2020-08-17 19:21:27 +08:00
|
|
|
Sandboxing PFFFT library
|
2020-08-17 19:20:56 +08:00
|
|
|
|
2020-08-17 19:21:27 +08:00
|
|
|
Builder: CMake
|
2020-08-17 19:20:56 +08:00
|
|
|
|
2020-08-17 19:21:27 +08:00
|
|
|
For testing:
|
|
|
|
|
`cd build`, then `./pffft_sandboxed`
|
2020-08-17 19:20:56 +08:00
|
|
|
|
2020-08-17 19:21:27 +08:00
|
|
|
For debug:
|
|
|
|
|
`SAPI_VLOG_LEVEL=1 ./pffft_sandboxed --v=100
|
|
|
|
|
--sandbox2_danger_danger_permit_all_and_log <auxiliar file>`
|
2020-08-17 19:20:56 +08:00
|
|
|
|
2020-08-17 19:21:27 +08:00
|
|
|
CMake observations:
|
|
|
|
|
* linking pffft and fftpack (which contains necessary functions for pffft)
|
|
|
|
|
* set math library
|
2020-08-17 19:20:56 +08:00
|
|
|
|
2020-08-17 19:21:27 +08:00
|
|
|
Sandboxed main observations:
|
|
|
|
|
* containing two testing parts (fft / pffft benchmarks)
|
|
|
|
|
! current stage: fft - works :)
|
2020-08-17 19:21:56 +08:00
|
|
|
pffft - implemented
|
|
|
|
|
* (Solved) pffft benchmark bug: "Sandbox not active"
|
|
|
|
|
N = 64, status OK, pffft_transform generates error
|
|
|
|
|
N > 64, status not OK
|
|
|
|
|
Problem on initialising sapi::StatusOr<PFFFT_Setup *> s;
|
|
|
|
|
the memory that stays for s is not the same with the address passed
|
|
|
|
|
in pffft_transform function.
|
|
|
|
|
(sapi :: v :: GenericPtr - to be changed)
|
|
|
|
|
|
|
|
|
|
Temporary solution: change the generated files to accept
|
|
|
|
|
uintptr_t instead of PFFFT_Setup
|
|
|
|
|
|
|
|
|
|
Solution: using "sapi :: v :: RemotePtr" instead of "sapi :: v :: GenericPtr"
|
|
|
|
|
to access the memory of object s
|
