[PATCH v4] bpf: Fix UAF in bpf_trampoline_multi_attach_free on update failure

[PATCH v4] bpf: Fix UAF in bpf_trampoline_multi_attach_free on update failure

[Hui Zhu]

From: Hui Zhu <zhuhui@kylinos.cn>

When bpf_trampoline_update() fails before modify_fentry_multi()/
unregister_fentry_multi() is called, cur_image is unchanged
(cur_image == old_image) and ftrace still calls into it.  Freeing
old_image in that case causes a UAF.

Only free old_image when it differs from cur_image.

Also change bpf_trampoline_multi_detach() to return void and drop
the WARN_ON_ONCE at its call site.

Fixes: aef4dfa790b2 ("bpf: Add bpf_trampoline_multi_attach/detach functions")
Signed-off-by: Hui Zhu <zhuhui@kylinos.cn>
---
Changelog:
v4:
According to the comments of bot+bpf-ci, add Fixes: and update comments
of bpf_trampoline_multi_attach_free.
v3:
According to the comments of Jiri Olsa, drop patches 2/3 and the
prog-side machinery.
keep only the simplified image-side fix in
bpf_trampoline_multi_attach_free() and make
bpf_trampoline_multi_detach() return void.
v2:
Folded v1's two detach patches into patch 1.
According to the comments of Jiri Olsa, Pin the prog (pinned_prog) on
cur_image so it stays alive while ftrace may still call into it.
Make bpf_trampoline_multi_detach() return void.
Fix the same UAF in standard (non-multi) trampolines.
According to the comments of sashiko, Fix the prog UAF in
bpf_trampoline_multi_attach() rollback.
Leak the trampoline in bpf_trampoline_put() when cur_image is left
by a rollback.

 include/linux/bpf.h      |  9 ++++-----
 kernel/bpf/trampoline.c  | 16 +++++++++++++---
 kernel/trace/bpf_trace.c |  2 +-
 3 files changed, 18 insertions(+), 9 deletions(-)

diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index 73bacfc6444d..3255c6f5ea6b 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -1518,8 +1518,8 @@ int arch_prepare_bpf_dispatcher(void *image, void *buf, s64 *funcs, int num_func
 
 int bpf_trampoline_multi_attach(struct bpf_prog *prog, u32 *ids,
 				struct bpf_tracing_multi_link *link);
-int bpf_trampoline_multi_detach(struct bpf_prog *prog,
-				struct bpf_tracing_multi_link *link);
+void bpf_trampoline_multi_detach(struct bpf_prog *prog,
+				 struct bpf_tracing_multi_link *link);
 void bpf_trampoline_set_flags(struct bpf_trampoline *tr, u32 flags);
 
 /*
@@ -1639,10 +1639,9 @@ static inline int bpf_trampoline_multi_attach(struct bpf_prog *prog, u32 *ids,
 {
 	return -ENOTSUPP;
 }
-static inline int bpf_trampoline_multi_detach(struct bpf_prog *prog,
-					      struct bpf_tracing_multi_link *link)
+static inline void bpf_trampoline_multi_detach(struct bpf_prog *prog,
+					       struct bpf_tracing_multi_link *link)
 {
-	return -ENOTSUPP;
 }
 static inline void bpf_trampoline_set_flags(struct bpf_trampoline *tr, u32 flags) {}
 #endif
diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
index ed7999ad6c66..e30425789487 100644
--- a/kernel/bpf/trampoline.c
+++ b/kernel/bpf/trampoline.c
@@ -1595,7 +1595,17 @@ static void bpf_trampoline_multi_attach_init(struct bpf_trampoline *tr)
 
 static void bpf_trampoline_multi_attach_free(struct bpf_trampoline *tr)
 {
-	if (tr->multi_attach.old_image)
+	/*
+	 * Only free old_image if it is no longer the active image.
+	 * When bpf_trampoline_update() fails before modify_fentry_multi()/
+	 * unregister_fentry_multi() is called, cur_image is unchanged
+	 * (cur_image == old_image) and ftrace still points to it. Freeing
+	 * it would cause a UAF when ftrace calls into the freed memory.
+	 * On success, cur_image is either a new image or NULL, so
+	 * old_image != cur_image means the image is stale.
+	 */
+	if (tr->multi_attach.old_image &&
+	    tr->multi_attach.old_image != tr->cur_image)
 		bpf_tramp_image_put(tr->multi_attach.old_image);
 
 	tr->multi_attach.old_image = NULL;
@@ -1719,7 +1729,8 @@ int bpf_trampoline_multi_attach(struct bpf_prog *prog, u32 *ids,
 	return err;
 }
 
-int bpf_trampoline_multi_detach(struct bpf_prog *prog, struct bpf_tracing_multi_link *link)
+void bpf_trampoline_multi_detach(struct bpf_prog *prog,
+				 struct bpf_tracing_multi_link *link)
 {
 	struct bpf_tracing_multi_data *data = &link->data;
 	struct bpf_tracing_multi_node *mnode;
@@ -1749,7 +1760,6 @@ int bpf_trampoline_multi_detach(struct bpf_prog *prog, struct bpf_tracing_multi_
 		bpf_trampoline_put(mnode->trampoline);
 
 	clear_tracing_multi_data(data);
-	return 0;
 }
 
 #undef for_each_mnode_cnt
diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
index 891897f8a1b3..29260951aa87 100644
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -3687,7 +3687,7 @@ static void bpf_tracing_multi_link_release(struct bpf_link *link)
 	struct bpf_tracing_multi_link *tr_link =
 		container_of(link, struct bpf_tracing_multi_link, link);
 
-	WARN_ON_ONCE(bpf_trampoline_multi_detach(link->prog, tr_link));
+	bpf_trampoline_multi_detach(link->prog, tr_link);
 }
 
 static void bpf_tracing_multi_link_dealloc(struct bpf_link *link)
-- 
2.53.0

Re: [PATCH v4] bpf: Fix UAF in bpf_trampoline_multi_attach_free on update failure

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:

Pre-existing issues:
- [High] Ignoring errors from `__bpf_trampoline_unlink_prog()` during detach leads to a Use-After-Free of the BPF program.
- [High] Ignoring errors from `update_ftrace_direct_del/mod()` during detach or attach rollback leads to a Use-After-Free of the trampoline image.
--

--- Patch [1]: [PATCH v4] bpf: Fix UAF in bpf_trampoline_multi_attach_free on update failure ---
commit 12049949b83ebddc54a6fdad57fa0b635706dbda
Author: Hui Zhu <zhuhui@kylinos.cn>
bpf: Fix UAF in bpf_trampoline_multi_attach_free on update failure

When bpf_trampoline_update() fails before modify_fentry_multi() or
unregister_fentry_multi() is called, cur_image is unchanged and ftrace still
calls into it. Freeing old_image in that case causes a UAF. This patch only
frees old_image when it differs from cur_image, and makes
bpf_trampoline_multi_detach() return void.

> diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
> index e07af35ed0402..90b70ea0d3709 100644
> --- a/kernel/bpf/trampoline.c
> +++ b/kernel/bpf/trampoline.c
> @@ -1632,7 +1632,17 @@ static void bpf_trampoline_multi_attach_init(struct bpf_trampoline *tr)
>  
>  static void bpf_trampoline_multi_attach_free(struct bpf_trampoline *tr)
>  {
> -	if (tr->multi_attach.old_image)
> +	/*
> +	 * Only free old_image if it is no longer the active image.
> +	 * When bpf_trampoline_update() fails before modify_fentry_multi()/
> +	 * unregister_fentry_multi() is called, cur_image is unchanged
> +	 * (cur_image == old_image) and ftrace still points to it. Freeing
> +	 * it would cause a UAF when ftrace calls into the freed memory.
> +	 * On success, cur_image is either a new image or NULL, so
> +	 * old_image != cur_image means the image is stale.
> +	 */
> +	if (tr->multi_attach.old_image &&
> +	    tr->multi_attach.old_image != tr->cur_image)

[Severity: High]
This is a pre-existing issue, but does this condition still allow a
use-after-free of the trampoline image if ftrace updates fail?

If __bpf_trampoline_unlink_prog() succeeds in bpf_trampoline_multi_detach(),
the active image is updated and old_image != cur_image will be true.

However, if the subsequent update to ftrace fails due to memory pressure:

    if (ftrace_hash_count(data->unreg))
        WARN_ON_ONCE(update_ftrace_direct_del(&direct_ops, data->unreg));

ftrace will return an error without updating its hooks, leaving ftrace
pointing to the original active image (old_image).

Since old_image != cur_image is true, won't bpf_trampoline_multi_attach_free()
then free old_image while ftrace is still executing it?

A similar issue appears to exist during error rollback in
kernel/bpf/trampoline.c:bpf_trampoline_multi_attach() if
update_ftrace_direct_del() fails, leaving ftrace executing newly allocated
images that are subsequently freed.

>  		bpf_tramp_image_put(tr->multi_attach.old_image);
>  
>  	tr->multi_attach.old_image = NULL;

[ ... ]

> diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
> index 891897f8a1b35..29260951aa871 100644
> --- a/kernel/trace/bpf_trace.c
> +++ b/kernel/trace/bpf_trace.c
> @@ -3687,7 +3687,7 @@ static void bpf_tracing_multi_link_release(struct bpf_link *link)
>  	struct bpf_tracing_multi_link *tr_link =
>  		container_of(link, struct bpf_tracing_multi_link, link);
>  
> -	WARN_ON_ONCE(bpf_trampoline_multi_detach(link->prog, tr_link));
> +	bpf_trampoline_multi_detach(link->prog, tr_link);

[Severity: High]
This is a pre-existing issue, but does dropping the program reference here
lead to a use-after-free if __bpf_trampoline_unlink_prog() fails?

In kernel/bpf/trampoline.c:bpf_trampoline_multi_detach(), if unlink fails:

    err = __bpf_trampoline_unlink_prog(&mnode->node, mnode->trampoline, NULL,
                                       &trampoline_multi_ops, data);
    WARN_ONCE(err, "__bpf_trampoline_unlink_prog failed: %d\n", err);

the detach path ignores the error. The ftrace direct hook remains active and
the active trampoline image remains unchanged, keeping its hardcoded call
to the BPF program.

Since bpf_tracing_multi_link_release() proceeds to drop the link's reference
to the BPF program, if the program's refcount drops to zero and it is freed,
could ftrace continue to jump into the freed program?

>  }
>  
>  static void bpf_tracing_multi_link_dealloc(struct bpf_link *link)

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260810061329.224276-1-hui.zhu@linux.dev?part=1

Re: [PATCH v4] bpf: Fix UAF in bpf_trampoline_multi_attach_free on update failure

[bot+bpf-ci]

[-- Attachment #1: Type: text/plain, Size: 2149 bytes --]

> diff --git a/include/linux/bpf.h b/include/linux/bpf.h
> index b4a10c9878cf..f4e8d372253a 100644
> --- a/include/linux/bpf.h
> +++ b/include/linux/bpf.h
> @@ -1518,8 +1518,8 @@ int arch_prepare_bpf_dispatcher(void *image, void *buf, s64 *funcs, int num_func
>
>  int bpf_trampoline_multi_attach(struct bpf_prog *prog, u32 *ids,
>  				struct bpf_tracing_multi_link *link);
> -int bpf_trampoline_multi_detach(struct bpf_prog *prog,
> -				struct bpf_tracing_multi_link *link);
> +void bpf_trampoline_multi_detach(struct bpf_prog *prog,
> +				 struct bpf_tracing_multi_link *link);

[ ... ]

> diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c
> index e07af35ed040..90b70ea0d370 100644
> --- a/kernel/bpf/trampoline.c
> +++ b/kernel/bpf/trampoline.c

[ ... ]

> @@ -1756,7 +1766,8 @@ int bpf_trampoline_multi_attach(struct bpf_prog *prog, u32 *ids,
>  	return err;
>  }
>
> -int bpf_trampoline_multi_detach(struct bpf_prog *prog, struct bpf_tracing_multi_link *link)
> +void bpf_trampoline_multi_detach(struct bpf_prog *prog,
> +				 struct bpf_tracing_multi_link *link)

[ ... ]

> diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
> index 891897f8a1b3..29260951aa87 100644
> --- a/kernel/trace/bpf_trace.c
> +++ b/kernel/trace/bpf_trace.c
> @@ -3687,7 +3687,7 @@ static void bpf_tracing_multi_link_release(struct bpf_link *link)
>  	struct bpf_tracing_multi_link *tr_link =
>  		container_of(link, struct bpf_tracing_multi_link, link);
>
> -	WARN_ON_ONCE(bpf_trampoline_multi_detach(link->prog, tr_link));
> +	bpf_trampoline_multi_detach(link->prog, tr_link);
>  }

This isn't a bug, but would it be worth splitting the return-type
change into its own patch?

The int-to-void conversion of bpf_trampoline_multi_detach() touches
both bpf.h prototypes, the definition and the bpf_trace.c call site,
but is independent of the old_image lifetime fix that the Fixes: tag
refers to.


---
AI reviewed your patch. Please fix the bug or email reply why it's not a bug.
See: https://github.com/kernel-patches/vmtest/blob/master/ci/claude/README.md

CI run summary: https://github.com/kernel-patches/bpf/actions/runs/31362089875