[PATCH bpf-next v3] bpf: verifier: Move desc->imm setup to sort_kfunc_descs_by_imm_off()

[PATCH bpf-next v3] bpf: verifier: Move desc->imm setup to sort_kfunc_descs_by_imm_off()

[Puranjay Mohan]

Metadata about a kfunc call is added to the kfunc_tab in
add_kfunc_call() but the call instruction itself could get removed by
opt_remove_dead_code() later if it is not reachable.

If the call instruction is removed, specialize_kfunc() is never called
for it and the desc->imm in the kfunc_tab is never initialized for this
kfunc call. In this case, sort_kfunc_descs_by_imm_off(env->prog); in
do_misc_fixups() doesn't sort the table correctly.
This is a problem for s390 as its JIT uses this table to find the
addresses for kfuncs, and if this table is not sorted properly, JIT may
fail to find addresses for valid kfunc calls.

This was exposed by:

commit d869d56ca848 ("bpf: verifier: refactor kfunc specialization")

as before this commit, desc->imm was initialised in add_kfunc_call()
which happens before dead code elimination.

Move desc->imm setup down to sort_kfunc_descs_by_imm_off(), this fixes
the problem and also saves us from having the same logic in
add_kfunc_call() and specialize_kfunc().

Suggested-by: Eduard Zingerman <eddyz87@gmail.com>
Signed-off-by: Puranjay Mohan <puranjay@kernel.org>
---

Changes in v2->v3:
v2: https://lore.kernel.org/all/20251113104053.18107-1-puranjay@kernel.org/
- Move desc->imm setup down to sort_kfunc_descs_by_imm_off() (Eduard)

Changes in v1->v2:
v1: https://lore.kernel.org/all/20251111160949.45623-1-puranjay@kernel.org/
- Removed fixes tag as the broken commit is not upstream yet.
- Initialize desc->imm with the correct value for both with and without
  bpf_jit_supports_far_kfunc_call() for completeness.
- Don't re-initialize desc->imm to func_id in specialize_kfunc() as it
  it already have that value, it only needs to be updated in the
  !bpf_jit_supports_far_kfunc_call() case where the imm can change.

This bug is not triggered by the CI currently, I am working on another
set for non-sleepbale arena allocations and as part of that I am adding
a new selftest that triggers this bug.

Selftest: https://github.com/kernel-patches/bpf/pull/10242/commits/1f681f022c6d685fd76695e5eafbe9d9ab4c0002
CI run: https://github.com/kernel-patches/bpf/actions/runs/19238699806/job/54996376908

---
 kernel/bpf/verifier.c | 54 ++++++++++++++++++++++++++++---------------
 1 file changed, 35 insertions(+), 19 deletions(-)

diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 1268fa075d4c..5f6f50f7116b 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -3391,16 +3391,43 @@ static int kfunc_desc_cmp_by_imm_off(const void *a, const void *b)
 	return 0;
 }
 
-static void sort_kfunc_descs_by_imm_off(struct bpf_prog *prog)
+static int set_kfunc_desc_imm(struct bpf_verifier_env *env, struct bpf_kfunc_desc *desc)
+{
+	unsigned long call_imm;
+
+	if (bpf_jit_supports_far_kfunc_call()) {
+		call_imm = desc->func_id;
+	} else {
+		call_imm = BPF_CALL_IMM(desc->addr);
+		/* Check whether the relative offset overflows desc->imm */
+		if ((unsigned long)(s32)call_imm != call_imm) {
+			verbose(env, "address of kernel func_id %u is out of range\n",
+				desc->func_id);
+			return -EINVAL;
+		}
+	}
+	desc->imm = call_imm;
+	return 0;
+}
+
+static int sort_kfunc_descs_by_imm_off(struct bpf_verifier_env *env)
 {
 	struct bpf_kfunc_desc_tab *tab;
+	int i, err;
 
-	tab = prog->aux->kfunc_tab;
+	tab = env->prog->aux->kfunc_tab;
 	if (!tab)
-		return;
+		return 0;
+
+	for (i = 0; i < tab->nr_descs; i++) {
+		err = set_kfunc_desc_imm(env, &tab->descs[i]);
+		if (err)
+			return err;
+	}
 
 	sort(tab->descs, tab->nr_descs, sizeof(tab->descs[0]),
 	     kfunc_desc_cmp_by_imm_off, NULL);
+	return 0;
 }
 
 bool bpf_prog_has_kfunc_call(const struct bpf_prog *prog)
@@ -22320,10 +22347,10 @@ static int specialize_kfunc(struct bpf_verifier_env *env, struct bpf_kfunc_desc
 	bool is_rdonly;
 	u32 func_id = desc->func_id;
 	u16 offset = desc->offset;
-	unsigned long addr = desc->addr, call_imm;
+	unsigned long addr = desc->addr;
 
 	if (offset) /* return if module BTF is used */
-		goto set_imm;
+		return 0;
 
 	if (bpf_dev_bound_kfunc_id(func_id)) {
 		xdp_kfunc = bpf_dev_bound_resolve_kfunc(prog, func_id);
@@ -22351,19 +22378,6 @@ static int specialize_kfunc(struct bpf_verifier_env *env, struct bpf_kfunc_desc
 		if (!env->insn_aux_data[insn_idx].non_sleepable)
 			addr = (unsigned long)bpf_dynptr_from_file_sleepable;
 	}
-
-set_imm:
-	if (bpf_jit_supports_far_kfunc_call()) {
-		call_imm = func_id;
-	} else {
-		call_imm = BPF_CALL_IMM(addr);
-		/* Check whether the relative offset overflows desc->imm */
-		if ((unsigned long)(s32)call_imm != call_imm) {
-			verbose(env, "address of kernel func_id %u is out of range\n", func_id);
-			return -EINVAL;
-		}
-	}
-	desc->imm = call_imm;
 	desc->addr = addr;
 	return 0;
 }
@@ -23441,7 +23455,9 @@ static int do_misc_fixups(struct bpf_verifier_env *env)
 		}
 	}
 
-	sort_kfunc_descs_by_imm_off(env->prog);
+	ret = sort_kfunc_descs_by_imm_off(env);
+	if (ret)
+		return ret;
 
 	return 0;
 }
-- 
2.47.3

Re: [PATCH bpf-next v3] bpf: verifier: Move desc->imm setup to sort_kfunc_descs_by_imm_off()

[Eduard Zingerman]

On Fri, 2025-11-14 at 15:40 +0000, Puranjay Mohan wrote:
> Metadata about a kfunc call is added to the kfunc_tab in
> add_kfunc_call() but the call instruction itself could get removed by
> opt_remove_dead_code() later if it is not reachable.
> 
> If the call instruction is removed, specialize_kfunc() is never called
> for it and the desc->imm in the kfunc_tab is never initialized for this
> kfunc call. In this case, sort_kfunc_descs_by_imm_off(env->prog); in
> do_misc_fixups() doesn't sort the table correctly.
> This is a problem for s390 as its JIT uses this table to find the
> addresses for kfuncs, and if this table is not sorted properly, JIT may
> fail to find addresses for valid kfunc calls.
> 
> This was exposed by:
> 
> commit d869d56ca848 ("bpf: verifier: refactor kfunc specialization")
> 
> as before this commit, desc->imm was initialised in add_kfunc_call()
> which happens before dead code elimination.
> 
> Move desc->imm setup down to sort_kfunc_descs_by_imm_off(), this fixes
> the problem and also saves us from having the same logic in
> add_kfunc_call() and specialize_kfunc().
> 
> Suggested-by: Eduard Zingerman <eddyz87@gmail.com>
> Signed-off-by: Puranjay Mohan <puranjay@kernel.org>
> ---

Acked-by: Eduard Zingerman <eddyz87@gmail.com>

[...]

Re: [PATCH bpf-next v3] bpf: verifier: Move desc->imm setup to sort_kfunc_descs_by_imm_off()

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to bpf/bpf-next.git (master)
by Alexei Starovoitov <ast@kernel.org>:

On Fri, 14 Nov 2025 15:40:22 +0000 you wrote:
> Metadata about a kfunc call is added to the kfunc_tab in
> add_kfunc_call() but the call instruction itself could get removed by
> opt_remove_dead_code() later if it is not reachable.
> 
> If the call instruction is removed, specialize_kfunc() is never called
> for it and the desc->imm in the kfunc_tab is never initialized for this
> kfunc call. In this case, sort_kfunc_descs_by_imm_off(env->prog); in
> do_misc_fixups() doesn't sort the table correctly.
> This is a problem for s390 as its JIT uses this table to find the
> addresses for kfuncs, and if this table is not sorted properly, JIT may
> fail to find addresses for valid kfunc calls.
> 
> [...]

Here is the summary with links:
  - [bpf-next,v3] bpf: verifier: Move desc->imm setup to sort_kfunc_descs_by_imm_off()
    https://git.kernel.org/bpf/bpf-next/c/4f7bc83b9837

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