Closed lacabra closed 5 years ago
It's also worth noting that in the machines where the build fails, it also throws the following error earlier:
error[E0463]: can't find crate for `std`
--> /root/.cargo/git/checkouts/rust-sgx-sdk-fc8771c5c45bde9a/212d9f4/xargo/sgx_tunittest/../../sgx_tunittest/src/lib.rs:88:1
|
88 | extern crate sgx_tstd as std;
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ can't find crate
error: aborting due to previous error
which is resolved by running the following as suggested in this repo's README:
rm -rf /root/.cargo/git/checkouts/rust-sgx-sdk-fc8771c5c45bde9a/212d9f4/xargo
But this error does not happen in the Linux machines where the build succeeds.
Figured it out: the JOBS=8
was the culprit.
The parameter JOBS
sets the number of parallel jobs, which defaults to the number of CPUs. The Macs in question have an i7 processor with 4 CPUs. It seems that setting JOBS
to a number higher than the number of processors in your system results in a crash or an unpredictable behavior.
Compiling with JOBS=1
, JOBS=4
or removing the JOBS
altogether, causes the build to succeed. Documenting this here for future reference, and closing.
Describe the bug Building core in some machine (MacOS) fails, while it succeeds in others (Linux/Ubuntu) in Simulation mode. This is only observed in Simulation mode because the machines in which the builds fail do not have SGX, and thus cannot build in Hardware mode.
The same build in Simulation mode used to work at some point in the recent past in these very same machines where it now fails.
Both @apalepu23 and I are experiencing the same issue in our respective machines.
To Reproduce Steps to reproduce the behavior:
enigmampc/enigma-core:latest
...Expected behavior A successful build
Backtrace If applicable, Please paste the whole error and backtrace, if it's too long use:
Desktop (please complete the following information):