Transfer RCU lock state across subprog calls
Subsystems (4)
Kumar Kartikeya Dwivedi
maintainercover Transfer RCU lock state across subprog calls
David suggested during the discussion in [0] that we should handle RCU locks in a similar fashion to spin locks where the verifier understands when a lock held in a caller is released in callee, or lock taken in callee is released in a caller, or the callee is called within a lock critical section. This set extends the same semantics to RCU read locks and adds a few selftests to verify correct behavior. This issue has also come up for sched-ext programs.
This would now allow static subprog calls to be made without errors within RCU read sections, for subprogs to release RCU locks of callers and return to them, or for subprogs to take RCU lock which is later released in the caller.
[0]: https://lore.kernel.org/bpf/[email protected]
Kumar Kartikeya Dwivedi (2):
bpf: Transfer RCU lock state between subprog calls selftests/bpf: Add tests for RCU lock transfer between subprogs
kernel/bpf/verifier.c | 3 +- .../selftests/bpf/prog_tests/rcu_read_lock.c | 3 + .../selftests/bpf/progs/rcu_read_lock.c | 64 +++++++++++++++++++ 3 files changed, 68 insertions(+), 2 deletions(-)
base-commit: 2a79690eae953daaac232f93e6c5ac47ac539f2d
Signature · 1 line
2.40.1
Kumar Kartikeya Dwivedi
maintainerPATCH 1/2 bpf: Transfer RCU lock state between subprog calls
Allow transferring an imbalanced RCU lock state between subprog calls during verification. This allows patterns where a subprog call returns with an RCU lock held, or a subprog call releases an RCU lock held by the caller. Currently, the verifier would end up complaining if the RCU lock is not released when processing an exit from a subprog, which is non-ideal if its execution is supposed to be enclosed in an RCU read section of the caller.
Instead, simply only check whether we are processing exit for frame#0 and do not complain on an active RCU lock otherwise. We only need to update the check when processing BPF_EXIT insn, as copy_verifier_state is already set up to do the right thing.
--- kernel/bpf/verifier.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
kernel/bpf/verifier.c +1 -2
@@ -17698,8 +17698,7 @@ static int do_check(struct bpf_verifier_env *env)return -EINVAL ; } -if (env->cur_state->active_rcu_lock && - !in_rbtree_lock_required_cb (env)) { +if (env->cur_state->active_rcu_lock && !env->cur_state->curframe) {verbose (env,"bpf_rcu_read_unlock is missing\n" );return -EINVAL ; }
Signature · 1 line
2.40.1
Re: [PATCH bpf-next v1 1/2] bpf: Transfer RCU lock state between subprog calls
Quoted 9 lines · Kumar Kartikeya Dwivedi
Allow transferring an imbalanced RCU lock state between subprog calls during verification. This allows patterns where a subprog call returns with an RCU lock held, or a subprog call releases an RCU lock held by the caller. Currently, the verifier would end up complaining if the RCU lock is not released when processing an exit from a subprog, which is non-ideal if its execution is supposed to be enclosed in an RCU read section of the caller.
Quoted 7 lines
Instead, simply only check whether we are processing exit for frame#0 and do not complain on an active RCU lock otherwise. We only need to update the check when processing
BPF_EXITinsn, ascopy_verifier_stateis already set up to do the right thing.Suggested-by David Vernet <[email protected]>Signed-off-by Kumar Kartikeya Dwivedi <[email protected]>
Quoted 22 lines
--- kernel/bpf/verifier.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)kernel/bpf/verifier.c +1
-3@@ -17698,8 +17698,7 @@ static int do_check(struct bpf_verifier_env *env)return -EINVAL ; } -if (env->cur_state->active_rcu_lock && - !in_rbtree_lock_required_cb (env)) { +if (env->cur_state->active_rcu_lock && !env->cur_state->curframe) {verbose (env,"bpf_rcu_read_unlock is missing\n" );return -EINVAL ; } --2.40.1
Signature · 2 lines
Regards Yafang
Re: [PATCH bpf-next v1 1/2] bpf: Transfer RCU lock state between subprog calls
In reply to Kumar Kartikeya DwivediQuoted 8 lines · Kumar Kartikeya Dwivedi
Allow transferring an imbalanced RCU lock state between subprog calls during verification. This allows patterns where a subprog call returns with an RCU lock held, or a subprog call releases an RCU lock held by the caller. Currently, the verifier would end up complaining if the RCU lock is not released when processing an exit from a subprog, which is non-ideal if its execution is supposed to be enclosed in an RCU read section of the caller.
Quoted 7 lines
Instead, simply only check whether we are processing exit for frame#0 and do not complain on an active RCU lock otherwise. We only need to update the check when processing
BPF_EXITinsn, ascopy_verifier_stateis already set up to do the right thing.Suggested-by David Vernet <[email protected]>Signed-off-by Kumar Kartikeya Dwivedi <[email protected]>
Kumar Kartikeya Dwivedi
maintainerPATCH 2/2 selftests/bpf: Add tests for RCU lock transfer between subprogs
In reply to Kumar Kartikeya DwivediAdd selftests covering the following cases:
- A static subprog called from within a RCU read section works
- A static subprog taking an RCU read lock which is released in caller works
- A static subprog releasing the caller's RCU read lock works
--- .../selftests/bpf/prog_tests/rcu_read_lock.c | 3 + .../selftests/bpf/progs/rcu_read_lock.c | 64 +++++++++++++++++++ 2 files changed, 67 insertions(+)
tools/testing/selftests/bpf/prog_tests/rcu_read_lock.c +3 -0
@@ -29,6 +29,9 @@ static void test_success(void)bpf_program__set_autoload (skel->progs.non_sleepable_1, true);bpf_program__set_autoload (skel->progs.non_sleepable_2, true);bpf_program__set_autoload (skel->progs.task_trusted_non_rcuptr, true); +bpf_program__set_autoload (skel->progs.rcu_read_lock_subprog, true); +bpf_program__set_autoload (skel->progs.rcu_read_lock_subprog_lock, true); +bpf_program__set_autoload (skel->progs.rcu_read_lock_subprog_unlock, true); err =rcu_read_lock__load (skel);if (!ASSERT_OK (err,"skel_load" )) goto out;
tools/testing/selftests/bpf/progs/rcu_read_lock.c +64 -0
67 lines not on this page · raw message
Signature · 1 line
2.40.1
Re: [PATCH bpf-next v1 2/2] selftests/bpf: Add tests for RCU lock transfer between subprogs
Quoted 5 lines · Kumar Kartikeya Dwivedi
Add selftests covering the following cases:
- A static subprog called from within a RCU read section works
- A static subprog taking an RCU read lock which is released in caller works
- A static subprog releasing the caller's RCU read lock works
Given the global subprog is not allowed, we'd better add failure cases for a global subprog.
Quoted 97 lines
Signed-off-by Kumar Kartikeya Dwivedi <[email protected]>--- .../selftests/bpf/prog_tests/rcu_read_lock.c | 3 + .../selftests/bpf/progs/rcu_read_lock.c | 64 +++++++++++++++++++ 2 files changed, 67 insertions(+)tools/testing/selftests/bpf/prog_tests/rcu_read_lock.c +3
-0@@ -29,6 +29,9 @@ static void test_success(void)bpf_program__set_autoload (skel->progs.non_sleepable_1, true);bpf_program__set_autoload (skel->progs.non_sleepable_2, true);bpf_program__set_autoload (skel->progs.task_trusted_non_rcuptr, true); +bpf_program__set_autoload (skel->progs.rcu_read_lock_subprog, true); +bpf_program__set_autoload (skel->progs.rcu_read_lock_subprog_lock, true); +bpf_program__set_autoload (skel->progs.rcu_read_lock_subprog_unlock, true); err =rcu_read_lock__load (skel);if (!ASSERT_OK (err,"skel_load" )) goto out;tools/testing/selftests/bpf/progs/rcu_read_lock.c +64
-1@@ -319,3 +319,67 @@ int cross_rcu_region(void *ctx)bpf_rcu_read_unlock ();return 0 ; } + +__noinline +static int static_subprog(void *ctx) +{ +volatile int ret =0 ; + +if (bpf_get_prandom_u32 ()) +return ret +42 ; +return ret +bpf_get_prandom_u32 (); +} + +__noinline +static int static_subprog_lock(void *ctx) +{ +volatile int ret =0 ; + +bpf_rcu_read_lock (); +if (bpf_get_prandom_u32 ()) +return ret +42 ; +return ret +bpf_get_prandom_u32 (); +} + +__noinline +static int static_subprog_unlock(void *ctx) +{ +volatile int ret =0 ; + +bpf_rcu_read_unlock (); +if (bpf_get_prandom_u32 ()) +return ret +42 ; +return ret +bpf_get_prandom_u32 (); +} + +SEC("?fentry.s/" SYS_PREFIX "sys_getpgid") +int rcu_read_lock_subprog (void *ctx) +{ +volatile int ret =0 ; + +bpf_rcu_read_lock (); +if (bpf_get_prandom_u32 ()) + ret +=static_subprog (ctx); +bpf_rcu_read_unlock (); +return 0 ; +} + +SEC("?fentry.s/" SYS_PREFIX "sys_getpgid") +int rcu_read_lock_subprog_lock (void *ctx) +{ +volatile int ret =0 ; + + ret +=static_subprog_lock (ctx); +bpf_rcu_read_unlock (); +return 0 ; +} + +SEC("?fentry.s/" SYS_PREFIX "sys_getpgid") +int rcu_read_lock_subprog_unlock (void *ctx) +{ +volatile int ret =0 ; + +bpf_rcu_read_lock (); + ret +=static_subprog_unlock (ctx); +return 0 ; +} --2.40.1
Signature · 2 lines
Regards Yafang
Kumar Kartikeya Dwivedi
maintainerRe: [PATCH bpf-next v1 2/2] selftests/bpf: Add tests for RCU lock transfer between subprogs
Quoted 11 lines · Yafang Shao
Quoted 5 lines · Kumar Kartikeya Dwivedi
Add selftests covering the following cases:
- A static subprog called from within a RCU read section works
- A static subprog taking an RCU read lock which is released in caller works
- A static subprog releasing the caller's RCU read lock worksGiven the global subprog is not allowed, we'd better add failure cases for a global subprog.
I will add tests for global subprogs, but just to be clear, it's not that they are disallowed, but that the verifier won't be able to see whether their caller holds an RCU read lock or not. Therefore, it would be similar to how the main subprog is verified, in the verifier state for them active_rcu_lock will be false, and whatever that entails will follow.
They can be called within an RCU read section, but they can't leave the lock in an imbalanced state on exit.
Quoted 1 line
[...]
Re: [PATCH bpf-next v1 2/2] selftests/bpf: Add tests for RCU lock transfer between subprogs
In reply to Kumar Kartikeya DwivediQuoted 6 lines · Kumar Kartikeya Dwivedi
Add selftests covering the following cases:
- A static subprog called from within a RCU read section works
- A static subprog taking an RCU read lock which is released in caller works
- A static subprog releasing the caller's RCU read lock worksSigned-off-by Kumar Kartikeya Dwivedi <[email protected]>