On 4/30/2021 11:32 AM, Yu, Yu-cheng wrote:

On 4/30/2021 10:47 AM, Andy Lutomirski wrote:

quoted

On Fri, Apr 30, 2021 at 10:00 AM Yu, Yu-cheng [off-list ref] 
wrote:

quoted

On 4/28/2021 4:03 PM, Andy Lutomirski wrote:

quoted

On Tue, Apr 27, 2021 at 1:44 PM Yu-cheng Yu [off-list ref] 
wrote:

quoted

When shadow stack is enabled, a task's shadow stack states must be 
saved
along with the signal context and later restored in sigreturn.  
However,
currently there is no systematic facility for extending a signal 
context.
There is some space left in the ucontext, but changing ucontext is 
likely
to create compatibility issues and there is not enough space for 
further
extensions.

Introduce a signal context extension struct 'sc_ext', which is used 
to save
shadow stack restore token address.  The extension is located above 
the fpu
states, plus alignment.  The struct can be extended (such as the ibt's
wait_endbr status to be introduced later), and sc_ext.total_size field
keeps track of total size.

I still don't like this.

Here's how the signal layout works, for better or for worse:

The kernel has:

struct rt_sigframe {
      char __user *pretcode;
      struct ucontext uc;
      struct siginfo info;
      /* fp state follows here */
};

This is roughly the actual signal frame.  But userspace does not have
this struct declared, and user code does not know the sizes of the
fields.  So it's accessed in a nonsensical way.  The signal handler
function is passed a pointer to the whole sigframe implicitly in RSP,
a pointer to &frame->info in RSI, anda pointer to &frame->uc in RDX.
User code can *find* the fp state by following a pointer from
mcontext, which is, in turn, found via uc:

struct ucontext {
      unsigned long      uc_flags;
      struct ucontext  *uc_link;
      stack_t          uc_stack;
      struct sigcontext uc_mcontext;  <-- fp pointer is in here
      sigset_t      uc_sigmask;    /* mask last for extensibility */
};

The kernel, in sigreturn, works a bit differently.  The sigreturn
variants know the base address of the frame but don't have the benefit
of receiving pointers to the fields.  So instead the kernel takes
advantage of the fact that it knows the offset to uc and parses uc
accordingly.  And the kernel follows the pointer in mcontext to find
the fp state.  The latter bit is quite important later.  The kernel
does not parse info at all.

The fp state is its own mess.  When XSAVE happened, Intel kindly (?)
gave us a software defined area between the "legacy" x87 region and
the modern supposedly extensible part.  Linux sticks the following
structure in that hole:

struct _fpx_sw_bytes {
      /*
       * If set to FP_XSTATE_MAGIC1 then this is an xstate context.
       * 0 if a legacy frame.
       */
      __u32                magic1;

      /*
       * Total size of the fpstate area:
       *
       *  - if magic1 == 0 then it's sizeof(struct _fpstate)
       *  - if magic1 == FP_XSTATE_MAGIC1 then it's sizeof(struct 
_xstate)
       *    plus extensions (if any)
       */
      __u32                extended_size;

      /*
       * Feature bit mask (including FP/SSE/extended state) that is 
present
       * in the memory layout:
       */
      __u64                xfeatures;

      /*
       * Actual XSAVE state size, based on the xfeatures saved in 
the layout.
       * 'extended_size' is greater than 'xstate_size':
       */
      __u32                xstate_size;

      /* For future use: */
      __u32                padding[7];
};


That's where we are right now upstream.  The kernel has a parser for
the FPU state that is bugs piled upon bugs and is going to have to be
rewritten sometime soon.  On top of all this, we have two upcoming
features, both of which require different kinds of extensions:

1. AVX-512.  (Yeah, you thought this story was over a few years ago,
but no.  And AMX makes it worse.)  To make a long story short, we
promised user code many years ago that a signal frame fit in 2048
bytes with some room to spare.  With AVX-512 this is false.  With AMX
it's so wrong it's not even funny.  The only way out of the mess
anyone has come up with involves making the length of the FPU state
vary depending on which features are INIT, i.e. making it more compact
than "compact" mode is.  This has a side effect: it's no longer
possible to modify the state in place, because enabling a feature with
no space allocated will make the structure bigger, and the stack won't
have room.  Fortunately, one can relocate the entire FPU state, update
the pointer in mcontext, and the kernel will happily follow the
pointer.  So new code on a new kernel using a super-compact state
could expand the state by allocating new memory (on the heap? very
awkwardly on the stack?) and changing the pointer.  For all we know,
some code already fiddles with the pointer.  This is great, except
that your patch sticks more data at the end of the FPU block that no
one is expecting, and your sigreturn code follows that pointer, and
will read off into lala land.

Then, what about we don't do that at all.  Is it possible from now on we
don't stick more data at the end, and take the relocating-fpu approach?

quoted

2. CET.  CET wants us to find a few more bytes somewhere, and those
bytes logically belong in ucontext, and here we are.

Fortunately, we can spare CET the need of ucontext extension.  When the
kernel handles sigreturn, the user-mode shadow stack pointer is right at
the restore token.  There is no need to put that in ucontext.

That seems entirely reasonable.  This might also avoid needing to
teach CRIU about CET at all.

quoted

However, the WAIT_ENDBR status needs to be saved/restored for signals.
Since IBT is now dependent on shadow stack, we can use a spare bit of
the shadow stack restore token for that.

That seems like unnecessary ABI coupling.  We have plenty of bits in
uc_flags, and we have an entire reserved word in sigcontext.  How
about just sticking this bit in one of those places?

Yes, I will make it UC_WAIT_ENDBR.

Personally, I think an explicit flag is cleaner than using a reserved 
word somewhere.  However, there is a small issue: ia32 has no uc_flags.

This series can support legacy apps up to now.  But, instead of creating 
too many special cases, perhaps we should drop CET support of ia32?

Thoughts?

Thanks,
Yu-cheng