Re: [PATCH v36 23/24] docs: x86/sgx: Document SGX micro architecture and kernel internals

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Tue, Jul 28, 2020 at 11:35:11PM +0200, Pavel Machek wrote:
> Hi!
> 
> > +CPUs starting from Icelake use Total Memory Encryption (TME) in the place of
> > +MEE. TME throws away the Merkle tree, which means losing integrity and
> > +anti-replay protection but also enables variable size memory pools for EPC.
> > +Using this attack for benefit would require an interposer on the system bus.
> 
> It is not exactly clear what "this attack" means.
> 
> (And it would be cool to explain against what SGX is protecting. I
> thought it was malicious RAM, but apparently not on Icelake+).

Icelake has the same capabilities against software attacks as
generations before that given the same CPU access control.

A custom interposer on a bus could use replay for the data coming out of
the CPU package. In pre-Icelake (i.e. MEE), this is prevented with a
Merkle tree.

However, DMA is still blocked by the PRMRR to the EPC memory even on
Icelake (SDM section 41.10), which adds some cost to do so (e.g. a
malicous peripheral).

Memory is of course encrypted in both architectures.

Agreed that the current paragraph looks confusing. I'll refine it based
on what I just wrote. I want to make it as transparent as possible but
it cannot be achieved without appropriate feedback.

> 
> > +Backing storage
> > +===============
> > +
> > +Backing storage is shared and not accounted. It is implemented as a private
> > +shmem file. Providing a backing storage in some form from user space is not
> > +possible - accounting would go to invalid state as reclaimed pages would get
> > +accounted to the processes of which behalf the kernel happened to be acting on.
> 
> "of which behalf" -- I can't parse that?

Executing in a process context.

> 
> > +Access control
> > +==============
> > +
> > +`mmap()` permissions are capped by the enclave permissions. A direct
> > +consequence of this is that all the pages for an address range must be added
> > +before `mmap()` can be applied. Effectively an enclave page with minimum
> > +permission in the address range sets the permission cap for the mapping
>    ~~~~~~~~~~
>     permissions?

Thanks, I'll fix that one.

> 									Pavel

Sorry for the late response. I came last week back from vacation and
have been purging the piled up stacks of email (looking at your domain
I have to add that I was in Brno, Czech).

/Jarkko



[Index of Archives]     [Kernel Newbies]     [Security]     [Netfilter]     [Bugtraq]     [Linux FS]     [Yosemite Forum]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Samba]     [Video 4 Linux]     [Device Mapper]     [Linux Resources]

  Powered by Linux