Test change in expand_heap

[SeanTAllen]

Per #25.

Check to see if the enclave_mmap call in expand_heap.c can be safely turned into an mmap call.

[SeanTAllen]

This is being tested in lsds/sgx-lkl#717

[SeanTAllen]

@davidchisnall still a no-go.

2020-08-04T11:34:01.9403773Z [[  SGX-LKL ]] ERROR: Exception received but LKL is unable to handle it. Printing stack trace saved by exception handler:
2020-08-04T11:34:01.9404729Z [[  SGX-LKL ]]     #0: 7fe000089532 in lkl_cpu_get(...)
2020-08-04T11:34:01.9406037Z [[  SGX-LKL ]]     #1: 7fe000088e73 in lkl_syscall(...)
2020-08-04T11:34:01.9407228Z [[  SGX-LKL ]]     #2: 7fe00055d32b in __filter_syscall6(...)
2020-08-04T11:34:01.9408542Z [[  SGX-LKL ]]     #3: 7fe00055d3e2 in __mmap(...)
2020-08-04T11:34:01.9409688Z [[  SGX-LKL ]]     #4: 7fe00053c72c in __expand_heap(...)
2020-08-04T11:34:01.9410408Z [[  SGX-LKL ]]     #5: 7fe00053caa5 in expand_heap(...)
2020-08-04T11:34:01.9411264Z [[  SGX-LKL ]]     #6: 7fe00053d4e5 in malloc(...)
2020-08-04T11:34:01.9411952Z [[  SGX-LKL ]]     #7: 7fe00052dfb5 in init_aux(...)
2020-08-04T11:34:01.9412628Z [[  SGX-LKL ]]     #8: 7fe00052e224 in __init_libc(...)
2020-08-04T11:34:01.9413361Z [[  SGX-LKL ]]     #9: 7fe0005a5c6a in __libc_init_enclave(...)
2020-08-04T11:34:01.9414287Z [[  SGX-LKL ]]     #10: 7fe0005a73db in __sgx_init_enclave(...)
2020-08-04T11:34:01.9415359Z [[  SGX-LKL ]]     #11: 7fe0005a7c23 in sgxlkl_enclave_init(...)
2020-08-04T11:34:01.9416138Z [[  SGX-LKL ]]     #12: 7fe0005d10fa in ecall_sgxlkl_enclave_init(...)
2020-08-04T11:34:01.9417318Z [[  SGX-LKL ]]     #13: 7fe0004f78ec in oe_handle_call_enclave_function(...)
2020-08-04T11:34:01.9418142Z [[  SGX-LKL ]]     #14: 7fe0004f997e in _handle_ecall(...)
2020-08-04T11:34:01.9418938Z [[  SGX-LKL ]]     #15: 7fe0004f92bf in __oe_handle_main(...)
2020-08-04T11:34:01.9419716Z [[  SGX-LKL ]]     #16: 7fe00051e6f4 in oe_enter(...)

[davidchisnall]

Once @mikbras's code is in, we probably need to make this conditional so that we call mmap in the userspace version and enclave_mmap in the kernel version. That said, hopefully once @vtikoo's work lands, we are just using oe_malloc before we start LKL.

[SeanTAllen]

@davidchisnall do you mean the cryptsetup replacement when you say @mikbras' code? If yes, that is in.

[davidchisnall]

No, the separation into two separately loaded components.