pnacl.c

Read, parse and interpret Portable Native Client Executables (pexes). See https://developers.google.com/native-client for more info about PNaCl.

Building

To build debug (out/pnacl) and optimized builds (out/pnacl-opt-assert):

$ make

To build other various configurations:

$ make everything

Tests

To run tests:

# Set your NACL_SDK_ROOT
$ export NACL_SDK_ROOT=...
$ make test

or run the Python test runner directly:

# run all tests
$ test/run-tests.py
[+38|-0] (0.72s)

# run tests that match a wildcard pattern
$ test/run-tests.py bench -v
. benchmark/binarytrees.txt (skipped)
+ benchmark/fannkuchredux.txt (0.035s)
+ benchmark/fasta.txt (0.034s)
+ benchmark/mandelbrot.txt (0.360s)
. benchmark/meteor.txt (skipped)
+ benchmark/nbody.txt (0.036s)
+ benchmark/spectralnorm.txt (0.150s)
[+5|-0] (0.62s)

# run slow tests
$ test/run-tests.py bench -v -s
+ benchmark/binarytrees.txt (1.395s)
+ benchmark/fannkuchredux.txt (0.041s)
+ benchmark/fasta.txt (0.033s)
+ benchmark/mandelbrot.txt (0.357s)
+ benchmark/meteor.txt (4.369s)
+ benchmark/nbody.txt (0.033s)
+ benchmark/spectralnorm.txt (0.142s)
[+7|-0] (6.38s)

Benchmarks

To run benchmarks:

# Set your NACL_SDK_ROOT
$ export NACL_SDK_ROOT=...
$ test/run-benchmarks.py
+ benchmark/binarytrees.10.bench O0 (4.238s) (0.078s) (54.3x)
+ benchmark/binarytrees.10.bench O2 (4.238s) (0.070s) (60.4x)
+ benchmark/binarytrees.12.bench O0 (21.166s) (0.158s) (133.6x)
+ benchmark/binarytrees.12.bench O2 (21.166s) (0.116s) (182.4x)
...
[+20|-0] (48.16s)

This runs the interpreter and compares it to a nexe translated using pnacl-translate -O0 and -O2.