bpf: Forbid trampoline attach for functions with variable arguments

Started by Jiri Olsa on May 5, 2021 · 5 messages from 5 authors · Last reply May 7, 2021

Subsystems (2)

Maintainers (6)

Show 2 more
STALE2015d LANDED

Status: Landed in mainline as 31379397dcc3 on 2021-05-06.

open the first message on its own page

Message details and related content
We can't currently allow to attach functions with variable arguments.
The problem is that we should save all the registers for arguments,
which is probably doable, but if caller uses more than 6 arguments,
we need stack data, which will be wrong, because of the extra stack
frame we do in bpf trampoline, so we could crash.

Also currently there's malformed trampoline code generated for such
functions at the moment as described in:
  https://lore.kernel.org/bpf/20210429212834.82621-1-jolsa@kernel.org/

Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
 kernel/bpf/btf.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)
diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
index 0600ed325fa0..161511bb3e51 100644
--- a/kernel/bpf/btf.c
+++ b/kernel/bpf/btf.c
@@ -5206,6 +5206,13 @@ int btf_distill_func_proto(struct bpf_verifier_log *log,
 	m->ret_size = ret;
 
 	for (i = 0; i < nargs; i++) {
+		if (i == nargs - 1 && args[i].type == 0) {
+			bpf_log(log,
+				"The function %s with variable args is unsupported.\n",
+				tname);
+			return -EINVAL;
+
+		}
 		ret = __get_type_size(btf, args[i].type, &t);
 		if (ret < 0) {
 			bpf_log(log,
@@ -5213,6 +5220,12 @@ int btf_distill_func_proto(struct bpf_verifier_log *log,
 				tname, i, btf_kind_str[BTF_INFO_KIND(t->info)]);
 			return -EINVAL;
 		}
+		if (ret == 0) {
+			bpf_log(log,
+				"The function %s has malformed void argument.\n",
+				tname);
+			return -EINVAL;
+		}
 		m->arg_size[i] = ret;
 	}
 	m->nr_args = nargs;
-- 
2.30.2
Message details and related content
On Wed, May 5, 2021 at 6:42 AM Jiri Olsa [off-list ref] wrote:
We can't currently allow to attach functions with variable arguments.
The problem is that we should save all the registers for arguments,
which is probably doable, but if caller uses more than 6 arguments,
we need stack data, which will be wrong, because of the extra stack
frame we do in bpf trampoline, so we could crash.

Also currently there's malformed trampoline code generated for such
functions at the moment as described in:
  https://lore.kernel.org/bpf/20210429212834.82621-1-jolsa@kernel.org/

Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
LGTM.

Acked-by: Andrii Nakryiko <andrii@kernel.org>
quoted hunk
 kernel/bpf/btf.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)
diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
index 0600ed325fa0..161511bb3e51 100644
--- a/kernel/bpf/btf.c
+++ b/kernel/bpf/btf.c
@@ -5206,6 +5206,13 @@ int btf_distill_func_proto(struct bpf_verifier_log *log,
        m->ret_size = ret;

        for (i = 0; i < nargs; i++) {
+               if (i == nargs - 1 && args[i].type == 0) {
+                       bpf_log(log,
+                               "The function %s with variable args is unsupported.\n",
+                               tname);
+                       return -EINVAL;
+
+               }
                ret = __get_type_size(btf, args[i].type, &t);
                if (ret < 0) {
                        bpf_log(log,
@@ -5213,6 +5220,12 @@ int btf_distill_func_proto(struct bpf_verifier_log *log,
                                tname, i, btf_kind_str[BTF_INFO_KIND(t->info)]);
                        return -EINVAL;
                }
+               if (ret == 0) {
+                       bpf_log(log,
+                               "The function %s has malformed void argument.\n",
+                               tname);
+                       return -EINVAL;
+               }
                m->arg_size[i] = ret;
        }
        m->nr_args = nargs;
--
2.30.2
Message details and related content
Hello:

This patch was applied to bpf/bpf.git (refs/heads/master):

On Wed,  5 May 2021 15:25:29 +0200 you wrote:
We can't currently allow to attach functions with variable arguments.
The problem is that we should save all the registers for arguments,
which is probably doable, but if caller uses more than 6 arguments,
we need stack data, which will be wrong, because of the extra stack
frame we do in bpf trampoline, so we could crash.

Also currently there's malformed trampoline code generated for such
functions at the moment as described in:
  https://lore.kernel.org/bpf/20210429212834.82621-1-jolsa@kernel.org/

[...]
Here is the summary with links:
  - bpf: Forbid trampoline attach for functions with variable arguments
    https://git.kernel.org/bpf/bpf/c/0a1a616720d9

You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html

Message details and related content
On 5/5/21 8:45 PM, Andrii Nakryiko wrote:
On Wed, May 5, 2021 at 6:42 AM Jiri Olsa [off-list ref] wrote:
quoted
We can't currently allow to attach functions with variable arguments.
The problem is that we should save all the registers for arguments,
which is probably doable, but if caller uses more than 6 arguments,
we need stack data, which will be wrong, because of the extra stack
frame we do in bpf trampoline, so we could crash.

Also currently there's malformed trampoline code generated for such
functions at the moment as described in:
   https://lore.kernel.org/bpf/20210429212834.82621-1-jolsa@kernel.org/

Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
LGTM.

Acked-by: Andrii Nakryiko <andrii@kernel.org>
quoted
  kernel/bpf/btf.c | 13 +++++++++++++
  1 file changed, 13 insertions(+)
diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
index 0600ed325fa0..161511bb3e51 100644
--- a/kernel/bpf/btf.c
+++ b/kernel/bpf/btf.c
@@ -5206,6 +5206,13 @@ int btf_distill_func_proto(struct bpf_verifier_log *log,
         m->ret_size = ret;

         for (i = 0; i < nargs; i++) {
+               if (i == nargs - 1 && args[i].type == 0) {
+                       bpf_log(log,
+                               "The function %s with variable args is unsupported.\n",
+                               tname);
+                       return -EINVAL;
+
(Jiri, fyi, I removed this extra newline while applying. Please scan for such
things before submitting.)
quoted
+               }
                 ret = __get_type_size(btf, args[i].type, &t);
                 if (ret < 0) {
                         bpf_log(log,
@@ -5213,6 +5220,12 @@ int btf_distill_func_proto(struct bpf_verifier_log *log,
                                 tname, i, btf_kind_str[BTF_INFO_KIND(t->info)]);
                         return -EINVAL;
                 }
+               if (ret == 0) {
+                       bpf_log(log,
+                               "The function %s has malformed void argument.\n",
+                               tname);
+                       return -EINVAL;
+               }
                 m->arg_size[i] = ret;
         }
         m->nr_args = nargs;
--
2.30.2
Message details and related content
On Fri, May 07, 2021 at 01:31:54AM +0200, Daniel Borkmann wrote:
On 5/5/21 8:45 PM, Andrii Nakryiko wrote:
quoted
On Wed, May 5, 2021 at 6:42 AM Jiri Olsa [off-list ref] wrote:
quoted
We can't currently allow to attach functions with variable arguments.
The problem is that we should save all the registers for arguments,
which is probably doable, but if caller uses more than 6 arguments,
we need stack data, which will be wrong, because of the extra stack
frame we do in bpf trampoline, so we could crash.

Also currently there's malformed trampoline code generated for such
functions at the moment as described in:
   https://lore.kernel.org/bpf/20210429212834.82621-1-jolsa@kernel.org/

Signed-off-by: Jiri Olsa <jolsa@kernel.org>
---
LGTM.

Acked-by: Andrii Nakryiko <andrii@kernel.org>
quoted
  kernel/bpf/btf.c | 13 +++++++++++++
  1 file changed, 13 insertions(+)
diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
index 0600ed325fa0..161511bb3e51 100644
--- a/kernel/bpf/btf.c
+++ b/kernel/bpf/btf.c
@@ -5206,6 +5206,13 @@ int btf_distill_func_proto(struct bpf_verifier_log *log,
         m->ret_size = ret;

         for (i = 0; i < nargs; i++) {
+               if (i == nargs - 1 && args[i].type == 0) {
+                       bpf_log(log,
+                               "The function %s with variable args is unsupported.\n",
+                               tname);
+                       return -EINVAL;
+
(Jiri, fyi, I removed this extra newline while applying. Please scan for such
things before submitting.)
sorry, will do

jirka
Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help