netdev - [PATCH AUTOSEL 6.15 048/118] bpf: Pass the same orig

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [day] [month] [year] [list]
Message-Id: <20250604005049.4147522-48-sashal@kernel.org>
Date: Tue,  3 Jun 2025 20:49:39 -0400
From: Sasha Levin <sashal@...nel.org>
To: patches@...ts.linux.dev,
	stable@...r.kernel.org
Cc: Ilya Leoshkevich <iii@...ux.ibm.com>,
	Martin KaFai Lau <martin.lau@...nel.org>,
	Alexei Starovoitov <ast@...nel.org>,
	Sasha Levin <sashal@...nel.org>,
	martin.lau@...ux.dev,
	daniel@...earbox.net,
	andrii@...nel.org,
	bpf@...r.kernel.org,
	netdev@...r.kernel.org
Subject: [PATCH AUTOSEL 6.15 048/118] bpf: Pass the same orig_call value to trampoline functions

From: Ilya Leoshkevich <iii@...ux.ibm.com>

[ Upstream commit 94bde253d3ae5d8a01cb958663b12daef1d06574 ]

There is currently some confusion in the s390x JIT regarding whether
orig_call can be NULL and what that means. Originally the NULL value
was used to distinguish the struct_ops case, but this was superseded by
BPF_TRAMP_F_INDIRECT (see commit 0c970ed2f87c ("s390/bpf: Fix indirect
trampoline generation").

The remaining reason to have this check is that NULL can actually be
passed to the arch_bpf_trampoline_size() call - but not to the
respective arch_prepare_bpf_trampoline()! call - by
bpf_struct_ops_prepare_trampoline().

Remove this asymmetry by passing stub_func to both functions, so that
JITs may rely on orig_call never being NULL.

Signed-off-by: Ilya Leoshkevich <iii@...ux.ibm.com>
Acked-by: Martin KaFai Lau <martin.lau@...nel.org>
Link: https://lore.kernel.org/r/20250512221911.61314-2-iii@linux.ibm.com
Signed-off-by: Alexei Starovoitov <ast@...nel.org>
Signed-off-by: Sasha Levin <sashal@...nel.org>
---

**YES** This commit should be backported to stable kernel trees. ##
Analysis **Nature of the Fix:** This commit addresses an asymmetry in
the BPF trampoline infrastructure where `NULL` could be passed to
`arch_bpf_trampoline_size()` but not to `arch_prepare_bpf_trampoline()`.
The fix ensures that `stub_func` is consistently passed to both
functions, allowing JIT implementations to rely on `orig_call` never
being `NULL`. **Code Changes Analysis:** The change is minimal and
surgical - only one line in `kernel/bpf/bpf_struct_ops.c`: ```c - size =
arch_bpf_trampoline_size(model, flags, tlinks, NULL); + size =
arch_bpf_trampoline_size(model, flags, tlinks, stub_func); ``` This
passes `stub_func` instead of `NULL` to `arch_bpf_trampoline_size()`,
creating consistency with the `arch_prepare_bpf_trampoline()` call on
line 620 which already receives `stub_func`. **Why This Should Be
Backported:** 1. **Fixes Architectural Inconsistency:** Based on the
repository analysis, this addresses confusion in JIT implementations
(particularly s390x) about when `orig_call` can be `NULL` and what that
signifies. 2. **Prevents Potential Crashes:** The repository history
shows that similar asymmetries in BPF trampoline handling caused crashes
on architectures like RISC-V and incorrect code generation on s390x. 3.
**Minimal Risk:** The change is extremely contained - it only affects
the parameter passed to `arch_bpf_trampoline_size()` in the struct_ops
path. Since this function is used for size calculation, passing a valid
function pointer instead of `NULL` should not break existing
functionality. 4. **Follows Stable Tree Criteria:** - **Important
bugfix:** Prevents JIT confusion and potential incorrect behavior -
**Minimal risk:** Single line change with clear semantics - **Confined
to subsystem:** Only affects BPF struct_ops trampoline generation - **No
architectural changes:** Does not modify core BPF infrastructure 5.
**Related Historical Precedent:** Looking at the similar commits in the
analysis, commit #3 (s390/bpf: Let arch_prepare_bpf_trampoline return
program size) was marked "YES" for backporting, and it was a similar
cleanup/consistency fix for the BPF trampoline infrastructure. 6.
**Prevents Future Issues:** This fix eliminates a source of confusion
for JIT maintainers and ensures all architectures can implement
consistent `NULL` checking logic. The fix aligns with the principle
established in commit 0c970ed2f87c that JITs should use the
`BPF_TRAMP_F_INDIRECT` flag rather than checking for `NULL` parameters,
and this change supports that by ensuring parameters are never `NULL` in
the first place.

 kernel/bpf/bpf_struct_ops.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/bpf/bpf_struct_ops.c b/kernel/bpf/bpf_struct_ops.c
index db13ee70d94d5..96113633e391a 100644
--- a/kernel/bpf/bpf_struct_ops.c
+++ b/kernel/bpf/bpf_struct_ops.c
@@ -601,7 +601,7 @@ int bpf_struct_ops_prepare_trampoline(struct bpf_tramp_links *tlinks,
 	if (model->ret_size > 0)
 		flags |= BPF_TRAMP_F_RET_FENTRY_RET;
 
-	size = arch_bpf_trampoline_size(model, flags, tlinks, NULL);
+	size = arch_bpf_trampoline_size(model, flags, tlinks, stub_func);
 	if (size <= 0)
 		return size ? : -EFAULT;
 
-- 
2.39.5