Skip to contentSkip to conversation

IRQ save/restore

PATCH Superseded Kumar Kartikeya Dwivedi 3 participants 13 messages Nov 29, 2024 – Dec 3, 2024

Subsystems (4)

v5 of 6 Superseded Latest is v6 →

lore
Share & tools
lore

Kumar Kartikeya Dwivedi

author maintainer
#0 lore raw

v5 cover IRQ save/restore

This set introduces support for managing IRQ state from BPF programs. Two new kfuncs, bpf_local_irq_save, and bpf_local_irq_restore are introduced to enable this functionality.

Intended use cases are writing IRQ safe data structures (e.g. memory allocator) in BPF programs natively, and use in new spin locking primitives intended to be introduced in the next few weeks.

The set begins with some refactoring patches before the actual functionality is introduced. Patch 1 consolidates all resource related state in bpf_verifier_state, and moves it out from bpf_func_state.

Patch 2 refactor acquire and release functions for reference state to make them reusable without duplication for other resource types.

After this, patch 3 refactors stack slot liveness marking logic to be shared between dynptr, and iterators, in preparation for introducing same logic for irq flag object on stack.

Finally, patch 4 and 7 introduce the new kfuncs and their selftests. For more details, please inspect the patch commit logs. Patch 5 makes the error message in case of resource leaks under BPF_EXIT a bit clearer. Patch 6 expands coverage of existing preempt-disable selftest to cover sleepable kfuncs.

See individual patches for more details.

Changelog:
----------
v4 -> v5
v4: https://lore.kernel.org/bpf/[email protected]

* Do regno - 1 when printing argument
* Pass verifier state explicitly into print_{insn,verifier}_state (Eduard)
* Pass frameno instead of bpf_func_state (Eduard)
* Move bpf_reference_state *refs after parent to fill two holes in bpf_verifier_state (Eduard). The hunk fixing that bug is in the commit adding IRQ save/restore kfuncs, as it is only needed then.
* Fix bug in release_reference_state breaking stack property (Eduard)
* Add selftest for triggering and reproducing bug found by Eduard irq_ooo_refs_array in final patch
* Print insn_idx and active_irq_id on error (Eduard)
* Add more acks

v3 -> v4
v3: https://lore.kernel.org/bpf/[email protected]

* Add yet another missing kfunc declaration to silence s390 CI

v2 -> v3
v2: https://lore.kernel.org/bpf/[email protected]

* Drop REF_TYPE_LOCK_MASK
* Add kfunc declarations to selftest to silence s390 CI errors

v1 -> v2
v1: https://lore.kernel.org/bpf/[email protected]

* Drop reference -> resource renaming in the verifier (Eduard, Alexei)
* Change verifier log for check_resource_leak for BPF_EXIT (Eduard)
* Remove id parameter from acquire_resource_state, read s->id (Eduard)
* Rename erase to release for reference state (Eduard)
* Move resource state to bpf_verifier_state (Eduard, Alexei)
* Drop unnecessary casting to/from u64 in helpers (Eduard)
* Add test for arg != PTR_TO_STACK (Eduard)
* Drop now redundant tests (Eduard)
* Address some other misc nits
* Add Reviewed-by and Acked-by from Eduard

Kumar Kartikeya Dwivedi (7):
bpf: Consolidate locks and reference state in verifier state
bpf: Refactor {acquire,release}_reference_state
bpf: Refactor mark_{dynptr,iter}_read
bpf: Introduce support for bpf_local_irq_{save,restore}
bpf: Improve verifier log for resource leak on exit selftests/bpf: Expand coverage of preempt tests to sleepable kfunc selftests/bpf: Add IRQ save/restore tests

 include/linux/bpf_verifier.h                  |  27 +-
 kernel/bpf/helpers.c                          |  17 +
 kernel/bpf/log.c                              |  21 +-
 kernel/bpf/verifier.c                         | 573 +++++++++++++-----
 .../selftests/bpf/prog_tests/verifier.c       |   2 +
 .../selftests/bpf/progs/exceptions_fail.c     |   4 +-
 tools/testing/selftests/bpf/progs/irq.c       | 444 ++++++++++++++
 .../selftests/bpf/progs/preempt_lock.c        |  28 +-
 .../selftests/bpf/progs/verifier_spin_lock.c  |   2 +-
 9 files changed, 950 insertions(+), 168 deletions(-)
 create mode 100644 tools/testing/selftests/bpf/progs/irq.c

base-commit: c8d02b547363880d996f80c38cc8b997c7b90725

Signature · 1 line
2.43.5
author maintainer
#1 lore raw

PATCH v5 1/7 bpf: Consolidate locks and reference state in verifier state

Currently, state for RCU read locks and preemption is in bpf_verifier_state, while locks and pointer reference state remains in bpf_func_state. There is no particular reason to keep the latter in bpf_func_state. Additionally, it is copied into a new frame's state and copied back to the caller frame's state everytime the verifier processes a pseudo call instruction. This is a bit wasteful, given this state is global for a given verification state / path.

Move all resource and reference related state in bpf_verifier_state structure in this patch, in preparation for introducing new reference state types in the future.

Since we switch print_verifier_state and friends to print using vstate, we now need to explicitly pass in the verifier state from the caller along with the bpf_func_state, so modify the prototype and callers to do so. To ensure func state matches the verifier state when we're printing data, take in frame number instead of bpf_func_state pointer instead and avoid inconsistencies induced by the caller.

---
 include/linux/bpf_verifier.h |  19 +++--
 kernel/bpf/log.c             |  20 ++---
 kernel/bpf/verifier.c        | 140 +++++++++++++++++------------------
 3 files changed, 88 insertions(+), 91 deletions(-)
include/linux/bpf_verifier.h +11 -8
index f4290c179bee..03e351c43fa8 100644
@@ -315,9 +315,6 @@ struct bpf_func_state {
 	u32 callback_depth;
 
 	/* The following fields should be last. See copy_func_state() */
-	int acquired_refs;
-	int active_locks;
-	struct bpf_reference_state *refs;
 	/* The state of the stack. Each element of the array describes BPF_REG_SIZE
 	 * (i.e. 8) bytes worth of stack memory.
 	 * stack[0] represents bytes [*(r10-8)..*(r10-1)]
@@ -370,6 +367,8 @@ struct bpf_verifier_state {
 	/* call stack tracking */
 	struct bpf_func_state *frame[MAX_CALL_FRAMES];
 	struct bpf_verifier_state *parent;
+	/* Acquired reference states */
+	struct bpf_reference_state *refs;
 	/*
 	 * 'branches' field is the number of branches left to explore:
 	 * 0 - all possible paths from this state reached bpf_exit or
@@ -419,9 +418,12 @@ struct bpf_verifier_state {
 	u32 insn_idx;
 	u32 curframe;
 
-	bool speculative;
+	u32 acquired_refs;
+	u32 active_locks;
+	u32 active_preempt_locks;
 	bool active_rcu_lock;
-	u32 active_preempt_lock;
+
+	bool speculative;
 	/* If this state was ever pointed-to by other state's loop_entry field
 	 * this flag would be set to true. Used to avoid freeing such states
 	 * while they are still in use.
@@ -979,8 +981,9 @@ const char *dynptr_type_str(enum bpf_dynptr_type type);
 const char *iter_type_str(const struct btf *btf, u32 btf_id);
 const char *iter_state_str(enum bpf_iter_state state);
 
-void print_verifier_state(struct bpf_verifier_env *env,
-			  const struct bpf_func_state *state, bool print_all);
-void print_insn_state(struct bpf_verifier_env *env, const struct bpf_func_state *state);
+void print_verifier_state(struct bpf_verifier_env *env, const struct bpf_verifier_state *vstate,
+			  u32 frameno, bool print_all);
+void print_insn_state(struct bpf_verifier_env *env, const struct bpf_verifier_state *vstate,
+		      u32 frameno);
 
 #endif /* _LINUX_BPF_VERIFIER_H */
kernel/bpf/log.c +11 -9
index 4a858fdb6476..2d28ce926053 100644
@@ -753,9 +753,10 @@ static void print_reg_state(struct bpf_verifier_env *env,
 	verbose(env, ")");
 }
 
-void print_verifier_state(struct bpf_verifier_env *env, const struct bpf_func_state *state,
-			  bool print_all)
+void print_verifier_state(struct bpf_verifier_env *env, const struct bpf_verifier_state *vstate,
+			  u32 frameno, bool print_all)
 {
+	const struct bpf_func_state *state = vstate->frame[frameno];
 	const struct bpf_reg_state *reg;
 	int i;
 
@@ -843,11 +844,11 @@ void print_verifier_state(struct bpf_verifier_env *env, const struct bpf_func_st
 			break;
 		}
 	}
-	if (state->acquired_refs && state->refs[0].id) {
-		verbose(env, " refs=%d", state->refs[0].id);
-		for (i = 1; i < state->acquired_refs; i++)
-			if (state->refs[i].id)
-				verbose(env, ",%d", state->refs[i].id);
+	if (vstate->acquired_refs && vstate->refs[0].id) {
+		verbose(env, " refs=%d", vstate->refs[0].id);
+		for (i = 1; i < vstate->acquired_refs; i++)
+			if (vstate->refs[i].id)
+				verbose(env, ",%d", vstate->refs[i].id);
 	}
 	if (state->in_callback_fn)
 		verbose(env, " cb");
@@ -864,7 +865,8 @@ static inline u32 vlog_alignment(u32 pos)
 			BPF_LOG_MIN_ALIGNMENT) - pos - 1;
 }
 
-void print_insn_state(struct bpf_verifier_env *env, const struct bpf_func_state *state)
+void print_insn_state(struct bpf_verifier_env *env, const struct bpf_verifier_state *vstate,
+		      u32 frameno)
 {
 	if (env->prev_log_pos && env->prev_log_pos == env->log.end_pos) {
 		/* remove new line character */
@@ -873,5 +875,5 @@ void print_insn_state(struct bpf_verifier_env *env, const struct bpf_func_state
 	} else {
 		verbose(env, "%d:", env->insn_idx);
 	}
-	print_verifier_state(env, state, false);
+	print_verifier_state(env, vstate, frameno, false);
 }
kernel/bpf/verifier.c +66 -74

444 lines not on this page · raw message

Signature · 1 line
2.43.5
author maintainer
#2 lore raw

PATCH v5 2/7 bpf: Refactor {acquire,release}_reference_state

In reply to Kumar Kartikeya Dwivedi

In preparation for introducing support for more reference types which have to add and remove reference state, refactor the acquire_reference_state and release_reference_state functions to share common logic.

The acquire_reference_state function simply handles growing the acquired refs and returning the pointer to the new uninitialized element, which can be filled in by the caller.

The release_reference_state function simply erases a reference state entry in the acquired_refs array and shrinks it. The callers are responsible for finding the suitable element by matching on various fields of the reference state and requesting deletion through this function. It is not supposed to be called directly.

Existing callers of release_reference_state were using it to find and remove state for a given ref_obj_id without scrubbing the associated registers in the verifier state. Introduce release_reference_nomark to provide this functionality and convert callers. We now use this new release_reference_nomark function within release_reference as well. It needs to operate on a verifier state instead of taking verifier env as mark_ptr_or_null_regs requires operating on verifier state of the two branches of a NULL condition check, therefore env->cur_state cannot be used directly.

---
 kernel/bpf/verifier.c | 110 +++++++++++++++++++++++-------------------
 1 file changed, 60 insertions(+), 50 deletions(-)
kernel/bpf/verifier.c +60 -50
index 019c56c782a2..91bcd84fabff 100644
@@ -196,7 +196,8 @@ struct bpf_verifier_stack_elem {
 
 #define BPF_PRIV_STACK_MIN_SIZE		64
 
-static int acquire_reference_state(struct bpf_verifier_env *env, int insn_idx);
+static int acquire_reference(struct bpf_verifier_env *env, int insn_idx);
+static int release_reference_nomark(struct bpf_verifier_state *state, int ref_obj_id);
 static int release_reference(struct bpf_verifier_env *env, int ref_obj_id);
 static void invalidate_non_owning_refs(struct bpf_verifier_env *env);
 static bool in_rbtree_lock_required_cb(struct bpf_verifier_env *env);
@@ -771,7 +772,7 @@ static int mark_stack_slots_dynptr(struct bpf_verifier_env *env, struct bpf_reg_
 		if (clone_ref_obj_id)
 			id = clone_ref_obj_id;
 		else
-			id = acquire_reference_state(env, insn_idx);
+			id = acquire_reference(env, insn_idx);
 
 		if (id < 0)
 			return id;
@@ -1033,7 +1034,7 @@ static int mark_stack_slots_iter(struct bpf_verifier_env *env,
 	if (spi < 0)
 		return spi;
 
-	id = acquire_reference_state(env, insn_idx);
+	id = acquire_reference(env, insn_idx);
 	if (id < 0)
 		return id;
 
@@ -1349,77 +1350,69 @@ static int grow_stack_state(struct bpf_verifier_env *env, struct bpf_func_state
  * On success, returns a valid pointer id to associate with the register
  * On failure, returns a negative errno.
  */
-static int acquire_reference_state(struct bpf_verifier_env *env, int insn_idx)
+static struct bpf_reference_state *acquire_reference_state(struct bpf_verifier_env *env, int insn_idx, bool gen_id)
 {
 	struct bpf_verifier_state *state = env->cur_state;
 	int new_ofs = state->acquired_refs;
-	int id, err;
+	int err;
 
 	err = resize_reference_state(state, state->acquired_refs + 1);
 	if (err)
-		return err;
-	id = ++env->id_gen;
-	state->refs[new_ofs].type = REF_TYPE_PTR;
-	state->refs[new_ofs].id = id;
+		return NULL;
+	if (gen_id)
+		state->refs[new_ofs].id = ++env->id_gen;
 	state->refs[new_ofs].insn_idx = insn_idx;
 
-	return id;
+	return &state->refs[new_ofs];
+}
+
+static int acquire_reference(struct bpf_verifier_env *env, int insn_idx)
+{
+	struct bpf_reference_state *s;
+
+	s = acquire_reference_state(env, insn_idx, true);
+	if (!s)
+		return -ENOMEM;
+	s->type = REF_TYPE_PTR;
+	return s->id;
 }
 
 static int acquire_lock_state(struct bpf_verifier_env *env, int insn_idx, enum ref_state_type type,
 			      int id, void *ptr)
 {
 	struct bpf_verifier_state *state = env->cur_state;
-	int new_ofs = state->acquired_refs;
-	int err;
+	struct bpf_reference_state *s;
 
-	err = resize_reference_state(state, state->acquired_refs + 1);
-	if (err)
-		return err;
-	state->refs[new_ofs].type = type;
-	state->refs[new_ofs].id = id;
-	state->refs[new_ofs].insn_idx = insn_idx;
-	state->refs[new_ofs].ptr = ptr;
+	s = acquire_reference_state(env, insn_idx, false);
+	s->type = type;
+	s->id = id;
+	s->ptr = ptr;
 
 	state->active_locks++;
 	return 0;
 }
 
-/* release function corresponding to acquire_reference_state(). Idempotent. */
-static int release_reference_state(struct bpf_verifier_state *state, int ptr_id)
+static void release_reference_state(struct bpf_verifier_state *state, int idx)
 {
-	int i, last_idx;
+	int last_idx;
 
 	last_idx = state->acquired_refs - 1;
-	for (i = 0; i < state->acquired_refs; i++) {
-		if (state->refs[i].type != REF_TYPE_PTR)
-			continue;
-		if (state->refs[i].id == ptr_id) {
-			if (last_idx && i != last_idx)
-				memcpy(&state->refs[i], &state->refs[last_idx],
-				       sizeof(*state->refs));
-			memset(&state->refs[last_idx], 0, sizeof(*state->refs));
-			state->acquired_refs--;
-			return 0;
-		}
-	}
-	return -EINVAL;
+	if (last_idx && idx != last_idx)
+		memcpy(&state->refs[idx], &state->refs[last_idx], sizeof(*state->refs));
+	memset(&state->refs[last_idx], 0, sizeof(*state->refs));
+	state->acquired_refs--;
+	return;
 }
 
 static int release_lock_state(struct bpf_verifier_state *state, int type, int id, void *ptr)
 {
-	int i, last_idx;
+	int i;
 
-	last_idx = state->acquired_refs - 1;
 	for (i = 0; i < state->acquired_refs; i++) {
 		if (state->refs[i].type != type)
 			continue;
 		if (state->refs[i].id == id && state->refs[i].ptr == ptr) {
-			if (last_idx && i != last_idx)
-				memcpy(&state->refs[i], &state->refs[last_idx],
-				       sizeof(*state->refs));
-			memset(&state->refs[last_idx], 0, sizeof(*state->refs));
-			state->acquired_refs--;
+			release_reference_state(state, i);
 			state->active_locks--;
 			return 0;
 		}
@@ -9666,21 +9659,38 @@ static void mark_pkt_end(struct bpf_verifier_state *vstate, int regn, bool range
 		reg->range = AT_PKT_END;
 }
 
+static int release_reference_nomark(struct bpf_verifier_state *state, int ref_obj_id)
+{
+	int i;
+
+	for (i = 0; i < state->acquired_refs; i++) {
+		if (state->refs[i].type != REF_TYPE_PTR)
+			continue;
+		if (state->refs[i].id == ref_obj_id) {
+			release_reference_state(state, i);
+			return 0;
+		}
+	}
+	return -EINVAL;
+}
+
 /* The pointer with the specified id has released its reference to kernel
  * resources. Identify all copies of the same pointer and clear the reference.
+ *
+ * This is the release function corresponding to acquire_reference(). Idempotent.
  */
-static int release_reference(struct bpf_verifier_env *env,
-			     int ref_obj_id)
+static int release_reference(struct bpf_verifier_env *env, int ref_obj_id)
 {
+	struct bpf_verifier_state *vstate = env->cur_state;
 	struct bpf_func_state *state;
 	struct bpf_reg_state *reg;
 	int err;
 
-	err = release_reference_state(env->cur_state, ref_obj_id);
+	err = release_reference_nomark(vstate, ref_obj_id);
 	if (err)
 		return err;
 
-	bpf_for_each_reg_in_vstate(env->cur_state, state, reg, ({
+	bpf_for_each_reg_in_vstate(vstate, state, reg, ({
 		if (reg->ref_obj_id == ref_obj_id)
 			mark_reg_invalid(env, reg);
 	}));
@@ -10774,7 +10784,7 @@ static int check_helper_call(struct bpf_verifier_env *env, struct bpf_insn *insn
 			struct bpf_func_state *state;
 			struct bpf_reg_state *reg;
 
-			err = release_reference_state(env->cur_state, ref_obj_id);
+			err = release_reference_nomark(env->cur_state, ref_obj_id);
 			if (!err) {
 				bpf_for_each_reg_in_vstate(env->cur_state, state, reg, ({
 					if (reg->ref_obj_id == ref_obj_id) {
@@ -11107,7 +11117,7 @@ static int check_helper_call(struct bpf_verifier_env *env, struct bpf_insn *insn
 		/* For release_reference() */
 		regs[BPF_REG_0].ref_obj_id = meta.ref_obj_id;
 	} else if (is_acquire_function(func_id, meta.map_ptr)) {
-		int id = acquire_reference_state(env, insn_idx);
+		int id = acquire_reference(env, insn_idx);
 
 		if (id < 0)
 			return id;
@@ -13087,7 +13097,7 @@ static int check_kfunc_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
 		}
 		mark_btf_func_reg_size(env, BPF_REG_0, sizeof(void *));
 		if (is_kfunc_acquire(&meta)) {
-			int id = acquire_reference_state(env, insn_idx);
+			int id = acquire_reference(env, insn_idx);
 
 			if (id < 0)
 				return id;
@@ -15387,7 +15397,7 @@ static void mark_ptr_or_null_regs(struct bpf_verifier_state *vstate, u32 regno,
 		 * No one could have freed the reference state before
 		 * doing the NULL check.
 		 */
-		WARN_ON_ONCE(release_reference_state(vstate, id));
+		WARN_ON_ONCE(release_reference_nomark(vstate, id));
 
 	bpf_for_each_reg_in_vstate(vstate, state, reg, ({
 		mark_ptr_or_null_reg(state, reg, id, is_null);
Signature · 1 line
2.43.5
maintainer Acked-by on v5 2/7
#3 lore raw

Re: [PATCH bpf-next v5 2/7] bpf: Refactor {acquire,release}_reference_state

Quoted 25 lines · Kumar Kartikeya Dwivedi

In preparation for introducing support for more reference types which have to add and remove reference state, refactor the acquire_reference_state and release_reference_state functions to share common logic.

The acquire_reference_state function simply handles growing the acquired refs and returning the pointer to the new uninitialized element, which can be filled in by the caller.

The release_reference_state function simply erases a reference state entry in the acquired_refs array and shrinks it. The callers are responsible for finding the suitable element by matching on various fields of the reference state and requesting deletion through this function. It is not supposed to be called directly.

Existing callers of release_reference_state were using it to find and remove state for a given ref_obj_id without scrubbing the associated registers in the verifier state. Introduce release_reference_nomark to provide this functionality and convert callers. We now use this new release_reference_nomark function within release_reference as well. It needs to operate on a verifier state instead of taking verifier env as mark_ptr_or_null_regs requires operating on verifier state of the two branches of a NULL condition check, therefore env->cur_state cannot be used directly.

Quoted 2 lines

---

[...]

#4 lore raw

Re: [PATCH bpf-next v5 2/7] bpf: Refactor {acquire,release}_reference_state

In reply to Kumar Kartikeya Dwivedi
Quoted 16 lines · Kumar Kartikeya Dwivedi
kernel/bpf/verifier.c +5 -5
+static struct bpf_reference_state *acquire_reference_state(struct bpf_verifier_env *env, int insn_idx, bool gen_id)
 {
        struct bpf_verifier_state *state = env->cur_state;
        int new_ofs = state->acquired_refs;
-       int id, err;
+       int err;
 
        err = resize_reference_state(state, state->acquired_refs + 1);
        if (err)
-               return err;
-       id = ++env->id_gen;
-       state->refs[new_ofs].type = REF_TYPE_PTR;
-       state->refs[new_ofs].id = id;
+               return NULL;
+       if (gen_id)
+               state->refs[new_ofs].id = ++env->id_gen;

...

kernel/bpf/verifier.c static int grow_stack_state(struct bpf_verifier_env *env, struct bpf_func_state · line 1377

Quoted 9 lines
kernel/bpf/verifier.c +9 -0
+static int acquire_reference(struct bpf_verifier_env *env, int insn_idx)
+{
+       struct bpf_reference_state *s;
+
+       s = acquire_reference_state(env, insn_idx, true);
+       if (!s)
+               return -ENOMEM;
+       s->type = REF_TYPE_PTR;
+       return s->id;

Small nit.
I think 'bool gen_id' is not very readable, since
the callsite is not obvious.
Let's drop the flag and instead do:
s->id = ++env->id_gen;
return s->id;

kernel/bpf/verifier.c static int grow_stack_state(struct bpf_verifier_env *env, struct bpf_func_state · line 1389

Quoted 4 lines
kernel/bpf/verifier.c +4 -0
+       s = acquire_reference_state(env, insn_idx, false);
+       s->type = type;
+       s->id = id;
+       s->ptr = ptr;

this bit will be easier to read too.

author maintainer
#5 lore raw

Re: [PATCH bpf-next v5 2/7] bpf: Refactor {acquire,release}_reference_state

Quoted 22 lines · Alexei Starovoitov
Quoted 16 lines · Kumar Kartikeya Dwivedi
kernel/bpf/verifier.c +5 -5
+static struct bpf_reference_state *acquire_reference_state(struct bpf_verifier_env *env, int insn_idx, bool gen_id)
 {
        struct bpf_verifier_state *state = env->cur_state;
        int new_ofs = state->acquired_refs;
-       int id, err;
+       int err;
 
        err = resize_reference_state(state, state->acquired_refs + 1);
        if (err)
-               return err;
-       id = ++env->id_gen;
-       state->refs[new_ofs].type = REF_TYPE_PTR;
-       state->refs[new_ofs].id = id;
+               return NULL;
+       if (gen_id)
+               state->refs[new_ofs].id = ++env->id_gen;

...

Quoted 16 lines

kernel/bpf/verifier.c static int grow_stack_state(struct bpf_verifier_env *env, struct bpf_func_state · line 1377

Quoted 9 lines
kernel/bpf/verifier.c +9 -0
+static int acquire_reference(struct bpf_verifier_env *env, int insn_idx)
+{
+       struct bpf_reference_state *s;
+
+       s = acquire_reference_state(env, insn_idx, true);
+       if (!s)
+               return -ENOMEM;
+       s->type = REF_TYPE_PTR;
+       return s->id;

Small nit.
I think 'bool gen_id' is not very readable, since
the callsite is not obvious.
Let's drop the flag and instead do:
s->id = ++env->id_gen;
return s->id;

Ok.

Quoted 7 lines

kernel/bpf/verifier.c static int grow_stack_state(struct bpf_verifier_env *env, struct bpf_func_state · line 1389

Quoted 5 lines
kernel/bpf/verifier.c +4 -0
+       s = acquire_reference_state(env, insn_idx, false);
+       s->type = type;
+       s->id = id;
+       s->ptr = ptr;

this bit will be easier to read too.

Ok.

author maintainer
#6 lore raw

PATCH v5 3/7 bpf: Refactor mark_{dynptr,iter}_read

In reply to Kumar Kartikeya Dwivedi

There is possibility of sharing code between mark_dynptr_read and mark_iter_read for updating liveness information of their stack slots. Consolidate common logic into mark_stack_slot_obj_read function in preparation for the next patch which needs the same logic for its own stack slots.

---
 kernel/bpf/verifier.c | 43 +++++++++++++++++++++----------------------
 1 file changed, 21 insertions(+), 22 deletions(-)
kernel/bpf/verifier.c +21 -22
index 91bcd84fabff..992992816308 100644
@@ -3192,10 +3192,27 @@ static int mark_reg_read(struct bpf_verifier_env *env,
 	return 0;
 }
 
-static int mark_dynptr_read(struct bpf_verifier_env *env, struct bpf_reg_state *reg)
+static int mark_stack_slot_obj_read(struct bpf_verifier_env *env, struct bpf_reg_state *reg,
+				    int spi, int nr_slots)
 {
 	struct bpf_func_state *state = func(env, reg);
-	int spi, ret;
+	int err, i;
+
+	for (i = 0; i < nr_slots; i++) {
+		struct bpf_reg_state *st = &state->stack[spi - i].spilled_ptr;
+
+		err = mark_reg_read(env, st, st->parent, REG_LIVE_READ64);
+		if (err)
+			return err;
+
+		mark_stack_slot_scratched(env, spi - i);
+	}
+	return 0;
+}
+
+static int mark_dynptr_read(struct bpf_verifier_env *env, struct bpf_reg_state *reg)
+{
+	int spi;
 
 	/* For CONST_PTR_TO_DYNPTR, it must have already been done by
 	 * check_reg_arg in check_helper_call and mark_btf_func_reg_size in
@@ -3210,31 +3227,13 @@ static int mark_dynptr_read(struct bpf_verifier_env *env, struct bpf_reg_state *
 	 * bounds and spi is the first dynptr slot. Simply mark stack slot as
 	 * read.
 	 */
-	ret = mark_reg_read(env, &state->stack[spi].spilled_ptr,
-			    state->stack[spi].spilled_ptr.parent, REG_LIVE_READ64);
-	if (ret)
-		return ret;
-	return mark_reg_read(env, &state->stack[spi - 1].spilled_ptr,
-			     state->stack[spi - 1].spilled_ptr.parent, REG_LIVE_READ64);
+	return mark_stack_slot_obj_read(env, reg, spi, BPF_DYNPTR_NR_SLOTS);
 }
 
 static int mark_iter_read(struct bpf_verifier_env *env, struct bpf_reg_state *reg,
 			  int spi, int nr_slots)
 {
-	struct bpf_func_state *state = func(env, reg);
-	int err, i;
-
-	for (i = 0; i < nr_slots; i++) {
-		struct bpf_reg_state *st = &state->stack[spi - i].spilled_ptr;
-
-		err = mark_reg_read(env, st, st->parent, REG_LIVE_READ64);
-		if (err)
-			return err;
-
-		mark_stack_slot_scratched(env, spi - i);
-	}
-
-	return 0;
+	return mark_stack_slot_obj_read(env, reg, spi, nr_slots);
 }
 
 /* This function is supposed to be used by the following 32-bit optimization
Signature · 1 line
2.43.5
author maintainer
#7 lore raw

PATCH v5 4/7 bpf: Introduce support for bpf_local_irq_{save,restore}

In reply to Kumar Kartikeya Dwivedi

Teach the verifier about IRQ-disabled sections through the introduction of two new kfuncs, bpf_local_irq_save, to save IRQ state and disable them, and bpf_local_irq_restore, to restore IRQ state and enable them back again.

For the purposes of tracking the saved IRQ state, the verifier is taught about a new special object on the stack of type STACK_IRQ_FLAG. This is a 8 byte value which saves the IRQ flags which are to be passed back to the IRQ restore kfunc.

Renumber the enums for REF_TYPE_* to simplify the check in find_lock_state, filtering out non-lock types as they grow will become cumbersome and is unecessary.

To track a dynamic number of IRQ-disabled regions and their associated saved states, a new resource type RES_TYPE_IRQ is introduced, which its state management functions: acquire_irq_state and release_irq_state, taking advantage of the refactoring and clean ups made in earlier commits.

One notable requirement of the kernel's IRQ save and restore API is that they cannot happen out of order. For this purpose, when releasing reference we keep track of the prev_id we saw with REF_TYPE_IRQ. Since reference states are inserted in increasing order of the index, this is used to remember the ordering of acquisitions of IRQ saved states, so that we maintain a logical stack in acquisition order of resource identities, and can enforce LIFO ordering when restoring IRQ state. The top of the stack is maintained using bpf_verifier_state's active_irq_id.

To maintain the stack property when releasing reference states, we need to modify release_reference_state to instead shift the remaining array left using memmove instead of swapping deleted element with last that might break the ordering. A selftest to test this subtle behavior is added in late patches.

The logic to detect initialized and unitialized irq flag slots, marking and unmarking is similar to how it's done for iterators. No additional checks are needed in refsafe for REF_TYPE_IRQ, apart from the usual check_id satisfiability check on the ref[i].id. We have to perform the same check_ids check on state->active_irq_id as well.

The kfuncs themselves are plain wrappers over local_irq_save and local_irq_restore macros.

---
 include/linux/bpf_verifier.h |   8 +-
 kernel/bpf/helpers.c         |  17 ++
 kernel/bpf/log.c             |   1 +
 kernel/bpf/verifier.c        | 298 ++++++++++++++++++++++++++++++++++-
 4 files changed, 320 insertions(+), 4 deletions(-)
include/linux/bpf_verifier.h +6 -2
index 03e351c43fa8..c8ea5efd147b 100644
@@ -233,6 +233,7 @@ enum bpf_stack_slot_type {
 	 */
 	STACK_DYNPTR,
 	STACK_ITER,
+	STACK_IRQ_FLAG,
 };
 
 #define BPF_REG_SIZE 8	/* size of eBPF register in bytes */
@@ -254,8 +255,10 @@ struct bpf_reference_state {
 	 * default to pointer reference on zero initialization of a state.
 	 */
 	enum ref_state_type {
-		REF_TYPE_PTR = 0,
-		REF_TYPE_LOCK,
+		REF_TYPE_PTR	= 1,
+		REF_TYPE_IRQ	= 2,
+
+		REF_TYPE_LOCK	= 3,
 	} type;
 	/* Track each reference created with a unique id, even if the same
 	 * instruction creates the reference multiple times (eg, via CALL).
@@ -421,6 +424,7 @@ struct bpf_verifier_state {
 	u32 acquired_refs;
 	u32 active_locks;
 	u32 active_preempt_locks;
+	u32 active_irq_id;
 	bool active_rcu_lock;
 
 	bool speculative;
kernel/bpf/helpers.c +17 -0
index 751c150f9e1c..532ea74d4850 100644
@@ -3057,6 +3057,21 @@ __bpf_kfunc int bpf_copy_from_user_str(void *dst, u32 dst__sz, const void __user
 	return ret + 1;
 }
 
+/* Keep unsinged long in prototype so that kfunc is usable when emitted to
+ * vmlinux.h in BPF programs directly, but note that while in BPF prog, the
+ * unsigned long always points to 8-byte region on stack, the kernel may only
+ * read and write the 4-bytes on 32-bit.
+ */
+__bpf_kfunc void bpf_local_irq_save(unsigned long *flags__irq_flag)
+{
+	local_irq_save(*flags__irq_flag);
+}
+
+__bpf_kfunc void bpf_local_irq_restore(unsigned long *flags__irq_flag)
+{
+	local_irq_restore(*flags__irq_flag);
+}
+
 __bpf_kfunc_end_defs();
 
 BTF_KFUNCS_START(generic_btf_ids)
@@ -3149,6 +3164,8 @@ BTF_ID_FLAGS(func, bpf_get_kmem_cache)
 BTF_ID_FLAGS(func, bpf_iter_kmem_cache_new, KF_ITER_NEW | KF_SLEEPABLE)
 BTF_ID_FLAGS(func, bpf_iter_kmem_cache_next, KF_ITER_NEXT | KF_RET_NULL | KF_SLEEPABLE)
 BTF_ID_FLAGS(func, bpf_iter_kmem_cache_destroy, KF_ITER_DESTROY | KF_SLEEPABLE)
+BTF_ID_FLAGS(func, bpf_local_irq_save)
+BTF_ID_FLAGS(func, bpf_local_irq_restore)
 BTF_KFUNCS_END(common_btf_ids)
 
 static const struct btf_kfunc_id_set common_kfunc_set = {
kernel/bpf/log.c +1 -0
index 2d28ce926053..38050f4ee400 100644
@@ -537,6 +537,7 @@ static char slot_type_char[] = {
 	[STACK_ZERO]	= '0',
 	[STACK_DYNPTR]	= 'd',
 	[STACK_ITER]	= 'i',
+	[STACK_IRQ_FLAG] = 'f'
 };
 
 static void print_liveness(struct bpf_verifier_env *env,
kernel/bpf/verifier.c +296 -2

446 lines not on this page · raw message

Signature · 1 line
2.43.5
#8 lore raw

Re: [PATCH bpf-next v5 4/7] bpf: Introduce support for bpf_local_irq_{save,restore}

On Thu, Nov 28, 2024 at 4:16 PM Kumar Kartikeya Dwivedi
<[email protected]> > enum ref_state_type {

include/linux/bpf_verifier.h struct bpf_reference_state { · line 262

Quoted 7 lines
include/linux/bpf_verifier.h +4 -2
-               REF_TYPE_PTR = 0,
-               REF_TYPE_LOCK,
+               REF_TYPE_PTR    = 1,
+               REF_TYPE_IRQ    = 2,
+
+               REF_TYPE_LOCK   = 3,
        } type;

why extra empty line?

why renumber ?

author maintainer
#9 lore raw

Re: [PATCH bpf-next v5 4/7] bpf: Introduce support for bpf_local_irq_{save,restore}

Quoted 11 lines · Alexei Starovoitov

On Thu, Nov 28, 2024 at 4:16 PM Kumar Kartikeya Dwivedi
<[email protected]> > enum ref_state_type {

include/linux/bpf_verifier.h struct bpf_reference_state { · line 262

Quoted 7 lines
include/linux/bpf_verifier.h +4 -2
-               REF_TYPE_PTR = 0,
-               REF_TYPE_LOCK,
+               REF_TYPE_PTR    = 1,
+               REF_TYPE_IRQ    = 2,
+
+               REF_TYPE_LOCK   = 3,
        } type;
Quoted 2 lines

why extra empty line?

To separate the lock types from others, but I can drop it.

Quoted 1 line

why renumber ?

To ensure we don't get assigned REF_TYPE_PTR by default after acquire_reference_state, if someone forgets to assign the type it will be REF_TYPE_PTR.
Right now with 1 it will get caught by refsafe's default WARN. I caused this myself so decided it's better to be more explicit.

author maintainer
#10 lore raw

PATCH v5 5/7 bpf: Improve verifier log for resource leak on exit

In reply to Kumar Kartikeya Dwivedi

The verifier log when leaking resources on BPF_EXIT may be a bit confusing, as it's a problem only when finally existing from the main prog, not from any of the subprogs. Hence, update the verifier error string and the corresponding selftests matching on it.

---
 kernel/bpf/verifier.c                              |  2 +-
 .../testing/selftests/bpf/progs/exceptions_fail.c  |  4 ++--
 tools/testing/selftests/bpf/progs/preempt_lock.c   | 14 +++++++-------
 .../selftests/bpf/progs/verifier_spin_lock.c       |  2 +-
 4 files changed, 11 insertions(+), 11 deletions(-)
kernel/bpf/verifier.c +1 -1
index 9c0315fffa07..a901af186400 100644
@@ -19102,7 +19102,7 @@ static int do_check(struct bpf_verifier_env *env)
 				 * match caller reference state when it exits.
 				 */
 				err = check_resource_leak(env, exception_exit, !env->cur_state->curframe,
-							  "BPF_EXIT instruction");
+							  "BPF_EXIT instruction in main prog");
 				if (err)
 					return err;
 
tools/testing/selftests/bpf/progs/exceptions_fail.c +2 -2
index fe0f3fa5aab6..8a0fdff89927 100644
@@ -131,7 +131,7 @@ int reject_subprog_with_lock(void *ctx)
 }
 
 SEC("?tc")
-__failure __msg("BPF_EXIT instruction cannot be used inside bpf_rcu_read_lock-ed region")
+__failure __msg("BPF_EXIT instruction in main prog cannot be used inside bpf_rcu_read_lock-ed region")
 int reject_with_rcu_read_lock(void *ctx)
 {
 	bpf_rcu_read_lock();
@@ -147,7 +147,7 @@ __noinline static int throwing_subprog(struct __sk_buff *ctx)
 }
 
 SEC("?tc")
-__failure __msg("BPF_EXIT instruction cannot be used inside bpf_rcu_read_lock-ed region")
+__failure __msg("BPF_EXIT instruction in main prog cannot be used inside bpf_rcu_read_lock-ed region")
 int reject_subprog_with_rcu_read_lock(void *ctx)
 {
 	bpf_rcu_read_lock();
tools/testing/selftests/bpf/progs/preempt_lock.c +7 -7
index 885377e83607..5269571cf7b5 100644
@@ -6,7 +6,7 @@
 #include "bpf_experimental.h"
 
 SEC("?tc")
-__failure __msg("BPF_EXIT instruction cannot be used inside bpf_preempt_disable-ed region")
+__failure __msg("BPF_EXIT instruction in main prog cannot be used inside bpf_preempt_disable-ed region")
 int preempt_lock_missing_1(struct __sk_buff *ctx)
 {
 	bpf_preempt_disable();
@@ -14,7 +14,7 @@ int preempt_lock_missing_1(struct __sk_buff *ctx)
 }
 
 SEC("?tc")
-__failure __msg("BPF_EXIT instruction cannot be used inside bpf_preempt_disable-ed region")
+__failure __msg("BPF_EXIT instruction in main prog cannot be used inside bpf_preempt_disable-ed region")
 int preempt_lock_missing_2(struct __sk_buff *ctx)
 {
 	bpf_preempt_disable();
@@ -23,7 +23,7 @@ int preempt_lock_missing_2(struct __sk_buff *ctx)
 }
 
 SEC("?tc")
-__failure __msg("BPF_EXIT instruction cannot be used inside bpf_preempt_disable-ed region")
+__failure __msg("BPF_EXIT instruction in main prog cannot be used inside bpf_preempt_disable-ed region")
 int preempt_lock_missing_3(struct __sk_buff *ctx)
 {
 	bpf_preempt_disable();
@@ -33,7 +33,7 @@ int preempt_lock_missing_3(struct __sk_buff *ctx)
 }
 
 SEC("?tc")
-__failure __msg("BPF_EXIT instruction cannot be used inside bpf_preempt_disable-ed region")
+__failure __msg("BPF_EXIT instruction in main prog cannot be used inside bpf_preempt_disable-ed region")
 int preempt_lock_missing_3_minus_2(struct __sk_buff *ctx)
 {
 	bpf_preempt_disable();
@@ -55,7 +55,7 @@ static __noinline void preempt_enable(void)
 }
 
 SEC("?tc")
-__failure __msg("BPF_EXIT instruction cannot be used inside bpf_preempt_disable-ed region")
+__failure __msg("BPF_EXIT instruction in main prog cannot be used inside bpf_preempt_disable-ed region")
 int preempt_lock_missing_1_subprog(struct __sk_buff *ctx)
 {
 	preempt_disable();
@@ -63,7 +63,7 @@ int preempt_lock_missing_1_subprog(struct __sk_buff *ctx)
 }
 
 SEC("?tc")
-__failure __msg("BPF_EXIT instruction cannot be used inside bpf_preempt_disable-ed region")
+__failure __msg("BPF_EXIT instruction in main prog cannot be used inside bpf_preempt_disable-ed region")
 int preempt_lock_missing_2_subprog(struct __sk_buff *ctx)
 {
 	preempt_disable();
@@ -72,7 +72,7 @@ int preempt_lock_missing_2_subprog(struct __sk_buff *ctx)
 }
 
 SEC("?tc")
-__failure __msg("BPF_EXIT instruction cannot be used inside bpf_preempt_disable-ed region")
+__failure __msg("BPF_EXIT instruction in main prog cannot be used inside bpf_preempt_disable-ed region")
 int preempt_lock_missing_2_minus_1_subprog(struct __sk_buff *ctx)
 {
 	preempt_disable();
tools/testing/selftests/bpf/progs/verifier_spin_lock.c +1 -1
index 3f679de73229..25599eac9a70 100644
@@ -187,7 +187,7 @@ l0_%=:	r6 = r0;					\
 
 SEC("cgroup/skb")
 __description("spin_lock: test6 missing unlock")
-__failure __msg("BPF_EXIT instruction cannot be used inside bpf_spin_lock-ed region")
+__failure __msg("BPF_EXIT instruction in main prog cannot be used inside bpf_spin_lock-ed region")
 __failure_unpriv __msg_unpriv("")
 __naked void spin_lock_test6_missing_unlock(void)
 {
Signature · 1 line
2.43.5
author maintainer
#11 lore raw

PATCH v5 6/7 selftests/bpf: Expand coverage of preempt tests to sleepable kfunc

In reply to Kumar Kartikeya Dwivedi

For preemption-related kfuncs, we don't test their interaction with sleepable kfuncs (we do test helpers) even though the verifier has code to protect against such a pattern. Expand coverage of the selftest to include this case.

---
 tools/testing/selftests/bpf/progs/preempt_lock.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)
tools/testing/selftests/bpf/progs/preempt_lock.c +14 -0
index 5269571cf7b5..6c5797bf0ead 100644
@@ -5,6 +5,8 @@
 #include "bpf_misc.h"
 #include "bpf_experimental.h"
 
+extern int bpf_copy_from_user_str(void *dst, u32 dst__sz, const void *unsafe_ptr__ign, u64 flags) __weak __ksym;
+
 SEC("?tc")
 __failure __msg("BPF_EXIT instruction in main prog cannot be used inside bpf_preempt_disable-ed region")
 int preempt_lock_missing_1(struct __sk_buff *ctx)
@@ -113,6 +115,18 @@ int preempt_sleepable_helper(void *ctx)
 	return 0;
 }
 
+SEC("?fentry.s/" SYS_PREFIX "sys_getpgid")
+__failure __msg("kernel func bpf_copy_from_user_str is sleepable within non-preemptible region")
+int preempt_sleepable_kfunc(void *ctx)
+{
+	u32 data;
+
+	bpf_preempt_disable();
+	bpf_copy_from_user_str(&data, sizeof(data), NULL, 0);
+	bpf_preempt_enable();
+	return 0;
+}
+
 int __noinline preempt_global_subprog(void)
 {
 	preempt_balance_subprog();
Signature · 1 line
2.43.5
author maintainer
#12 lore raw

PATCH v5 7/7 selftests/bpf: Add IRQ save/restore tests

In reply to Kumar Kartikeya Dwivedi

Include tests that check for rejection in erroneous cases, like unbalanced IRQ-disabled counts, within and across subprogs, invalid IRQ flag state or input to kfuncs, behavior upon overwriting IRQ saved state on stack, interaction with sleepable kfuncs/helpers, global functions, and out of order restore. Include some success scenarios as well to demonstrate usage.

#128/1   irq/irq_save_bad_arg:OK
#128/2   irq/irq_restore_bad_arg:OK
#128/3   irq/irq_restore_missing_2:OK
#128/4   irq/irq_restore_missing_3:OK
#128/5   irq/irq_restore_missing_3_minus_2:OK
#128/6   irq/irq_restore_missing_1_subprog:OK
#128/7   irq/irq_restore_missing_2_subprog:OK
#128/8   irq/irq_restore_missing_3_subprog:OK
#128/9   irq/irq_restore_missing_3_minus_2_subprog:OK
#128/10  irq/irq_balance:OK
#128/11  irq/irq_balance_n:OK
#128/12  irq/irq_balance_subprog:OK
#128/13  irq/irq_global_subprog:OK
#128/14  irq/irq_restore_ooo:OK
#128/15  irq/irq_restore_ooo_3:OK
#128/16  irq/irq_restore_3_subprog:OK
#128/17  irq/irq_restore_4_subprog:OK
#128/18  irq/irq_restore_ooo_3_subprog:OK
#128/19  irq/irq_restore_invalid:OK
#128/20  irq/irq_save_invalid:OK
#128/21  irq/irq_restore_iter:OK
#128/22  irq/irq_save_iter:OK
#128/23  irq/irq_flag_overwrite:OK
#128/24  irq/irq_flag_overwrite_partial:OK
#128/25  irq/irq_ooo_refs_array:OK
#128/26  irq/irq_sleepable_helper:OK
#128/27  irq/irq_sleepable_kfunc:OK
#128     irq:OK

Summary: 1/27 PASSED, 0 SKIPPED, 0 FAILED

---
 .../selftests/bpf/prog_tests/verifier.c       |   2 +
 tools/testing/selftests/bpf/progs/irq.c       | 444 ++++++++++++++++++
 2 files changed, 446 insertions(+)
 create mode 100644 tools/testing/selftests/bpf/progs/irq.c
tools/testing/selftests/bpf/prog_tests/verifier.c +2 -0
index d9f65adb456b..b1b4d69c407a 100644
@@ -98,6 +98,7 @@
 #include "verifier_xdp_direct_packet_access.skel.h"
 #include "verifier_bits_iter.skel.h"
 #include "verifier_lsm.skel.h"
+#include "irq.skel.h"
 
 #define MAX_ENTRIES 11
 
@@ -225,6 +226,7 @@ void test_verifier_xdp(void)                  { RUN(verifier_xdp); }
 void test_verifier_xdp_direct_packet_access(void) { RUN(verifier_xdp_direct_packet_access); }
 void test_verifier_bits_iter(void) { RUN(verifier_bits_iter); }
 void test_verifier_lsm(void)                  { RUN(verifier_lsm); }
+void test_irq(void)			      { RUN(irq); }
 
 void test_verifier_mtu(void)
 {
tools/testing/selftests/bpf/progs/irq.c new file +444 -0

444 lines not on this page · raw message

Signature · 1 line
2.43.5