Catch Segmentation Fault Error in an Enclave

Irene__GP — Sat, 01 Jul 2017 20:48:05 GMT

Dear all,

I am working on a project which would include some "inline assembly" code in the enclave. While in general this should work fine, one issue is that the relocation information of the assembly code is missing. In other words, there could probably exist a lot of segmentation fault issue in the enclave, something like:

...
call 0x8080460  <--- this address (0x8080460) is incorrect, its new destination in the enclave should be 0x4040404
...

To handle such issue, currently what I am thinking is to intercept all the segmentation fault (SIGSEGV) within the enclave, and re-direct the execution flow to its new destination during the runtime. This seems generally sound to me.

I knew that SGX provides an exception handling routine (sgx_register_exception_handler); however, according to this post, it seems that SGX exception handler simply cannot catch SIGSEGV exception.. I am unable to play with a SGX hardware so at this moment I assume this is true.

So here is my question:

1. Can I catch this SIGSEGV exception within the sgx_register_exception_handler?

2. If not, what is the recommended way, the best practice to capture SIGSEGV exception in an enclave for my problem?

Thank you, any advice or suggestion would be strongly appreciated!

One more question, is it

Irene__GP — Sat, 01 Jul 2017 20:51:12 GMT

One more question, is it possible to catch the SIGSEGV within the secure world (within SGX Enclave or some trusted computing base)? Or I have to catch it in the OS-level, which is not assumed as "trusted" in the SGX model...

topic Hi, in Intel® Software Guard Extensions (Intel® SGX)

Catch Segmentation Fault Error in an Enclave

One more question, is it

Hi,