_fd == -1, "accept(listen_fd)",
 		  "accept_fd:%d errno:%d\n",
 		  accept_fd, errno))
 		goto done;
 
-	ev.data.fd = cli_fd;
-	err = epoll_ctl(epfd, EPOLL_CTL_ADD, cli_fd, &ev);
-	if (CHECK(err, "epoll_ctl(EPOLL_CTL_ADD, cli_fd)",
-		  "err:%d errno:%d\n", err, errno))
-		goto done;
-
 	if (init_sk_storage(accept_fd, 0xeB9F))
 		goto done;
 
@@ -350,14 +291,6 @@ static void do_test(void)
 			  "err:%d errno:%d\n", err, errno))
 			goto done;
 
-		/* Have some timeout in recv(cli_fd). Just in case. */
-		err = epoll_wait(epfd, &ev, 1, 1000);
-		if (CHECK(err != 1 || ev.data.fd != cli_fd,
-			  "epoll_wait(cli_fd)",
-			  "err:%d errno:%d ev.data.fd:%d cli_fd:%d\n",
-			  err, errno, ev.data.fd, cli_fd))
-			goto done;
-
 		err = recv(cli_fd, buf, DATA_LEN, 0);
 		if (CHECK(err != DATA_LEN, "recv(cli_fd)", "err:%d errno:%d\n",
 			  err, errno))
@@ -384,7 +317,6 @@ static void do_test(void)
 		close(cli_fd);
 	if (listen_fd != -1)
 		close(listen_fd);
-	close(epfd);
 }
 
 void test_sock_fields(void)
-- 
2.24.1

[PATCH v3 bpf-next 11/11] bpf: selftest: Use bpf_skc_to_tcp_sock() in the sock_fields test

[Martin KaFai Lau]

This test uses bpf_skc_to_tcp_sock() to get a kernel tcp_sock ptr "ktp".
Access the ktp->lsndtime and also pass it to bpf_sk_storage_get().

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
---
 tools/testing/selftests/bpf/prog_tests/sock_fields.c |  2 ++
 tools/testing/selftests/bpf/progs/test_sock_fields.c | 11 +++++++++--
 2 files changed, 11 insertions(+), 2 deletions(-)

diff --git a/tools/testing/selftests/bpf/prog_tests/sock_fields.c b/tools/testing/selftests/bpf/prog_tests/sock_fields.c
index 23d14e2d0d28..9b81600a1ef4 100644
--- a/tools/testing/selftests/bpf/prog_tests/sock_fields.c
+++ b/tools/testing/selftests/bpf/prog_tests/sock_fields.c
@@ -144,6 +144,8 @@ static void check_result(void)
 	      "srv_sk", "Unexpected. Check srv_sk output. egress_linum:%u\n",
 	      egress_linum);
 
+	CHECK(!skel->bss->lsndtime, "srv_tp", "Unexpected lsndtime:0\n");
+
 	CHECK(cli_sk.state == 10 ||
 	      !cli_sk.state ||
 	      cli_sk.family != AF_INET6 ||
diff --git a/tools/testing/selftests/bpf/progs/test_sock_fields.c b/tools/testing/selftests/bpf/progs/test_sock_fields.c
index 370e33a858db..24fdf2b2747e 100644
--- a/tools/testing/selftests/bpf/progs/test_sock_fields.c
+++ b/tools/testing/selftests/bpf/progs/test_sock_fields.c
@@ -7,6 +7,7 @@
 
 #include <bpf/bpf_helpers.h>
 #include <bpf/bpf_endian.h>
+#include "bpf_tcp_helpers.h"
 
 enum bpf_linum_array_idx {
 	EGRESS_LINUM_IDX,
@@ -47,6 +48,7 @@ struct bpf_tcp_sock srv_tp = {};
 struct bpf_sock listen_sk = {};
 struct bpf_sock srv_sk = {};
 struct bpf_sock cli_sk = {};
+__u64 lsndtime = 0;
 
 static bool is_loopback6(__u32 *a6)
 {
@@ -121,6 +123,7 @@ int egress_read_sock_fields(struct __sk_buff *skb)
 	struct bpf_tcp_sock *tp, *tp_ret;
 	struct bpf_sock *sk, *sk_ret;
 	__u32 linum, linum_idx;
+	struct tcp_sock *ktp;
 
 	linum_idx = EGRESS_LINUM_IDX;
 
@@ -165,9 +168,13 @@ int egress_read_sock_fields(struct __sk_buff *skb)
 	tpcpy(tp_ret, tp);
 
 	if (sk_ret == &srv_sk) {
+		ktp = bpf_skc_to_tcp_sock(sk);
+		if (!ktp)
+			RET_LOG();
+		lsndtime = ktp->lsndtime;
 		/* The userspace has created it for srv sk */
-		pkt_out_cnt = bpf_sk_storage_get(&sk_pkt_out_cnt, sk, 0, 0);
-		pkt_out_cnt10 = bpf_sk_storage_get(&sk_pkt_out_cnt10, sk,
+		pkt_out_cnt = bpf_sk_storage_get(&sk_pkt_out_cnt, ktp, 0, 0);
+		pkt_out_cnt10 = bpf_sk_storage_get(&sk_pkt_out_cnt10, ktp,
 						   0, 0);
 	} else {
 		pkt_out_cnt = bpf_sk_storage_get(&sk_pkt_out_cnt, sk,
-- 
2.24.1

Re: [PATCH v3 bpf-next 11/11] bpf: selftest: Use bpf_skc_to_tcp_sock() in the sock_fields test

[Alexei Starovoitov]

On Tue, Sep 22, 2020 at 12:05:18AM -0700, Martin KaFai Lau wrote:
>  
> @@ -165,9 +168,13 @@ int egress_read_sock_fields(struct __sk_buff *skb)
>  	tpcpy(tp_ret, tp);
>  
>  	if (sk_ret == &srv_sk) {
> +		ktp = bpf_skc_to_tcp_sock(sk);
> +		if (!ktp)
> +			RET_LOG();
> +		lsndtime = ktp->lsndtime;

If I understood the patches correctly they extend the use cases significantly,
but this test only covers cgroup/egress type with one particular helper.
Could you please expand the test cases to cover more of things from earlier
patches. Patch 7 covers some, but imo not enough.
Similarly the case of sk_release(sk) in one place and sk_release(ktp)
in another branch (one you mention in commit log) needs to have
C based selftest as well.