[PATCH 5.10.y] bpf: Enforce expected_attach_type for tailcall compatibility

[PATCH 5.10.y] bpf: Enforce expected_attach_type for tailcall compatibility

[Miguel Gazquez (Schneider Electric)]

From: Daniel Borkmann <daniel@iogearbox.net>

[ Upstream commit 4540aed51b12bc13364149bf95f6ecef013197c0 ]

Yinhao et al. recently reported:

  Our fuzzer tool discovered an uninitialized pointer issue in the
  bpf_prog_test_run_xdp() function within the Linux kernel's BPF subsystem.
  This leads to a NULL pointer dereference when a BPF program attempts to
  deference the txq member of struct xdp_buff object.

The test initializes two programs of BPF_PROG_TYPE_XDP: progA acts as the
entry point for bpf_prog_test_run_xdp() and its expected_attach_type can
neither be of be BPF_XDP_DEVMAP nor BPF_XDP_CPUMAP. progA calls into a slot
of a tailcall map it owns. progB's expected_attach_type must be BPF_XDP_DEVMAP
to pass xdp_is_valid_access() validation. The program returns struct xdp_md's
egress_ifindex, and the latter is only allowed to be accessed under mentioned
expected_attach_type. progB is then inserted into the tailcall which progA
calls.

The underlying issue goes beyond XDP though. Another example are programs
of type BPF_PROG_TYPE_CGROUP_SOCK_ADDR. sock_addr_is_valid_access() as well
as sock_addr_func_proto() have different logic depending on the programs'
expected_attach_type. Similarly, a program attached to BPF_CGROUP_INET4_GETPEERNAME
should not be allowed doing a tailcall into a program which calls bpf_bind()
out of BPF which is only enabled for BPF_CGROUP_INET4_CONNECT.

In short, specifying expected_attach_type allows to open up additional
functionality or restrictions beyond what the basic bpf_prog_type enables.
The use of tailcalls must not violate these constraints. Fix it by enforcing
expected_attach_type in __bpf_prog_map_compatible().

Note that we only enforce this for tailcall maps, but not for BPF devmaps or
cpumaps: There, the programs are invoked through dev_map_bpf_prog_run*() and
cpu_map_bpf_prog_run*() which set up a new environment / context and therefore
these situations are not prone to this issue.

[ Fixed conflict, applied the changes to bpf_prog_array_compatible
instead of __bpf_prog_map_compatible. Dropped the guard testing for
BPF_MAP_TYPE_PROG_ARRAY as bpf_prog_array_compatible is only called on maps whose
map_type is already BPF_MAP_TYPE_PROG_ARRAY, so the check was always true ]

Fixes: 5e43f899b03a ("bpf: Check attach type at prog load time")
Reported-by: Yinhao Hu <dddddd@hust.edu.cn>
Reported-by: Kaiyan Mei <M202472210@hust.edu.cn>
Reviewed-by: Dongliang Mu <dzm91@hust.edu.cn>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/r/20250926171201.188490-1-daniel@iogearbox.net
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Signed-off-by: Miguel Gazquez (Schneider Electric) <miguel.gazquez@bootlin.com>
---
 include/linux/bpf.h | 1 +
 kernel/bpf/core.c   | 4 ++++
 2 files changed, 5 insertions(+)

diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index 5d5d0bc7ca50..fb213524ccac 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -876,6 +876,7 @@ struct bpf_array_aux {
 		spinlock_t lock;
 		enum bpf_prog_type type;
 		bool jited;
+		enum bpf_attach_type expected_attach_type;
 	} owner;
 	/* Programs with direct jumps into programs part of this array. */
 	struct list_head poke_progs;
diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c
index db613a97ee5f..c9b3fe8b250a 100644
--- a/kernel/bpf/core.c
+++ b/kernel/bpf/core.c
@@ -1795,10 +1795,14 @@ bool bpf_prog_array_compatible(struct bpf_array *array,
 		 */
 		array->aux->owner.type  = fp->type;
 		array->aux->owner.jited = fp->jited;
+		array->aux->owner.expected_attach_type = fp->expected_attach_type;
 		ret = true;
 	} else {
 		ret = array->aux->owner.type  == fp->type &&
 		      array->aux->owner.jited == fp->jited;
+		if (ret &&
+		    array->aux->owner.expected_attach_type != fp->expected_attach_type)
+			ret = false;
 	}
 	spin_unlock(&array->aux->owner.lock);
 	return ret;

---
base-commit: 2a3da1f4966798b0b48ce302944ad356b2c98b5d
change-id: 20260907-cve-2025-40123-e08968057032

Best regards,
--  
Miguel Gazquez (Schneider Electric) <miguel.gazquez@bootlin.com>

Re: [PATCH 5.10.y] bpf: Enforce expected_attach_type for tailcall compatibility

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] Incomplete fix for tailcall expected_attach_type bypass: BPF_PROG_TYPE_EXT programs bypass the check.
--

commit d87774423a605def68b6715a3edf5dc1feea5da7
Author: Daniel Borkmann <daniel@iogearbox.net>

bpf: Enforce expected_attach_type for tailcall compatibility

This patch aims to prevent context confusion by enforcing that programs
inserted into a tailcall map share the same expected_attach_type.
This restricts tailcalls from violating attach type constraints.

Link: https://lore.kernel.org/r/20250926171201.188490-1-daniel@iogearbox.net

> diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c
> --- a/kernel/bpf/core.c
> +++ b/kernel/bpf/core.c
[ ... ]
> @@ -1795,10 +1795,14 @@ bool bpf_prog_array_compatible(struct bpf_array *array,
[ ... ]
>  	} else {
>  		ret = array->aux->owner.type  == fp->type &&
>  		      array->aux->owner.jited == fp->jited;
> +		if (ret &&
> +		    array->aux->owner.expected_attach_type != fp->expected_attach_type)
> +			ret = false;

[Severity: High]
This is a pre-existing issue, but does this check effectively become 0 != 0
for all BPF_PROG_TYPE_EXT programs, allowing arbitrary mixing of EXT programs
with different underlying target contexts?

It appears that for BPF_PROG_TYPE_EXT programs, expected_attach_type is
reset to 0 at the end of verification in bpf_check().

If userspace loads two BPF_PROG_TYPE_EXT programs targeting entirely
different functions (e.g., XDP_DEVMAP vs XDP_CPUMAP) and inserts them both
into the same BPF_MAP_TYPE_PROG_ARRAY, couldn't a tailcall between them
bypass this expected_attach_type check?

Can this allow a program verified for one target context to execute in a
different context, potentially leading to out-of-bounds accesses or NULL
pointer dereferences?

>  	}
>  	spin_unlock(&array->aux->owner.lock);
>  	return ret;

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260908-cve-2025-40123-v1-1-be62451196fb@bootlin.com?part=1

Re: [PATCH 5.10.y] bpf: Enforce expected_attach_type for tailcall compatibility

[Sasha Levin]

> The underlying issue goes beyond XDP though. Another example are programs
> of type BPF_PROG_TYPE_CGROUP_SOCK_ADDR. sock_addr_is_valid_access() as well
> as sock_addr_func_proto() have different logic depending on the programs'
> expected_attach_type.

Thanks for the 5.10.y backport of 4540aed51b12 ("bpf: Enforce
expected_attach_type for tailcall compatibility"). Holding this until a 5.15.y
version exists too, since fixes need to land on newer trees first.

-- 
Thanks,
Sasha

Re: [PATCH 5.10.y] bpf: Enforce expected_attach_type for tailcall compatibility

[Miguel Gazquez]

Hi

Le 09/09/2026 à 22:26, Sasha Levin a écrit :
>> The underlying issue goes beyond XDP though. Another example are programs
>> of type BPF_PROG_TYPE_CGROUP_SOCK_ADDR. sock_addr_is_valid_access() as well
>> as sock_addr_func_proto() have different logic depending on the programs'
>> expected_attach_type.
> 
> Thanks for the 5.10.y backport of 4540aed51b12 ("bpf: Enforce
> expected_attach_type for tailcall compatibility"). Holding this until a 5.15.y
> version exists too, since fixes need to land on newer trees first.
> 

Sorry I missed that there was no 5.15 backport, I sent it.

-- 
Miguel Gazquez, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com