[PATCH v4 tip 3/7] tracing: allow BPF programs to call ktime_get_ns()

[PATCH v4 tip 3/7] tracing: allow BPF programs to call ktime_get_ns()

[Alexei Starovoitov]

bpf_ktime_get_ns() is used by programs to compue time delta between events
or as a timestamp

Signed-off-by: Alexei Starovoitov <ast-uqk4Ao+rVK5Wk0Htik3J/w@public.gmane.org>
---
 include/uapi/linux/bpf.h |    1 +
 kernel/trace/bpf_trace.c |   10 ++++++++++
 2 files changed, 11 insertions(+)

diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index 4486d36d2e9e..101e509d1001 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -165,6 +165,7 @@ enum bpf_func_id {
 	BPF_FUNC_map_update_elem, /* int map_update_elem(&map, &key, &value, flags) */
 	BPF_FUNC_map_delete_elem, /* int map_delete_elem(&map, &key) */
 	BPF_FUNC_probe_read,      /* int bpf_probe_read(void *dst, int size, void *src) */
+	BPF_FUNC_ktime_get_ns,    /* u64 bpf_ktime_get_ns(void) */
 	__BPF_FUNC_MAX_ID,
 };
 
diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
index bcce9b238dad..64b918488607 100644
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -36,6 +36,11 @@ static u64 bpf_probe_read(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5)
 	return probe_kernel_read(dst, unsafe_ptr, size);
 }
 
+static u64 bpf_ktime_get_ns(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5)
+{
+	return ktime_get_ns();
+}
+
 static struct bpf_func_proto kprobe_prog_funcs[] = {
 	[BPF_FUNC_probe_read] = {
 		.func = bpf_probe_read,
@@ -45,6 +50,11 @@ static struct bpf_func_proto kprobe_prog_funcs[] = {
 		.arg2_type = ARG_CONST_STACK_SIZE,
 		.arg3_type = ARG_ANYTHING,
 	},
+	[BPF_FUNC_ktime_get_ns] = {
+		.func = bpf_ktime_get_ns,
+		.gpl_only = true,
+		.ret_type = RET_INTEGER,
+	},
 };
 
 static const struct bpf_func_proto *kprobe_prog_func_proto(enum bpf_func_id func_id)
-- 
1.7.9.5

Re: [PATCH v4 tip 3/7] tracing: allow BPF programs to call ktime_get_ns()

[Peter Zijlstra]

On Fri, Feb 27, 2015 at 04:08:51PM -0800, Alexei Starovoitov wrote:
> bpf_ktime_get_ns() is used by programs to compue time delta between events
> or as a timestamp
> 
> Signed-off-by: Alexei Starovoitov <ast-uqk4Ao+rVK5Wk0Htik3J/w@public.gmane.org>
> ---
>  include/uapi/linux/bpf.h |    1 +
>  kernel/trace/bpf_trace.c |   10 ++++++++++
>  2 files changed, 11 insertions(+)
> 
> diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> index 4486d36d2e9e..101e509d1001 100644
> --- a/include/uapi/linux/bpf.h
> +++ b/include/uapi/linux/bpf.h
> @@ -165,6 +165,7 @@ enum bpf_func_id {
>  	BPF_FUNC_map_update_elem, /* int map_update_elem(&map, &key, &value, flags) */
>  	BPF_FUNC_map_delete_elem, /* int map_delete_elem(&map, &key) */
>  	BPF_FUNC_probe_read,      /* int bpf_probe_read(void *dst, int size, void *src) */
> +	BPF_FUNC_ktime_get_ns,    /* u64 bpf_ktime_get_ns(void) */
>  	__BPF_FUNC_MAX_ID,
>  };
>  
> diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
> index bcce9b238dad..64b918488607 100644
> --- a/kernel/trace/bpf_trace.c
> +++ b/kernel/trace/bpf_trace.c
> @@ -36,6 +36,11 @@ static u64 bpf_probe_read(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5)
>  	return probe_kernel_read(dst, unsafe_ptr, size);
>  }
>  
> +static u64 bpf_ktime_get_ns(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5)
> +{
> +	return ktime_get_ns();
> +}

Please use ktime_get_mono_fast_ns() instead. If you ever want to allow
running BPF stuff from general tracepoints and the like you need to
be NMI safe.

Re: [PATCH v4 tip 3/7] tracing: allow BPF programs to call ktime_get_ns()

[Alexei Starovoitov]

On Sat, Feb 28, 2015 at 2:20 AM, Peter Zijlstra <peterz@infradead.org> wrote:
>>
>> +static u64 bpf_ktime_get_ns(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5)
>> +{
>> +     return ktime_get_ns();
>> +}
>
> Please use ktime_get_mono_fast_ns() instead. If you ever want to allow
> running BPF stuff from general tracepoints and the like you need to
> be NMI safe.

good point. will do.
Currently in trace_call_bpf() I have:
if (in_nmi()) /* not supported yet */
  return 1;
to tackle nmi after all pieces are in place.
Likely nmi-safe programs will have access to reduced set
of helper functions.