[PATCH bpf-next v3 1/2] bpf: Remove bpf_probe_write_user() warning message

[PATCH bpf-next v3 1/2] bpf: Remove bpf_probe_write_user() warning message

[Marco Elver]

The warning message for bpf_probe_write_user() was introduced in
96ae52279594 ("bpf: Add bpf_probe_write_user BPF helper to be called in
tracers"), with the following in the commit message:

    Given this feature is meant for experiments, and it has a risk of
    crashing the system, and running programs, we print a warning on
    when a proglet that attempts to use this helper is installed,
    along with the pid and process name.

After 8 years since 96ae52279594, bpf_probe_write_user() has found
successful applications beyond experiments [1, 2], with no other good
alternatives. Despite its intended purpose for "experiments", that
doesn't stop Hyrum's law, and there are likely many more users depending
on this helper: "[..] it does not matter what you promise [..] all
observable behaviors of your system will be depended on by somebody."

The ominous "helper that may corrupt user memory!" has offered no real
benefit, and has been found to lead to confusion where the system
administrator is loading programs with valid use cases.

As such, remove the warning message.

Link: https://lore.kernel.org/lkml/20240404190146.1898103-1-elver@google.com/ [1]
Link: https://lore.kernel.org/r/lkml/CAAn3qOUMD81-vxLLfep0H6rRd74ho2VaekdL4HjKq+Y1t9KdXQ@mail.gmail.com/ [2]
Link: https://lore.kernel.org/all/CAEf4Bzb4D_=zuJrg3PawMOW3KqF8JvJm9SwF81_XHR2+u5hkUg@mail.gmail.com/
Signed-off-by: Marco Elver <elver@google.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
---
v3:
* Collect Ack from Jiri.

v2:
* Just delete the message entirely (suggested by Andrii Nakryiko)
---
 kernel/trace/bpf_trace.c | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
index 630b763e5240..0ab56af2e298 100644
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -362,9 +362,6 @@ static const struct bpf_func_proto *bpf_get_probe_write_proto(void)
 	if (!capable(CAP_SYS_ADMIN))
 		return NULL;
 
-	pr_warn_ratelimited("%s[%d] is installing a program with bpf_probe_write_user helper that may corrupt user memory!",
-			    current->comm, task_pid_nr(current));
-
 	return &bpf_probe_write_user_proto;
 }
 
-- 
2.47.0.338.g60cca15819-goog

[PATCH bpf-next v3 2/2] bpf: Refactor bpf_tracing_func_proto() and remove bpf_get_probe_write_proto()

[Marco Elver]

With bpf_get_probe_write_proto() no longer printing a message, we can
avoid it being a special case with its own permission check.

Refactor bpf_tracing_func_proto() similar to bpf_base_func_proto() to
have a section conditional on bpf_token_capable(CAP_SYS_ADMIN), where
the proto for bpf_probe_write_user() is returned. Finally, remove the
unnecessary bpf_get_probe_write_proto().

This simplifies the code, and adding additional CAP_SYS_ADMIN-only
helpers in future avoids duplicating the same CAP_SYS_ADMIN check.

Suggested-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Marco Elver <elver@google.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
---
v3:
* Fix where bpf_base_func_proto() is called - it needs to be last,
  because we may override protos (as is e.g. done for
  BPF_FUNC_get_smp_processor_id).

v2:
* New patch.
---
 kernel/trace/bpf_trace.c | 25 +++++++++++++------------
 1 file changed, 13 insertions(+), 12 deletions(-)

diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
index 0ab56af2e298..9b1d1fa4c06c 100644
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -357,14 +357,6 @@ static const struct bpf_func_proto bpf_probe_write_user_proto = {
 	.arg3_type	= ARG_CONST_SIZE,
 };
 
-static const struct bpf_func_proto *bpf_get_probe_write_proto(void)
-{
-	if (!capable(CAP_SYS_ADMIN))
-		return NULL;
-
-	return &bpf_probe_write_user_proto;
-}
-
 #define MAX_TRACE_PRINTK_VARARGS	3
 #define BPF_TRACE_PRINTK_SIZE		1024
 
@@ -1458,9 +1450,6 @@ bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 		return &bpf_perf_event_read_proto;
 	case BPF_FUNC_get_prandom_u32:
 		return &bpf_get_prandom_u32_proto;
-	case BPF_FUNC_probe_write_user:
-		return security_locked_down(LOCKDOWN_BPF_WRITE_USER) < 0 ?
-		       NULL : bpf_get_probe_write_proto();
 	case BPF_FUNC_probe_read_user:
 		return &bpf_probe_read_user_proto;
 	case BPF_FUNC_probe_read_kernel:
@@ -1539,8 +1528,20 @@ bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 	case BPF_FUNC_trace_vprintk:
 		return bpf_get_trace_vprintk_proto();
 	default:
-		return bpf_base_func_proto(func_id, prog);
+		break;
 	}
+
+	if (bpf_token_capable(prog->aux->token, CAP_SYS_ADMIN)) {
+		switch (func_id) {
+		case BPF_FUNC_probe_write_user:
+			return security_locked_down(LOCKDOWN_BPF_WRITE_USER) < 0 ?
+			       NULL : &bpf_probe_write_user_proto;
+		default:
+			break;
+		}
+	}
+
+	return bpf_base_func_proto(func_id, prog);
 }
 
 static bool is_kprobe_multi(const struct bpf_prog *prog)
-- 
2.47.0.338.g60cca15819-goog

Re: [PATCH bpf-next v3 2/2] bpf: Refactor bpf_tracing_func_proto() and remove bpf_get_probe_write_proto()

[Alexei Starovoitov]

On Wed, Nov 27, 2024 at 6:10 AM Marco Elver <elver@google.com> wrote:
>
> With bpf_get_probe_write_proto() no longer printing a message, we can
> avoid it being a special case with its own permission check.
>
> Refactor bpf_tracing_func_proto() similar to bpf_base_func_proto() to
> have a section conditional on bpf_token_capable(CAP_SYS_ADMIN), where
> the proto for bpf_probe_write_user() is returned. Finally, remove the
> unnecessary bpf_get_probe_write_proto().
>
> This simplifies the code, and adding additional CAP_SYS_ADMIN-only
> helpers in future avoids duplicating the same CAP_SYS_ADMIN check.
>
> Suggested-by: Andrii Nakryiko <andrii@kernel.org>
> Signed-off-by: Marco Elver <elver@google.com>
> Acked-by: Jiri Olsa <jolsa@kernel.org>
> ---
> v3:
> * Fix where bpf_base_func_proto() is called - it needs to be last,
>   because we may override protos (as is e.g. done for
>   BPF_FUNC_get_smp_processor_id).
>
> v2:
> * New patch.
> ---
>  kernel/trace/bpf_trace.c | 25 +++++++++++++------------
>  1 file changed, 13 insertions(+), 12 deletions(-)
>
> diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
> index 0ab56af2e298..9b1d1fa4c06c 100644
> --- a/kernel/trace/bpf_trace.c
> +++ b/kernel/trace/bpf_trace.c
> @@ -357,14 +357,6 @@ static const struct bpf_func_proto bpf_probe_write_user_proto = {
>         .arg3_type      = ARG_CONST_SIZE,
>  };
>
> -static const struct bpf_func_proto *bpf_get_probe_write_proto(void)
> -{
> -       if (!capable(CAP_SYS_ADMIN))
> -               return NULL;
> -
> -       return &bpf_probe_write_user_proto;
> -}
> -
>  #define MAX_TRACE_PRINTK_VARARGS       3
>  #define BPF_TRACE_PRINTK_SIZE          1024
>
> @@ -1458,9 +1450,6 @@ bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
>                 return &bpf_perf_event_read_proto;
>         case BPF_FUNC_get_prandom_u32:
>                 return &bpf_get_prandom_u32_proto;
> -       case BPF_FUNC_probe_write_user:
> -               return security_locked_down(LOCKDOWN_BPF_WRITE_USER) < 0 ?
> -                      NULL : bpf_get_probe_write_proto();
>         case BPF_FUNC_probe_read_user:
>                 return &bpf_probe_read_user_proto;
>         case BPF_FUNC_probe_read_kernel:
> @@ -1539,8 +1528,20 @@ bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
>         case BPF_FUNC_trace_vprintk:
>                 return bpf_get_trace_vprintk_proto();
>         default:
> -               return bpf_base_func_proto(func_id, prog);
> +               break;
>         }
> +
> +       if (bpf_token_capable(prog->aux->token, CAP_SYS_ADMIN)) {
> +               switch (func_id) {
> +               case BPF_FUNC_probe_write_user:
> +                       return security_locked_down(LOCKDOWN_BPF_WRITE_USER) < 0 ?
> +                              NULL : &bpf_probe_write_user_proto;
> +               default:
> +                       break;
> +               }
> +       }
> +
> +       return bpf_base_func_proto(func_id, prog);

Moving bpf_base_func_proto() all the way to the top was incorrect,
but here we can move it just above this bpf_token_capable() check
and remove extra indent like:

func_proto = bpf_base_func_proto();
if (func_proto)
   return func_proto;
if (!bpf_token_capable(prog->aux->token, CAP_SYS_ADMIN))
   return NULL;
switch (func_id) {
case BPF_FUNC_probe_write_user:

that will align it with the style of bpf_base_func_proto().

pw-bot: cr

Re: [PATCH bpf-next v3 2/2] bpf: Refactor bpf_tracing_func_proto() and remove bpf_get_probe_write_proto()

[Marco Elver]

On Thu, Nov 28, 2024 at 10:22AM -0800, Alexei Starovoitov wrote:
[..]
> Moving bpf_base_func_proto() all the way to the top was incorrect,
> but here we can move it just above this bpf_token_capable() check
> and remove extra indent like:
> 
> func_proto = bpf_base_func_proto();
> if (func_proto)
>    return func_proto;
> if (!bpf_token_capable(prog->aux->token, CAP_SYS_ADMIN))
>    return NULL;
> switch (func_id) {
> case BPF_FUNC_probe_write_user:
> 
> that will align it with the style of bpf_base_func_proto().
> 
> pw-bot: cr

Ack, let me change that.

Below is preview of v4 for this bit.

@@ -1417,6 +1409,8 @@ late_initcall(bpf_key_sig_kfuncs_init);
 static const struct bpf_func_proto *
 bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 {
+	const struct bpf_func_proto *func_proto;
+
 	switch (func_id) {
 	case BPF_FUNC_map_lookup_elem:
 		return &bpf_map_lookup_elem_proto;
@@ -1458,9 +1452,6 @@ bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 		return &bpf_perf_event_read_proto;
 	case BPF_FUNC_get_prandom_u32:
 		return &bpf_get_prandom_u32_proto;
-	case BPF_FUNC_probe_write_user:
-		return security_locked_down(LOCKDOWN_BPF_WRITE_USER) < 0 ?
-		       NULL : bpf_get_probe_write_proto();
 	case BPF_FUNC_probe_read_user:
 		return &bpf_probe_read_user_proto;
 	case BPF_FUNC_probe_read_kernel:
@@ -1539,7 +1530,22 @@ bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 	case BPF_FUNC_trace_vprintk:
 		return bpf_get_trace_vprintk_proto();
 	default:
-		return bpf_base_func_proto(func_id, prog);
+		break;
+	}
+
+	func_proto = bpf_base_func_proto(func_id, prog);
+	if (func_proto)
+		return func_proto;
+
+	if (!bpf_token_capable(prog->aux->token, CAP_SYS_ADMIN))
+		return NULL;
+
+	switch (func_id) {
+	case BPF_FUNC_probe_write_user:
+		return security_locked_down(LOCKDOWN_BPF_WRITE_USER) < 0 ?
+		       NULL : &bpf_probe_write_user_proto;
+	default:
+		return NULL;
 	}
 }
 

Re: [PATCH bpf-next v3 2/2] bpf: Refactor bpf_tracing_func_proto() and remove bpf_get_probe_write_proto()

[Alexei Starovoitov]

On Thu, Nov 28, 2024 at 10:36 AM Marco Elver <elver@google.com> wrote:
>
> On Thu, Nov 28, 2024 at 10:22AM -0800, Alexei Starovoitov wrote:
> [..]
> > Moving bpf_base_func_proto() all the way to the top was incorrect,
> > but here we can move it just above this bpf_token_capable() check
> > and remove extra indent like:
> >
> > func_proto = bpf_base_func_proto();
> > if (func_proto)
> >    return func_proto;
> > if (!bpf_token_capable(prog->aux->token, CAP_SYS_ADMIN))
> >    return NULL;
> > switch (func_id) {
> > case BPF_FUNC_probe_write_user:
> >
> > that will align it with the style of bpf_base_func_proto().
> >
> > pw-bot: cr
>
> Ack, let me change that.
>
> Below is preview of v4 for this bit.

lgtm