Re: [PATCH v32 00/21] Intel SGX foundations

[Sean Christopherson]

On Tue, Jun 16, 2020 at 11:09:58PM +0300, Jarkko Sakkinen wrote:
> On Wed, Jun 10, 2020 at 01:59:03PM -0700, Sean Christopherson wrote:
> > On Mon, Jun 01, 2020 at 10:51:57AM +0300, Jarkko Sakkinen wrote:
> > > v29:
> > > * The selftest has been moved to selftests/sgx. Because SGX is an execution
> > >   environment of its own, it really isn't a great fit with more "standard"
> > >   x86 tests.
> > > 
> > >   The RSA key is now generated on fly and the whole signing process has
> > >   been made as part of the enclave loader instead of signing the enclave
> > >   during the compilation time.
> > > 
> > >   Finally, the enclave loader loads now the test enclave directly from its
> > >   ELF file, which means that ELF file does not need to be coverted as raw
> > >   binary during the build process.
> > 
> > Something in the above rework broke the selftest.  I'm getting intermittent
> > EINIT failures with SGX_INVALID_SIGNATURE.  I'm guessing it's related to
> > the dynamic RSA key generation, e.g. only ~15% of runs fail.  Verified that
> > v29 selftest fails and v28 passes.  My internal tests also pass, i.e. it's
> > all but guaranteed to be a selftest issue, not a kernel issue.
> > 
> > Jarkko, I don't have bandwidth to dig into this right now, hopefully this
> > reproduces in your environment.  Let me know if that's not the case.
> 
> I haven't experienced but I'll try to stress test it.
> 
> Just to know how complex test should reproduce your issue, can you
> reproduce the issue by running the selftest sequentially in a loop or
> do I need to do something more complex than that?

I didn't even get that complex, just running the selftest manually will
eventually fail for me, e.g. the first failure I saw was a one-off run of
the selftest.