New kfunc to return ARCH-specific timecounter. The main reason to
implement this kfunc is to avoid extra overhead of benchmark
measurements, which are usually done by a pair of bpf_ktime_get_ns()
at the beginnig and at the end of the code block under benchmark.
When fully JITed this function doesn't implement conversion to the
monotonic clock and saves some CPU cycles by receiving timecounter
values in single-digit amount of instructions. The delta values can be
translated into nanoseconds using kfunc introduced in the next patch.
For x86 BPF JIT converts this kfunc into rdtsc ordered call. Other
architectures will get JIT implementation too if supported. The fallback
is to get CLOCK_MONOTONIC_RAW value in ns.

JIT version of the function uses "LFENCE; RDTSC" variant because it
doesn't care about cookie value returned by "RDTSCP" and it doesn't want
to trash RCX value. LFENCE option provides the same ordering guarantee as
RDTSCP variant.

The simplest use-case is added in 4th patch, where we calculate the time
spent by bpf_get_ns_current_pid_tgid() kfunc. More complex example is to
use session cookie to store timecounter value at kprobe/uprobe using
kprobe.session/uprobe.session, and calculate the difference at
kretprobe/uretprobe.

Acked-by: Eduard Zingerman <[email protected]>
Acked-by: Andrii Nakryiko <[email protected]>
Acked-by: Yonghong Song <[email protected]>
Signed-off-by: Vadim Fedorenko <[email protected]>

The new helper should be used to convert deltas of values
received by bpf_get_cpu_time_counter() into nanoseconds. It is not
designed to do full conversion of time counter values to
CLOCK_MONOTONIC_RAW nanoseconds and cannot guarantee monotonicity of 2
independent values, but rather to convert the difference of 2 close
enough values of CPU timestamp counter into nanoseconds.

This function is JITted into just several instructions and adds as
low overhead as possible and perfectly suits benchmark use-cases.

When the kfunc is not JITted it returns the value provided as argument
because the kfunc in previous patch will return values in nanoseconds.

Reviewed-by: Eduard Zingerman <[email protected]>
Acked-by: Andrii Nakryiko <[email protected]>
Signed-off-by: Vadim Fedorenko <[email protected]>

bpf_get_cpu_time_counter() is replaced with rdtsc instruction on x86_64.
Add tests to check that JIT works as expected.

Acked-by: Eduard Zingerman <[email protected]>
Signed-off-by: Vadim Fedorenko <[email protected]>

The selftest provides an example of how to measure the latency of bpf
kfunc/helper call using time stamp counter and how to convert measured
value into nanoseconds.

Signed-off-by: Vadim Fedorenko <[email protected]>