Date:   Thu, 4 Nov 2021 09:46:16 -0700
From:   Yonghong Song <yhs@...com>
To:     Song Liu <songliubraving@...com>, <bpf@...r.kernel.org>,
        <netdev@...r.kernel.org>
CC:     <ast@...nel.org>, <daniel@...earbox.net>, <andrii@...nel.org>,
        <kernel-team@...com>, <kpsingh@...nel.org>
Subject: Re: [PATCH v2 bpf-next 1/2] bpf: introduce helper bpf_find_vma



On 11/4/21 12:00 AM, Song Liu wrote:
> In some profiler use cases, it is necessary to map an address to the
> backing file, e.g., a shared library. bpf_find_vma helper provides a
> flexible way to achieve this. bpf_find_vma maps an address of a task to
> the vma (vm_area_struct) for this address, and feed the vma to an callback
> BPF function. The callback function is necessary here, as we need to
> ensure mmap_sem is unlocked.
> 
> It is necessary to lock mmap_sem for find_vma. To lock and unlock mmap_sem
> safely when irqs are disable, we use the same mechanism as stackmap with
> build_id. Specifically, when irqs are disabled, the unlocked is postponed
> in an irq_work. Refactor stackmap.c so that the irq_work is shared among
> bpf_find_vma and stackmap helpers.
> 
> Signed-off-by: Song Liu <songliubraving@...com>
> ---
>   include/linux/bpf.h            |  1 +
>   include/uapi/linux/bpf.h       | 20 ++++++++++
>   kernel/bpf/mmap_unlock_work.h  | 65 +++++++++++++++++++++++++++++++
>   kernel/bpf/stackmap.c          | 71 ++++++++--------------------------
>   kernel/bpf/task_iter.c         | 49 +++++++++++++++++++++++
>   kernel/bpf/verifier.c          | 36 +++++++++++++++++
>   kernel/trace/bpf_trace.c       |  2 +
>   tools/include/uapi/linux/bpf.h | 20 ++++++++++
>   8 files changed, 209 insertions(+), 55 deletions(-)
>   create mode 100644 kernel/bpf/mmap_unlock_work.h
> 
> diff --git a/include/linux/bpf.h b/include/linux/bpf.h
> index 2be6dfd68df99..df3410bff4b06 100644
> --- a/include/linux/bpf.h
> +++ b/include/linux/bpf.h
> @@ -2157,6 +2157,7 @@ extern const struct bpf_func_proto bpf_btf_find_by_name_kind_proto;
>   extern const struct bpf_func_proto bpf_sk_setsockopt_proto;
>   extern const struct bpf_func_proto bpf_sk_getsockopt_proto;
>   extern const struct bpf_func_proto bpf_kallsyms_lookup_name_proto;
> +extern const struct bpf_func_proto bpf_find_vma_proto;
>   
>   const struct bpf_func_proto *tracing_prog_func_proto(
>     enum bpf_func_id func_id, const struct bpf_prog *prog);
> diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> index ba5af15e25f5c..22fa7b74de451 100644
> --- a/include/uapi/linux/bpf.h
> +++ b/include/uapi/linux/bpf.h
> @@ -4938,6 +4938,25 @@ union bpf_attr {
>    *		**-ENOENT** if symbol is not found.
>    *
>    *		**-EPERM** if caller does not have permission to obtain kernel address.
> + *
> + * long bpf_find_vma(struct task_struct *task, u64 addr, void *callback_fn, void *callback_ctx, u64 flags)
> + *	Description
> + *		Find vma of *task* that contains *addr*, call *callback_fn*
> + *		function with *task*, *vma*, and *callback_ctx*.
> + *		The *callback_fn* should be a static function and
> + *		the *callback_ctx* should be a pointer to the stack.
> + *		The *flags* is used to control certain aspects of the helper.
> + *		Currently, the *flags* must be 0.
> + *
> + *		The expected callback signature is
> + *
> + *		long (\*callback_fn)(struct task_struct \*task, struct vm_area_struct \*vma, void \*ctx);

ctx => callback_ctx
this should make it clear what this 'ctx' is.

> + *
> + *	Return
> + *		0 on success.
> + *		**-ENOENT** if *task->mm* is NULL, or no vma contains *addr*.
> + *		**-EBUSY** if failed to try lock mmap_lock.
> + *		**-EINVAL** for invalid **flags**.
>    */
>   #define __BPF_FUNC_MAPPER(FN)		\
>   	FN(unspec),			\
> @@ -5120,6 +5139,7 @@ union bpf_attr {
>   	FN(trace_vprintk),		\
>   	FN(skc_to_unix_sock),		\
>   	FN(kallsyms_lookup_name),	\
> +	FN(find_vma),			\
>   	/* */
>   
>   /* integer value in 'imm' field of BPF_CALL instruction selects which helper
[...]
> +
>   static void stack_map_get_build_id_offset(struct bpf_stack_build_id *id_offs,
>   					  u64 *ips, u32 trace_nr, bool user)
>   {
> -	int i;
> +	struct mmap_unlock_irq_work *work = NULL;
> +	bool irq_work_busy = bpf_mmap_unlock_get_irq_work(&work);
>   	struct vm_area_struct *vma;
> -	bool irq_work_busy = false;
> -	struct stack_map_irq_work *work = NULL;
> -
> -	if (irqs_disabled()) {
> -		if (!IS_ENABLED(CONFIG_PREEMPT_RT)) {
> -			work = this_cpu_ptr(&up_read_work);
> -			if (irq_work_is_busy(&work->irq_work)) {
> -				/* cannot queue more up_read, fallback */
> -				irq_work_busy = true;
> -			}
> -		} else {
> -			/*
> -			 * PREEMPT_RT does not allow to trylock mmap sem in
> -			 * interrupt disabled context. Force the fallback code.
> -			 */
> -			irq_work_busy = true;
> -		}
> -	}
> +	int i;

I think moving 'int i' is unnecessary here since there is no 
functionality change.

>   
> -	/*
> -	 * We cannot do up_read() when the irq is disabled, because of
> -	 * risk to deadlock with rq_lock. To do build_id lookup when the
> -	 * irqs are disabled, we need to run up_read() in irq_work. We use
> -	 * a percpu variable to do the irq_work. If the irq_work is
> -	 * already used by another lookup, we fall back to report ips.
> -	 *
> -	 * Same fallback is used for kernel stack (!user) on a stackmap
> -	 * with build_id.
> +	/* If the irq_work is in use, fall back to report ips. Same
> +	 * fallback is used for kernel stack (!user) on a stackmap with
> +	 * build_id.
>   	 */
>   	if (!user || !current || !current->mm || irq_work_busy ||
>   	    !mmap_read_trylock(current->mm)) {
[...]
> +
> +	irq_work_busy = bpf_mmap_unlock_get_irq_work(&work);
> +
> +	if (irq_work_busy || !mmap_read_trylock(mm))
> +		return -EBUSY;
> +
> +	vma = find_vma(mm, start);
> +
> +	if (vma && vma->vm_start <= start && vma->vm_end > start) {
> +		callback_fn((u64)(long)task, (u64)(long)vma,
> +			    (u64)(long)callback_ctx, 0, 0);
> +		ret = 0;
> +	}
> +	bpf_mmap_unlock_mm(work, mm);
> +	return ret;
> +}
> +
> +BTF_ID_LIST_SINGLE(btf_find_vma_ids, struct, task_struct)

We have global btf_task_struct_ids, maybe reuse it?

> +
> +const struct bpf_func_proto bpf_find_vma_proto = {
> +	.func		= bpf_find_vma,
> +	.ret_type	= RET_INTEGER,
> +	.arg1_type	= ARG_PTR_TO_BTF_ID,
> +	.arg1_btf_id	= &btf_find_vma_ids[0],
> +	.arg2_type	= ARG_ANYTHING,
> +	.arg3_type	= ARG_PTR_TO_FUNC,
> +	.arg4_type	= ARG_PTR_TO_STACK_OR_NULL,
> +	.arg5_type	= ARG_ANYTHING,
> +};
> +
>   static int __init task_iter_init(void)
>   {
>   	int ret;
> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index f0dca726ebfde..a65526112924a 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c
> @@ -6132,6 +6132,35 @@ static int set_timer_callback_state(struct bpf_verifier_env *env,
>   	return 0;
>   }
>   
> +BTF_ID_LIST_SINGLE(btf_set_find_vma_ids, struct, vm_area_struct)

In task_iter.c, we have

BTF_ID_LIST(btf_task_file_ids)
BTF_ID(struct, file)
BTF_ID(struct, vm_area_struct)

Maybe it is worthwhile to separate them so we can put vm_area_struct as 
global to be reused.

> +
> +static int set_find_vma_callback_state(struct bpf_verifier_env *env,
> +				       struct bpf_func_state *caller,
> +				       struct bpf_func_state *callee,
> +				       int insn_idx)
> +{
> +	/* bpf_find_vma(struct task_struct *task, u64 start,

start => addr ?

> +	 *               void *callback_fn, void *callback_ctx, u64 flags)
> +	 * (callback_fn)(struct task_struct *task,
> +	 *               struct vm_area_struct *vma, void *ctx);

ctx => callback_ctx ?

> +	 */
> +	callee->regs[BPF_REG_1] = caller->regs[BPF_REG_1];
> +
> +	callee->regs[BPF_REG_2].type = PTR_TO_BTF_ID;
> +	__mark_reg_known_zero(&callee->regs[BPF_REG_2]);
> +	callee->regs[BPF_REG_2].btf =  btf_vmlinux;
> +	callee->regs[BPF_REG_2].btf_id = btf_set_find_vma_ids[0];
> +
> +	/* pointer to stack or null */
> +	callee->regs[BPF_REG_3] = caller->regs[BPF_REG_4];
> +
> +	/* unused */
> +	__mark_reg_not_init(env, &callee->regs[BPF_REG_4]);
> +	__mark_reg_not_init(env, &callee->regs[BPF_REG_5]);
> +	callee->in_callback_fn = true;
> +	return 0;
> +}
> +
[...]

Powered by blists - more mailing lists