Peter Pietzuch

@wintersteiger ok, I see. So the question is if the _parsed_ JSON app_config goes into a single app_config struct or into many separate structs/variables for the different SGX-LKL components.

David wrote: > I had a look in `lkl/mm/nommu.c`. If I found all of the relevant places in OpenJDK, it looks as if the logic in `do_munmap` contains the checks...

Some findings: * When removing the intercept for the `mmap` syscall in musl, it segfaults during the execution of the Stage 3 dynamic linker (here: https://github.com/lsds/sgx-lkl-musl/blob/92eaf58d34574d76745322af82c5f097c23e7b9f/ldso/dynlink.c#L1597). * Calling the `execve`...

This will be broken because we are not delivering signals to the correct thread (see https://github.com/lsds/sgx-lkl/issues/644).

@davidchisnall yes, any configuration parameter that affects enclave execution but cannot be checked by the enclave and thus can be changed behind the back of the enclave should not be...

I agree that Java and cross-compilation should go, and instead become sample applications. I still think that there is value in supporting a particular distro (Alpine) directly.

Right, I would only keep it for the moment when Alpine is the only supported distribution. Once that limitations has disappeared, I agree that it should be removed.

@hukoyu this still needs cleanup.

@shaikshavali1 1. Does only the first sub-test fail and the others pass? 2. Please use GDB to find out what the illegal assembly instruction is. Then provide a stracktrace with...

> If the JVM is running in a configuration where it depends on SIGSEGV for functional correctness then it is insecure on current SGX. We should not be running benchmarks...