All of lore.kernel.org
 help / color / mirror / Atom feed
From: Yonghong Song <yonghong.song@linux.dev>
To: Dave Marchevsky <davemarchevsky@fb.com>, bpf@vger.kernel.org
Cc: Alexei Starovoitov <ast@kernel.org>,
	Daniel Borkmann <daniel@iogearbox.net>,
	Andrii Nakryiko <andrii@kernel.org>,
	Martin KaFai Lau <martin.lau@kernel.org>,
	Kernel Team <kernel-team@fb.com>
Subject: Re: [PATCH v1 bpf-next 6/7] [RFC] bpf: Allow bpf_spin_{lock,unlock} in sleepable prog's RCU CS
Date: Tue, 1 Aug 2023 23:33:50 -0700	[thread overview]
Message-ID: <322d07e2-0620-9fd6-504e-19cb63772fe1@linux.dev> (raw)
In-Reply-To: <20230801203630.3581291-7-davemarchevsky@fb.com>



On 8/1/23 1:36 PM, Dave Marchevsky wrote:
> Commit 9e7a4d9831e8 ("bpf: Allow LSM programs to use bpf spin locks")
> disabled bpf_spin_lock usage in sleepable progs, stating:
> 
>   Sleepable LSM programs can be preempted which means that allowng spin
>   locks will need more work (disabling preemption and the verifier
>   ensuring that no sleepable helpers are called when a spin lock is
>   held).
> 
> It seems that some of this 'ensuring that no sleepable helpers are
> called' was done for RCU critical section in commit 9bb00b2895cb ("bpf:
> Add kfunc bpf_rcu_read_lock/unlock()"), specifically the check which
> fails with verbose "sleepable helper %s#%d in rcu_read_lock region"
> message in check_helper_call and similar in check_kfunc_call. These
> checks prevent sleepable helper and kfunc calls in RCU critical
> sections. Accordingly, it should be safe to allow bpf_spin_{lock,unlock}
> in RCU CS. This patch does so, replacing the broad "sleepable progs cannot use
> bpf_spin_lock yet" check with a more targeted !in_rcu_cs.
> 
> [
>    RFC: Does preemption still need to be disabled here?

Good question. My hunch is that we do not need it since
   - spin lock is inside rcu cs, which is similar to a
     spin lock in a non-sleepable program.

I could be wrong as preemption with a sleepable program
may allow explicit blocking.


> ]
> 
> Signed-off-by: Dave Marchevsky <davemarchevsky@fb.com>
> ---
>   kernel/bpf/verifier.c | 9 ++++-----
>   1 file changed, 4 insertions(+), 5 deletions(-)
> 
> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index 4bda365000d3..d1b8e8964aec 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c
> @@ -8270,6 +8270,10 @@ static int check_func_arg(struct bpf_verifier_env *env, u32 arg,
>   			verbose(env, "can't spin_{lock,unlock} in rbtree cb\n");
>   			return -EACCES;
>   		}
> +		if (!in_rcu_cs(env)) {
> +			verbose(env, "sleepable progs may only spin_{lock,unlock} in RCU CS\n");
> +			return -EACCES;
> +		}
>   		if (meta->func_id == BPF_FUNC_spin_lock) {
>   			err = process_spin_lock(env, regno, true);
>   			if (err)
> @@ -16972,11 +16976,6 @@ static int check_map_prog_compatibility(struct bpf_verifier_env *env,
>   			verbose(env, "tracing progs cannot use bpf_spin_lock yet\n");
>   			return -EINVAL;
>   		}
> -
> -		if (prog->aux->sleepable) {
> -			verbose(env, "sleepable progs cannot use bpf_spin_lock yet\n");
> -			return -EINVAL;
> -		}
>   	}
>   
>   	if (btf_record_has_field(map->record, BPF_TIMER)) {

  reply	other threads:[~2023-08-02  6:33 UTC|newest]

Thread overview: 25+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-08-01 20:36 [PATCH v1 bpf-next 0/7] BPF Refcount followups 3: bpf_mem_free_rcu refcounted nodes Dave Marchevsky
2023-08-01 20:36 ` [PATCH v1 bpf-next 1/7] bpf: Ensure kptr_struct_meta is non-NULL for collection insert and refcount_acquire Dave Marchevsky
2023-08-02  3:57   ` Yonghong Song
2023-08-02 19:23     ` Dave Marchevsky
2023-08-02 21:41       ` Yonghong Song
2023-08-04  6:17         ` David Marchevsky
2023-08-04 15:37           ` Yonghong Song
2023-08-01 20:36 ` [PATCH v1 bpf-next 2/7] bpf: Consider non-owning refs trusted Dave Marchevsky
2023-08-02  4:11   ` Yonghong Song
2023-08-01 20:36 ` [PATCH v1 bpf-next 3/7] bpf: Use bpf_mem_free_rcu when bpf_obj_dropping refcounted nodes Dave Marchevsky
2023-08-02  4:15   ` Yonghong Song
2023-08-01 20:36 ` [PATCH v1 bpf-next 4/7] bpf: Reenable bpf_refcount_acquire Dave Marchevsky
2023-08-02  5:21   ` Yonghong Song
2023-08-01 20:36 ` [PATCH v1 bpf-next 5/7] bpf: Consider non-owning refs to refcounted nodes RCU protected Dave Marchevsky
2023-08-02  5:59   ` Yonghong Song
2023-08-04  6:47     ` David Marchevsky
2023-08-04 15:43       ` Yonghong Song
2023-08-02 22:50   ` Alexei Starovoitov
2023-08-04  6:55     ` David Marchevsky
2023-08-01 20:36 ` [PATCH v1 bpf-next 6/7] [RFC] bpf: Allow bpf_spin_{lock,unlock} in sleepable prog's RCU CS Dave Marchevsky
2023-08-02  6:33   ` Yonghong Song [this message]
2023-08-02 22:55   ` Alexei Starovoitov
2023-08-01 20:36 ` [PATCH v1 bpf-next 7/7] selftests/bpf: Add tests for rbtree API interaction in sleepable progs Dave Marchevsky
2023-08-02 23:07   ` Alexei Starovoitov
2023-08-02  3:07 ` [PATCH v1 bpf-next 0/7] BPF Refcount followups 3: bpf_mem_free_rcu refcounted nodes Yonghong Song

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=322d07e2-0620-9fd6-504e-19cb63772fe1@linux.dev \
    --to=yonghong.song@linux.dev \
    --cc=andrii@kernel.org \
    --cc=ast@kernel.org \
    --cc=bpf@vger.kernel.org \
    --cc=daniel@iogearbox.net \
    --cc=davemarchevsky@fb.com \
    --cc=kernel-team@fb.com \
    --cc=martin.lau@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.