Date:   Fri, 1 Mar 2019 15:58:52 +0000
From:   Lorenz Bauer <lmb@...udflare.com>
To:     Martin KaFai Lau <kafai@...com>
Cc:     netdev@...r.kernel.org, Alexei Starovoitov <ast@...com>,
        Daniel Borkmann <daniel@...earbox.net>, kernel-team@...com
Subject: Re: [PATCH bpf-next 1/2] bpf: Fix bpf_tcp_sock and bpf_sk_fullsock
 issue related to bpf_sk_release

On Fri, 1 Mar 2019 at 07:03, Martin KaFai Lau <kafai@...com> wrote:
>
> Lorenz Bauer [thanks!] reported that a ptr returned by bpf_tcp_sock(sk)
> can still be accessed after bpf_sk_release(sk).
> Both bpf_tcp_sock() and bpf_sk_fullsock() have the same issue.
> This patch addresses them together.
>
> A simple reproducer looks like this:
>
> sk = bpf_sk_lookup_tcp();
> /* if (!sk) ... */
> tp = bpf_tcp_sock(sk);
> /* if (!tp) ... */
> bpf_sk_release(sk);
> snd_cwnd = tp->snd_cwnd; /* oops! The verifier does not complain. */
>
> The problem is the verifier did not scrub the register's states of
> the tcp_sock ptr (tp) after bpf_sk_release(sk).
>
> [ Note that when calling bpf_tcp_sock(sk), the sk is not always
>   refcount-acquired. e.g. bpf_tcp_sock(skb->sk). The verifier works
>   fine for this case. ]
>
> Currently, the verifier does not track if a helper's return ptr (in REG_0)
> is "carry"-ing one of its argument's refcount status. To carry this info,
> the reg1->id needs to be stored in reg0.  The reg0->id has already
> been used for NULL checking purpose.  Hence, a new "refcount_id"
> is needed in "struct bpf_reg_state".
>
> With refcount_id, when bpf_sk_release(sk) is called, the verifier can scrub
> all reg states which has a refcount_id match.  It is done with the changes
> in release_reg_references().
>
> When acquiring and releasing a refcount, the reg->id is still used.
> Hence, we cannot do "bpf_sk_release(tp)" in the above reproducer
> example.
>
> Misc change notes:
> - With the new refcount_id, reg_is_refcounted() test can now be
>   done with "reg->refcount_id && reg->id == reg->refcount_id" instead of
>   testing the ptr type.
>
>   The type_is_refcounted() and type_is_refcounted_or_null()
>   are no longer needed, so removed.
>
> - An anonymous struct is added to bpf_call_arg_meta to store
>   the reg->id and reg->refcount_id of the arg.  Otherwise,
>   they will be unavailable after check_helper_call()
>   has cleared all CALLER_SAVED_REGS.
>
> - The check_func_arg() can only allow one refcount-ed arg.  It is
>   guaranteed by check_refcount_ok() which ensures at most one arg can be
>   refcount-ed.  Hence, it is a verifier internal error if >1 refcount arg
>   found in check_func_arg().
>
> - The check_func_arg() also complains if a "is_acquire_function(func_id)"
>   helper is having a refcount-ed arg.  No func_id is doing this now
>   and should have been rejected earlier, so it is treated as
>   verifier internal error also.
>
> - In check_func_arg(), the "!reg->id" check is removed under
>   the ARG_PTR_TO_SOCKET case.  It is because a PTR_TO_SOCKET
>   can be obtained from bpf_sk_fullsock() and it does not
>   take a refcount.  The verifier will still complain during
>   release_reference() but it does not treat it as a
>   verifier internal error anymore.
>
> - In release_reference(), release_reference_state() is called
>   first to ensure a match on "reg->id" can be found before
>   scrubbing the reg states with release_reg_references().
>
> Fixes: 655a51e536c0 ("bpf: Add struct bpf_tcp_sock and BPF_FUNC_tcp_sock")
> Cc: Lorenz Bauer <lmb@...udflare.com>
> Reported-by: Lorenz Bauer <lmb@...udflare.com>
> Signed-off-by: Martin KaFai Lau <kafai@...com>
> ---
>  include/linux/bpf_verifier.h |  9 ++++
>  kernel/bpf/verifier.c        | 89 ++++++++++++++++++++++--------------
>  2 files changed, 64 insertions(+), 34 deletions(-)
>
> diff --git a/include/linux/bpf_verifier.h b/include/linux/bpf_verifier.h
> index 69f7a3449eda..b7698d0534cb 100644
> --- a/include/linux/bpf_verifier.h
> +++ b/include/linux/bpf_verifier.h
> @@ -66,6 +66,15 @@ struct bpf_reg_state {
>          * same reference to the socket, to determine proper reference freeing.
>          */
>         u32 id;
> +       /* For PTR_TO_SOCKET and PTR_TO_TCP_SOCK, this ptr may not actually
> +        * hold a refcount of a socket but instead it is a ptr
> +        * returned from a helper which is based on its refcount-ed
> +        * ptr argument (e.g. bpf_tcp_sock()).
> +        * "refcount_id" stores which refcount-ed argument it
> +        * originally derived from.  When this original argument's
> +        * refcount is released, this ptr will also be invalidated.
> +        */
> +       u32 refcount_id;
>         /* For scalar types (SCALAR_VALUE), this represents our knowledge of
>          * the actual value.
>          * For pointer types, this represents the variable part of the offset
> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index 1b9496c41383..682b7a6f0d2a 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c
> @@ -212,7 +212,10 @@ struct bpf_call_arg_meta {
>         int access_size;
>         s64 msize_smax_value;
>         u64 msize_umax_value;
> -       int ptr_id;
> +       struct {
> +               u32 id;
> +               u32 refcount_id;
> +       } refcount_reg;
>         int func_id;
>  };
>
> @@ -346,19 +349,9 @@ static bool reg_type_may_be_null(enum bpf_reg_type type)
>                type == PTR_TO_TCP_SOCK_OR_NULL;
>  }
>
> -static bool type_is_refcounted(enum bpf_reg_type type)
> -{
> -       return type == PTR_TO_SOCKET;
> -}
> -
> -static bool type_is_refcounted_or_null(enum bpf_reg_type type)
> -{
> -       return type == PTR_TO_SOCKET || type == PTR_TO_SOCKET_OR_NULL;
> -}
> -
>  static bool reg_is_refcounted(const struct bpf_reg_state *reg)
>  {
> -       return type_is_refcounted(reg->type);
> +       return reg->refcount_id && reg->id == reg->refcount_id;
>  }
>
>  static bool reg_may_point_to_spin_lock(const struct bpf_reg_state *reg)
> @@ -367,14 +360,10 @@ static bool reg_may_point_to_spin_lock(const struct bpf_reg_state *reg)
>                 map_value_has_spin_lock(reg->map_ptr);
>  }
>
> -static bool reg_is_refcounted_or_null(const struct bpf_reg_state *reg)
> -{
> -       return type_is_refcounted_or_null(reg->type);
> -}
> -
>  static bool arg_type_is_refcounted(enum bpf_arg_type type)
>  {
> -       return type == ARG_PTR_TO_SOCKET;
> +       return type == ARG_PTR_TO_SOCKET ||
> +               type == ARG_PTR_TO_SOCK_COMMON;
>  }
>
>  /* Determine whether the function releases some resources allocated by another
> @@ -392,6 +381,12 @@ static bool is_acquire_function(enum bpf_func_id func_id)
>                 func_id == BPF_FUNC_sk_lookup_udp;
>  }
>
> +static bool is_refcount_carrying_function(enum bpf_func_id func_id)
> +{
> +       return func_id == BPF_FUNC_tcp_sock ||
> +               func_id == BPF_FUNC_sk_fullsock;
> +}
> +
>  /* string representation of 'enum bpf_reg_type' */
>  static const char * const reg_type_str[] = {
>         [NOT_INIT]              = "?",
> @@ -465,7 +460,8 @@ static void print_verifier_state(struct bpf_verifier_env *env,
>                         if (t == PTR_TO_STACK)
>                                 verbose(env, ",call_%d", func(env, reg)->callsite);
>                 } else {
> -                       verbose(env, "(id=%d", reg->id);
> +                       verbose(env, "(id=%d refcount_id=%d", reg->id,
> +                               reg->refcount_id);
>                         if (t != SCALAR_VALUE)
>                                 verbose(env, ",off=%d", reg->off);
>                         if (type_is_pkt_pointer(t))
> @@ -2418,12 +2414,6 @@ static int check_func_arg(struct bpf_verifier_env *env, u32 regno,
>                 expected_type = PTR_TO_SOCKET;
>                 if (type != expected_type)
>                         goto err_type;
> -               if (meta->ptr_id || !reg->id) {
> -                       verbose(env, "verifier internal error: mismatched references meta=%d, reg=%d\n",
> -                               meta->ptr_id, reg->id);
> -                       return -EFAULT;
> -               }
> -               meta->ptr_id = reg->id;
>         } else if (arg_type == ARG_PTR_TO_SPIN_LOCK) {
>                 if (meta->func_id == BPF_FUNC_spin_lock) {
>                         if (process_spin_lock(env, regno, true))
> @@ -2532,6 +2522,26 @@ static int check_func_arg(struct bpf_verifier_env *env, u32 regno,
>                                               zero_size_allowed, meta);
>         }
>
> +       if (reg->refcount_id) {
> +               if (meta->refcount_reg.refcount_id) {
> +                       verbose(env, "verifier internal error: more than one arg with refcount_id R%d %u %u\n",
> +                               regno, reg->refcount_id,
> +                               meta->refcount_reg.refcount_id);
> +                       return -EFAULT;
> +               }
> +
> +               if (is_acquire_function(meta->func_id)) {
> +                       verbose(env,
> +                               "verifier internal error: func %s#%d taking an already refcount-ed arg R%d\n",
> +                               func_id_name(meta->func_id), meta->func_id,
> +                               regno);
> +                       return -EFAULT;
> +               }
> +
> +               meta->refcount_reg.id = reg->id;
> +               meta->refcount_reg.refcount_id = reg->refcount_id;
> +       }
> +
>         return err;
>  err_type:
>         verbose(env, "R%d type=%s expected=%s\n", regno,
> @@ -2799,19 +2809,20 @@ static void clear_all_pkt_pointers(struct bpf_verifier_env *env)
>  }
>
>  static void release_reg_references(struct bpf_verifier_env *env,
> -                                  struct bpf_func_state *state, int id)
> +                                  struct bpf_func_state *state,
> +                                  int refcount_id)

Could this just take id?  release_reference could then also only take
id. The way I see it, release_reg_references should clear all
registers and stack spills where either id == id, or reference_id ==
id.
In the current code that is given because id == reference_id if
release_reference_state succeeds, but I find that harder to
understand.

>  {
>         struct bpf_reg_state *regs = state->regs, *reg;
>         int i;
>
>         for (i = 0; i < MAX_BPF_REG; i++)
> -               if (regs[i].id == id)
> +               if (regs[i].refcount_id == refcount_id)
>                         mark_reg_unknown(env, regs, i);
>
>         bpf_for_each_spilled_reg(i, state, reg) {
>                 if (!reg)
>                         continue;
> -               if (reg_is_refcounted(reg) && reg->id == id)
> +               if (reg->refcount_id == refcount_id)
>                         __mark_reg_unknown(reg);
>         }
>  }
> @@ -2819,16 +2830,21 @@ static void release_reg_references(struct bpf_verifier_env *env,
>  /* The pointer with the specified id has released its reference to kernel
>   * resources. Identify all copies of the same pointer and clear the reference.
>   */
> -static int release_reference(struct bpf_verifier_env *env,
> -                            struct bpf_call_arg_meta *meta)
> +static int release_reference(struct bpf_verifier_env *env, u32 id,
> +                            u32 refcount_id)
>  {
>         struct bpf_verifier_state *vstate = env->cur_state;
> +       int err;
>         int i;
>
> +       err = release_reference_state(cur_func(env), id);
> +       if (err)
> +               return err;
> +
>         for (i = 0; i <= vstate->curframe; i++)
> -               release_reg_references(env, vstate->frame[i], meta->ptr_id);
> +               release_reg_references(env, vstate->frame[i], refcount_id);
>
> -       return release_reference_state(cur_func(env), meta->ptr_id);
> +       return 0;
>  }
>
>  static int check_func_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
> @@ -3093,7 +3109,8 @@ static int check_helper_call(struct bpf_verifier_env *env, int func_id, int insn
>                         return err;
>                 }
>         } else if (is_release_function(func_id)) {
> -               err = release_reference(env, &meta);
> +               err = release_reference(env, meta.refcount_reg.id,
> +                                       meta.refcount_reg.refcount_id);
>                 if (err) {
>                         verbose(env, "func %s#%d reference has not been acquired before\n",
>                                 func_id_name(func_id), func_id);
> @@ -3156,6 +3173,7 @@ static int check_helper_call(struct bpf_verifier_env *env, int func_id, int insn
>                                 return id;
>                         /* For release_reference() */
>                         regs[BPF_REG_0].id = id;
> +                       regs[BPF_REG_0].refcount_id = id;
>                 } else {
>                         /* For mark_ptr_or_null_reg() */
>                         regs[BPF_REG_0].id = ++env->id_gen;
> @@ -3170,6 +3188,9 @@ static int check_helper_call(struct bpf_verifier_env *env, int func_id, int insn
>                 return -EINVAL;
>         }
>
> +       if (is_refcount_carrying_function(func_id))
> +               regs[BPF_REG_0].refcount_id = meta.refcount_reg.refcount_id;
> +
>         do_refine_retval_range(regs, fn->ret_type, func_id, &meta);
>
>         err = check_map_func_compatibility(env, meta.map_ptr, func_id);
> @@ -4687,7 +4708,7 @@ static void mark_ptr_or_null_regs(struct bpf_verifier_state *vstate, u32 regno,
>         u32 id = regs[regno].id;
>         int i, j;
>
> -       if (reg_is_refcounted_or_null(&regs[regno]) && is_null)
> +       if (reg_is_refcounted(&regs[regno]) && is_null)
>                 release_reference_state(state, id);

Since reg_is_refcounted() is true, release_reference_state should
always succeed. Check the error code?

>
>         for (i = 0; i < MAX_BPF_REG; i++)
> --
> 2.17.1
>

How about resetting reg->refcount_id in mark_ptr_or_null_reg as well?
(Aside: where does the
state pruning referenced by the comment in that function take place?)

-- 
Lorenz Bauer  |  Systems Engineer
25 Lavington St., London SE1 0NZ

www.cloudflare.com

Powered by blists - more mailing lists