Message-ID: <20221203021500.okerdcfwhgykrxxg@macbook-pro-6.dhcp.thefacebook.com>
Date:   Fri, 2 Dec 2022 18:15:00 -0800
From:   Alexei Starovoitov <alexei.starovoitov@...il.com>
To:     David Vernet <void@...ifault.com>
Cc:     bpf@...r.kernel.org, ast@...nel.org, daniel@...earbox.net,
        andrii@...nel.org, martin.lau@...ux.dev, song@...nel.org,
        yhs@...a.com, john.fastabend@...il.com, kpsingh@...nel.org,
        sdf@...gle.com, haoluo@...gle.com, jolsa@...nel.org,
        linux-kernel@...r.kernel.org, kernel-team@...a.com
Subject: Re: [PATCH bpf-next 1/2] bpf/docs: Document struct task_struct *
 kfuncs

On Fri, Dec 02, 2022 at 04:07:35PM -0600, David Vernet wrote:
> bpf_task_acquire(), bpf_task_release(), bpf_task_kptr_get(), and
> bpf_task_from_pid() are kfuncs that were recently added to
> kernel/bpf/helpers.c. These are "core" kfuncs in that they're available
> for use for any tracepoint or struct_ops BPF program. Though they have
> no ABI stability guarantees, we should still document them. This patch
> adds a new Core kfuncs section to the BPF kfuncs doc, and adds entries
> for all of these task kfuncs.
> 
> Signed-off-by: David Vernet <void@...ifault.com>
> ---
>  Documentation/bpf/kfuncs.rst | 148 +++++++++++++++++++++++++++++++++++
>  kernel/bpf/helpers.c         |   8 +-
>  2 files changed, 152 insertions(+), 4 deletions(-)
> 
> diff --git a/Documentation/bpf/kfuncs.rst b/Documentation/bpf/kfuncs.rst
> index 90774479ab7a..b0c35ad6fad4 100644
> --- a/Documentation/bpf/kfuncs.rst
> +++ b/Documentation/bpf/kfuncs.rst
> @@ -213,3 +213,151 @@ type. An example is shown below::
>                  return register_btf_kfunc_id_set(BPF_PROG_TYPE_TRACING, &bpf_task_kfunc_set);
>          }
>          late_initcall(init_subsystem);
> +
> +3. Core kfuncs
> +==============
> +
> +The BPF subsystem provides a number of "core" kfuncs that are potentially
> +applicable to a wide variety of different possible use cases and programs.
> +Those kfuncs are documented here.
> +
> +3.1 struct task_struct * kfuncs
> +-------------------------------
> +
> +There are a number of kfuncs that allow ``struct task_struct *`` objects to be
> +used as kptrs:
> +
> +.. kernel-doc:: kernel/bpf/helpers.c
> +   :identifiers: bpf_task_acquire bpf_task_release
> +
> +These kfuncs are useful when you want to acquire or release a reference to a
> +``struct task_struct *`` that was passed as e.g. a tracepoint arg, or a
> +struct_ops callback arg. For example:
> +
> +.. code-block:: c
> +
> +	/**
> +	 * A trivial example tracepoint program that shows how to
> +	 * acquire and release a struct task_struct * pointer.
> +	 */
> +	SEC("tp_btf/task_newtask")
> +	int BPF_PROG(task_acquire_release_example, struct task_struct *task, u64 clone_flags)
> +	{
> +		struct task_struct *acquired;
> +
> +		acquired = bpf_task_acquire(task);
> +
> +		/*
> +		 * In a typical program you'd do something like store
> +		 * the task in a map. Here, we just release it.

There is a sentence later in this patch about what happens with the pointer
that was stored in a map, but I would add some part of it here as well. Like:

 * In a typical program you'd do something like store
 * the task in a map and the map will automatically release it later.
 * Here, we release it manually.

> +		 */
> +		bpf_task_release(acquired);
> +		return 0;
> +	}
> +
> +If you want to acquire a reference to a ``struct task_struct`` kptr that's
> +already stored in a map, you can use bpf_task_kptr_get():
> +
> +.. kernel-doc:: kernel/bpf/helpers.c
> +   :identifiers: bpf_task_kptr_get
> +
> +Here's an example of how it can be used:
> +
> +.. code-block:: c
> +
> +	/* struct containing the struct task_struct kptr which is actually stored in the map. */
> +	struct __tasks_kfunc_map_value {
> +		struct task_struct __kptr_ref * task;
> +	};
> +
> +	/* The map containing struct __tasks_kfunc_map_value entries. */
> +	struct hash_map {
> +		__uint(type, BPF_MAP_TYPE_HASH);
> +		__type(key, int);
> +		__type(value, struct __tasks_kfunc_map_value);
> +		__uint(max_entries, 1);
> +	} __tasks_kfunc_map SEC(".maps");
> +
> +	/* ... */
> +
> +	/**
> +	 * A simple example tracepoint program showing how a
> +	 * struct task_struct kptr that is stored in a map can
> +	 * be acquired using the bpf_task_kptr_get() kfunc.
> +	 */
> +	 SEC("tp_btf/task_newtask")
> +	 int BPF_PROG(task_kptr_get_example, struct task_struct *task, u64 clone_flags)
> +	 {
> +		struct task_struct *kptr;
> +		struct __tasks_kfunc_map_value *v;
> +		s32 pid;
> +		long status;
> +
> +		status = bpf_probe_read_kernel(&pid, sizeof(pid), &task->pid);

why use the slow bpf_probe_read_kernel() here?
I think the example should follow modern coding practices.
Just: pid = task->pid; instead ?

> +		if (status)
> +			return status;
> +
> +		/* Assume a task kptr was previously stored in the map. */
> +		v = bpf_map_lookup_elem(&__tasks_kfunc_map, &pid);
> +		if (!v)
> +			return -ENOENT;
> +
> +		/* Acquire a reference to the task kptr that's already stored in the map. */
> +		kptr = bpf_task_kptr_get(&v->task);
> +		if (!kptr)
> +			/* If no task was present in the map, it's because
> +			 * we're racing with another CPU that removed it with
> +			 * bpf_kptr_xchg() between the bpf_map_lookup_elem()
> +			 * above, and our call to bpf_task_kptr_get().
> +			 * bpf_task_kptr_get() internally safely handles this
> +			 * race, and will return NULL if the task is no longer
> +			 * present in the map by the time we invoke the kfunc.
> +			 */
> +			return -EBUSY;
> +
> +		/* Free the reference we just took above. Note that the
> +		 * original struct task_struct kptr is still in the map.
> +		 * It will be freed either at a later time if another
> +		 * context deletes it from the map, or automatically by
> +		 * the BPF subsystem if it's still present when the map
> +		 * is destroyed.
> +		 */
> +		bpf_task_release(kptr);
> +
> +		return 0;
> +        }
> +
> +Finally, a BPF program can also look up a task from a pid. This can be useful
> +if the caller doesn't have a trusted pointer to a ``struct task_struct *``
> +object that it can acquire a reference on with bpf_task_acquire().
> +
> +.. kernel-doc:: kernel/bpf/helpers.c
> +   :identifiers: bpf_task_from_pid
> +
> +Here is an example of it being used:
> +
> +.. code-block:: c
> +
> +	SEC("tp_btf/task_newtask")
> +	int BPF_PROG(task_get_pid_example, struct task_struct *task, u64 clone_flags)
> +	{
> +		struct task_struct *lookup;
> +
> +		lookup = bpf_task_from_pid(task->pid);
> +		if (!lookup)
> +			/* A task should always be found, as %task is a tracepoint arg. */
> +			return -ENOENT;
> +
> +		if (lookup->pid != task->pid) {
> +			/* The pid of the lookup task should be the same as the input task. */

I suspect both "errors" are actually possible in practice,
since bpf_task_from_pid is using init_pid_ns.
But this taskd might be in different pid_ns. See task_active_pid_ns.
Probably worth mentioning this aspect of bpf_task_from_pid.

> +			bpf_task_release(lookup);
> +			return -EINVAL;
> +		}
> +
> +		/* bpf_task_from_pid() returns an acquired reference,
> +		 * so it must be dropped before returning from the
> +		 * tracepoint handler.
> +		 */
> +		bpf_task_release(lookup);
> +		return 0;
> +	}
> diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c
> index a5a511430f2a..004afbc14bbf 100644
> --- a/kernel/bpf/helpers.c
> +++ b/kernel/bpf/helpers.c
> @@ -1868,10 +1868,10 @@ struct task_struct *bpf_task_kptr_get(struct task_struct **pp)
>  }
>  
>  /**
> - * bpf_task_release - Release the reference acquired on a struct task_struct *.
> - * If this kfunc is invoked in an RCU read region, the task_struct is
> - * guaranteed to not be freed until the current grace period has ended, even if
> - * its refcount drops to 0.
> + * bpf_task_release - Release the reference acquired on a task.  If this kfunc
> + * is invoked in an RCU read region, the task_struct is guaranteed to not be
> + * freed until the current grace period has ended, even if its refcount drops
> + * to 0.
>   * @p: The task on which a reference is being released.
>   */
>  void bpf_task_release(struct task_struct *p)
> -- 
> 2.38.1
> 

Powered by blists - more mailing lists