context: use CompareAndSwap in cancelCtx.Done

context: use CompareAndSwap in cancelCtx.Done

So it is not necessary to use a mutex, speedup the code a bit:

name                                         old time/op    new time/op    delta
CommonParentCancel-8                            756ns ± 1%     760ns ± 1%    ~     (p=0.184 n=10+10)
WithTimeout/concurrency=40-8                   1.36µs ± 6%    1.38µs ± 5%    ~     (p=0.367 n=10+9)
WithTimeout/concurrency=4000-8                 1.30µs ± 1%    1.30µs ± 1%    ~     (p=0.444 n=9+9)
WithTimeout/concurrency=400000-8               1.25µs ± 1%    1.25µs ± 1%    ~     (p=0.268 n=10+10)
CancelTree/depth=1/Root=Background-8           62.3ns ± 1%    61.9ns ± 1%  -0.72%  (p=0.046 n=9+10)
CancelTree/depth=1/Root=OpenCanceler-8          447ns ± 1%     437ns ± 0%  -2.09%  (p=0.000 n=9+9)
CancelTree/depth=1/Root=ClosedCanceler-8        210ns ± 1%     211ns ± 1%  +0.35%  (p=0.024 n=10+10)
CancelTree/depth=10/Root=Background-8          2.46µs ± 0%    2.42µs ± 1%  -1.86%  (p=0.000 n=10+10)
CancelTree/depth=10/Root=OpenCanceler-8        3.50µs ± 1%    3.46µs ± 4%    ~     (p=0.063 n=9+9)
CancelTree/depth=10/Root=ClosedCanceler-8      1.21µs ± 0%    1.22µs ± 1%  +0.72%  (p=0.001 n=8+10)
CancelTree/depth=100/Root=Background-8         26.3µs ± 1%    25.7µs ± 1%  -2.20%  (p=0.000 n=9+10)
CancelTree/depth=100/Root=OpenCanceler-8       34.0µs ± 1%    33.2µs ± 1%  -2.15%  (p=0.000 n=9+10)
CancelTree/depth=100/Root=ClosedCanceler-8     11.2µs ± 1%    11.2µs ± 0%    ~     (p=0.562 n=10+9)
CancelTree/depth=1000/Root=Background-8         265µs ± 1%     260µs ± 1%  -2.15%  (p=0.000 n=10+10)
CancelTree/depth=1000/Root=OpenCanceler-8       341µs ± 1%     334µs ± 0%  -1.90%  (p=0.000 n=10+10)
CancelTree/depth=1000/Root=ClosedCanceler-8     110µs ± 0%     111µs ± 1%  +0.53%  (p=0.001 n=9+10)
CheckCanceled/Err-8                            14.2ns ± 0%    14.2ns ± 0%  -0.08%  (p=0.012 n=8+8)
CheckCanceled/Done-8                           6.19ns ± 1%    5.69ns ± 1%  -8.11%  (p=0.000 n=8+9)
ContextCancelDone-8                            1.40ns ± 0%    1.31ns ± 0%  -6.50%  (p=0.000 n=9+9)
DeepValueNewGoRoutine/depth=10-8                488ns ± 0%     490ns ± 0%  +0.62%  (p=0.000 n=9+9)
DeepValueNewGoRoutine/depth=20-8                529ns ± 0%     531ns ± 1%  +0.46%  (p=0.004 n=10+10)
DeepValueNewGoRoutine/depth=30-8                589ns ± 1%     594ns ± 0%  +0.82%  (p=0.004 n=9+9)
DeepValueNewGoRoutine/depth=50-8                664ns ± 0%     668ns ± 0%  +0.52%  (p=0.000 n=10+9)
DeepValueNewGoRoutine/depth=100-8               916ns ± 2%     915ns ± 2%    ~     (p=0.912 n=10+10)
DeepValueSameGoRoutine/depth=10-8              39.6ns ± 1%    38.8ns ± 2%  -2.01%  (p=0.001 n=9+10)
DeepValueSameGoRoutine/depth=20-8              76.9ns ± 1%    74.4ns ± 1%  -3.25%  (p=0.000 n=9+10)
DeepValueSameGoRoutine/depth=30-8               136ns ± 1%     125ns ± 1%  -8.53%  (p=0.000 n=9+10)
DeepValueSameGoRoutine/depth=50-8               196ns ± 1%     192ns ± 1%  -1.90%  (p=0.000 n=10+10)
DeepValueSameGoRoutine/depth=100-8              383ns ± 2%     372ns ± 2%  -2.86%  (p=0.000 n=10+10)

Change-Id: Ifb12affed2d6eda1104e4074d63d3f602be4c46b
Reviewed-on: https://go-review.googlesource.com/c/go/+/405674
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
Reviewed-by: Damien Neil <dneil@google.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>