Merge branch 'bpf-verifier-retval-logic-fixes'

Andrii Nakryiko says:

====================
BPF verifier retval logic fixes

This patch set fixes BPF verifier logic around validating and enforcing return
values for BPF programs that have specific range of expected return values.
Both sync and async callbacks have similar logic and are fixes as well.
A few tests are added that would fail without the fixes in this patch set.

Also, while at it, we update retval checking logic to use smin/smax range
instead of tnum, avoiding future potential issues if expected range cannot be
represented precisely by tnum (e.g., [0, 2] is not representable by tnum and
is treated as [0, 3]).

There is a little bit of refactoring to unify async callback and program exit
logic to avoid duplication of checks as much as possible.

v4->v5:
  - fix timer_bad_ret test on no-alu32 flavor (CI);
v3->v4:
  - add back bpf_func_state rearrangement patch;
  - simplified patch #4 as suggested (Shung-Hsi);
v2->v3:
  - more carefullly switch from umin/umax to smin/smax;
v1->v2:
  - drop tnum from retval checks (Eduard);
  - use smin/smax instead of umin/umax (Alexei).
====================

Link: https://lore.kernel.org/r/20231202175705.885270-1-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

11 files changed, 123 insertions, 21 deletions

diff --git a/tools/testing/selftests/bpf/progs/exceptions_assert.c b/tools/testing/selftests/bpf/progs/exceptions_assert.c
index 49efaed143fc..0ef81040da59 100644
--- a/tools/testing/selftests/bpf/progs/exceptions_assert.c
+++ b/tools/testing/selftests/bpf/progs/exceptions_assert.c
@@ -125,7 +125,7 @@ int check_assert_generic(struct __sk_buff *ctx)
 }
 
 SEC("?fentry/bpf_check")
-__failure __msg("At program exit the register R0 has value (0x40; 0x0)")
+__failure __msg("At program exit the register R1 has smin=64 smax=64")
 int check_assert_with_return(void *ctx)
 {
 	bpf_assert_with(!ctx, 64);
diff --git a/tools/testing/selftests/bpf/progs/exceptions_fail.c b/tools/testing/selftests/bpf/progs/exceptions_fail.c
index 8c0ef2742208..9cceb6521143 100644
--- a/tools/testing/selftests/bpf/progs/exceptions_fail.c
+++ b/tools/testing/selftests/bpf/progs/exceptions_fail.c
@@ -308,7 +308,7 @@ int reject_set_exception_cb_bad_ret1(void *ctx)
 }
 
 SEC("?fentry/bpf_check")
-__failure __msg("At program exit the register R0 has value (0x40; 0x0) should")
+__failure __msg("At program exit the register R1 has smin=64 smax=64 should")
 int reject_set_exception_cb_bad_ret2(void *ctx)
 {
 	bpf_throw(64);
diff --git a/tools/testing/selftests/bpf/progs/test_global_func15.c b/tools/testing/selftests/bpf/progs/test_global_func15.c
index b512d6a6c75e..b4e089d6981d 100644
--- a/tools/testing/selftests/bpf/progs/test_global_func15.c
+++ b/tools/testing/selftests/bpf/progs/test_global_func15.c
@@ -13,7 +13,7 @@ __noinline int foo(unsigned int *v)
 }
 
 SEC("cgroup_skb/ingress")
-__failure __msg("At program exit the register R0 has value")
+__failure __msg("At program exit the register R0 has ")
 int global_func15(struct __sk_buff *skb)
 {
 	unsigned int v = 1;
@@ -22,3 +22,35 @@ int global_func15(struct __sk_buff *skb)
 
 	return v;
 }
+
+SEC("cgroup_skb/ingress")
+__log_level(2) __flag(BPF_F_TEST_STATE_FREQ)
+__failure
+/* check that fallthrough code path marks r0 as precise */
+__msg("mark_precise: frame0: regs=r0 stack= before 2: (b7) r0 = 1")
+/* check that branch code path marks r0 as precise */
+__msg("mark_precise: frame0: regs=r0 stack= before 0: (85) call bpf_get_prandom_u32#7")
+__msg("At program exit the register R0 has ")
+__naked int global_func15_tricky_pruning(void)
+{
+	asm volatile (
+		"call %[bpf_get_prandom_u32];"
+		"if r0 s> 1000 goto 1f;"
+		"r0 = 1;"
+	"1:"
+		"goto +0;" /* checkpoint */
+		/* cgroup_skb/ingress program is expected to return [0, 1]
+		 * values, so branch above makes sure that in a fallthrough
+		 * case we have a valid 1 stored in R0 register, but in
+		 * a branch case we assign some random value to R0.  So if
+		 * there is something wrong with precision tracking for R0 at
+		 * program exit, we might erronenously prune branch case,
+		 * because R0 in fallthrough case is imprecise (and thus any
+		 * value is valid from POV of verifier is_state_equal() logic)
+		 */
+		"exit;"
+		:
+		: __imm(bpf_get_prandom_u32)
+		: __clobber_common
+	);
+}
diff --git a/tools/testing/selftests/bpf/progs/timer_failure.c b/tools/testing/selftests/bpf/progs/timer_failure.c
index 226d33b5a05c..9fbc69c77bbb 100644
--- a/tools/testing/selftests/bpf/progs/timer_failure.c
+++ b/tools/testing/selftests/bpf/progs/timer_failure.c
@@ -21,17 +21,37 @@ struct {
 	__type(value, struct elem);
 } timer_map SEC(".maps");
 
-static int timer_cb_ret1(void *map, int *key, struct bpf_timer *timer)
+__naked __noinline __used
+static unsigned long timer_cb_ret_bad()
 {
-	if (bpf_get_smp_processor_id() % 2)
-		return 1;
-	else
-		return 0;
+	asm volatile (
+		"call %[bpf_get_prandom_u32];"
+		"if r0 s> 1000 goto 1f;"
+		"r0 = 0;"
+	"1:"
+		"goto +0;" /* checkpoint */
+		/* async callback is expected to return 0, so branch above
+		 * skipping r0 = 0; should lead to a failure, but if exit
+		 * instruction doesn't enforce r0's precision, this callback
+		 * will be successfully verified
+		 */
+		"exit;"
+		:
+		: __imm(bpf_get_prandom_u32)
+		: __clobber_common
+	);
 }
 
 SEC("fentry/bpf_fentry_test1")
-__failure __msg("should have been in (0x0; 0x0)")
-int BPF_PROG2(test_ret_1, int, a)
+__log_level(2)
+__flag(BPF_F_TEST_STATE_FREQ)
+__failure
+/* check that fallthrough code path marks r0 as precise */
+__msg("mark_precise: frame0: regs=r0 stack= before 22: (b7) r0 = 0")
+/* check that branch code path marks r0 as precise */
+__msg("mark_precise: frame0: regs=r0 stack= before 24: (85) call bpf_get_prandom_u32#7")
+__msg("should have been in [0, 0]")
+long BPF_PROG2(test_bad_ret, int, a)
 {
 	int key = 0;
 	struct bpf_timer *timer;
@@ -39,7 +59,7 @@ int BPF_PROG2(test_ret_1, int, a)
 	timer = bpf_map_lookup_elem(&timer_map, &key);
 	if (timer) {
 		bpf_timer_init(timer, &timer_map, CLOCK_BOOTTIME);
-		bpf_timer_set_callback(timer, timer_cb_ret1);
+		bpf_timer_set_callback(timer, timer_cb_ret_bad);
 		bpf_timer_start(timer, 1000, 0);
 	}
 
diff --git a/tools/testing/selftests/bpf/progs/user_ringbuf_fail.c b/tools/testing/selftests/bpf/progs/user_ringbuf_fail.c
index 03ee946c6bf7..11ab25c42c36 100644
--- a/tools/testing/selftests/bpf/progs/user_ringbuf_fail.c
+++ b/tools/testing/selftests/bpf/progs/user_ringbuf_fail.c
@@ -184,7 +184,7 @@ invalid_drain_callback_return(struct bpf_dynptr *dynptr, void *context)
  * not be able to write to that pointer.
  */
 SEC("?raw_tp")
-__failure __msg("At callback return the register R0 has value")
+__failure __msg("At callback return the register R0 has ")
 int user_ringbuf_callback_invalid_return(void *ctx)
 {
 	bpf_user_ringbuf_drain(&user_ringbuf, invalid_drain_callback_return, NULL, 0);
diff --git a/tools/testing/selftests/bpf/progs/verifier_cgroup_inv_retcode.c b/tools/testing/selftests/bpf/progs/verifier_cgroup_inv_retcode.c
index d6c4a7f3f790..6e0f349f8f15 100644
--- a/tools/testing/selftests/bpf/progs/verifier_cgroup_inv_retcode.c
+++ b/tools/testing/selftests/bpf/progs/verifier_cgroup_inv_retcode.c
@@ -7,7 +7,7 @@
 
 SEC("cgroup/sock")
 __description("bpf_exit with invalid return code. test1")
-__failure __msg("R0 has value (0x0; 0xffffffff)")
+__failure __msg("smin=0 smax=4294967295 should have been in [0, 1]")
 __naked void with_invalid_return_code_test1(void)
 {
 	asm volatile ("					\
@@ -30,7 +30,7 @@ __naked void with_invalid_return_code_test2(void)
 
 SEC("cgroup/sock")
 __description("bpf_exit with invalid return code. test3")
-__failure __msg("R0 has value (0x0; 0x3)")
+__failure __msg("smin=0 smax=3 should have been in [0, 1]")
 __naked void with_invalid_return_code_test3(void)
 {
 	asm volatile ("					\
@@ -53,7 +53,7 @@ __naked void with_invalid_return_code_test4(void)
 
 SEC("cgroup/sock")
 __description("bpf_exit with invalid return code. test5")
-__failure __msg("R0 has value (0x2; 0x0)")
+__failure __msg("smin=2 smax=2 should have been in [0, 1]")
 __naked void with_invalid_return_code_test5(void)
 {
 	asm volatile ("					\
@@ -75,7 +75,7 @@ __naked void with_invalid_return_code_test6(void)
 
 SEC("cgroup/sock")
 __description("bpf_exit with invalid return code. test7")
-__failure __msg("R0 has unknown scalar value")
+__failure __msg("R0 has unknown scalar value should have been in [0, 1]")
 __naked void with_invalid_return_code_test7(void)
 {
 	asm volatile ("					\
diff --git a/tools/testing/selftests/bpf/progs/verifier_direct_packet_access.c b/tools/testing/selftests/bpf/progs/verifier_direct_packet_access.c
index 99a23dea8233..be95570ab382 100644
--- a/tools/testing/selftests/bpf/progs/verifier_direct_packet_access.c
+++ b/tools/testing/selftests/bpf/progs/verifier_direct_packet_access.c
@@ -411,7 +411,7 @@ l0_%=:	r0 = 0;						\
 
 SEC("tc")
 __description("direct packet access: test17 (pruning, alignment)")
-__failure __msg("misaligned packet access off 2+(0x0; 0x0)+15+-4 size 4")
+__failure __msg("misaligned packet access off 2+0+15+-4 size 4")
 __flag(BPF_F_STRICT_ALIGNMENT)
 __naked void packet_access_test17_pruning_alignment(void)
 {
diff --git a/tools/testing/selftests/bpf/progs/verifier_int_ptr.c b/tools/testing/selftests/bpf/progs/verifier_int_ptr.c
index b054f9c48143..74d9cad469d9 100644
--- a/tools/testing/selftests/bpf/progs/verifier_int_ptr.c
+++ b/tools/testing/selftests/bpf/progs/verifier_int_ptr.c
@@ -67,7 +67,7 @@ __naked void ptr_to_long_half_uninitialized(void)
 
 SEC("cgroup/sysctl")
 __description("ARG_PTR_TO_LONG misaligned")
-__failure __msg("misaligned stack access off (0x0; 0x0)+-20+0 size 8")
+__failure __msg("misaligned stack access off 0+-20+0 size 8")
 __naked void arg_ptr_to_long_misaligned(void)
 {
 	asm volatile ("					\
diff --git a/tools/testing/selftests/bpf/progs/verifier_netfilter_retcode.c b/tools/testing/selftests/bpf/progs/verifier_netfilter_retcode.c
index 353ae6da00e1..e1ffa5d32ff0 100644
--- a/tools/testing/selftests/bpf/progs/verifier_netfilter_retcode.c
+++ b/tools/testing/selftests/bpf/progs/verifier_netfilter_retcode.c
@@ -39,7 +39,7 @@ __naked void with_valid_return_code_test3(void)
 
 SEC("netfilter")
 __description("bpf_exit with invalid return code. test4")
-__failure __msg("R0 has value (0x2; 0x0)")
+__failure __msg("R0 has smin=2 smax=2 should have been in [0, 1]")
 __naked void with_invalid_return_code_test4(void)
 {
 	asm volatile ("					\
diff --git a/tools/testing/selftests/bpf/progs/verifier_stack_ptr.c b/tools/testing/selftests/bpf/progs/verifier_stack_ptr.c
index e0f77e3e7869..417c61cd4b19 100644
--- a/tools/testing/selftests/bpf/progs/verifier_stack_ptr.c
+++ b/tools/testing/selftests/bpf/progs/verifier_stack_ptr.c
@@ -37,7 +37,7 @@ __naked void ptr_to_stack_store_load(void)
 
 SEC("socket")
 __description("PTR_TO_STACK store/load - bad alignment on off")
-__failure __msg("misaligned stack access off (0x0; 0x0)+-8+2 size 8")
+__failure __msg("misaligned stack access off 0+-8+2 size 8")
 __failure_unpriv
 __naked void load_bad_alignment_on_off(void)
 {
@@ -53,7 +53,7 @@ __naked void load_bad_alignment_on_off(void)
 
 SEC("socket")
 __description("PTR_TO_STACK store/load - bad alignment on reg")
-__failure __msg("misaligned stack access off (0x0; 0x0)+-10+8 size 8")
+__failure __msg("misaligned stack access off 0+-10+8 size 8")
 __failure_unpriv
 __naked void load_bad_alignment_on_reg(void)
 {
diff --git a/tools/testing/selftests/bpf/progs/verifier_subprog_precision.c b/tools/testing/selftests/bpf/progs/verifier_subprog_precision.c
index b5efcaeaa1ae..0dfe3f8b69ac 100644
--- a/tools/testing/selftests/bpf/progs/verifier_subprog_precision.c
+++ b/tools/testing/selftests/bpf/progs/verifier_subprog_precision.c
@@ -117,6 +117,56 @@ __naked int global_subprog_result_precise(void)
 	);
 }
 
+__naked __noinline __used
+static unsigned long loop_callback_bad()
+{
+	/* bpf_loop() callback that can return values outside of [0, 1] range */
+	asm volatile (
+		"call %[bpf_get_prandom_u32];"
+		"if r0 s> 1000 goto 1f;"
+		"r0 = 0;"
+	"1:"
+		"goto +0;" /* checkpoint */
+		/* bpf_loop() expects [0, 1] values, so branch above skipping
+		 * r0 = 0; should lead to a failure, but if exit instruction
+		 * doesn't enforce r0's precision, this callback will be
+		 * successfully verified
+		 */
+		"exit;"
+		:
+		: __imm(bpf_get_prandom_u32)
+		: __clobber_common
+	);
+}
+
+SEC("?raw_tp")
+__failure __log_level(2)
+__flag(BPF_F_TEST_STATE_FREQ)
+/* check that fallthrough code path marks r0 as precise */
+__msg("mark_precise: frame1: regs=r0 stack= before 11: (b7) r0 = 0")
+/* check that we have branch code path doing its own validation */
+__msg("from 10 to 12: frame1: R0=scalar(smin=umin=1001")
+/* check that branch code path marks r0 as precise, before failing */
+__msg("mark_precise: frame1: regs=r0 stack= before 9: (85) call bpf_get_prandom_u32#7")
+__msg("At callback return the register R0 has smin=1001 should have been in [0, 1]")
+__naked int callback_precise_return_fail(void)
+{
+	asm volatile (
+		"r1 = 1;"			/* nr_loops */
+		"r2 = %[loop_callback_bad];"	/* callback_fn */
+		"r3 = 0;"			/* callback_ctx */
+		"r4 = 0;"			/* flags */
+		"call %[bpf_loop];"
+
+		"r0 = 0;"
+		"exit;"
+		:
+		: __imm_ptr(loop_callback_bad),
+		  __imm(bpf_loop)
+		: __clobber_common
+	);
+}
+
 SEC("?raw_tp")
 __success __log_level(2)
 /* First simulated path does not include callback body,