Re: [PATCH v7 tip 3/8] tracing: allow BPF programs to call bpf_ktime_get_ns()

On 3/19/15 8:11 AM, Steven Rostedt wrote:

On Mon, 16 Mar 2015 14:49:39 -0700
Alexei Starovoitov [off-list ref] wrote:

quoted

bpf_ktime_get_ns() is used by programs to compue time delta between events

  "compute"

ok :)

quoted

+	[BPF_FUNC_ktime_get_ns] = {
+		.func = bpf_ktime_get_ns,
+		.gpl_only = true,
+		.ret_type = RET_INTEGER,

Hmm, a nanosecond value returned as integer? Is there a way to make
this a 64 bit return type, or is RET_INTEGER default to 64 bits in BPF
functions?

RET_INTEGER doesn't mean C 'int' width. It means non-pointer and
non-void value. The return value is always promoted to full
register width which is 64-bit.