Re: [PATCH v27 24/31] x86/cet/shstk: Handle thread shadow stack

On 5/22/2021 4:39 PM, Andy Lutomirski wrote:

On Fri, May 21, 2021 at 3:14 PM Yu-cheng Yu [off-list ref] wrote:

quoted

diff --git a/arch/x86/kernel/shstk.c b/arch/x86/kernel/shstk.c
index 5ea2b494e9f9..8e5f772181b9 100644
--- a/arch/x86/kernel/shstk.c
+++ b/arch/x86/kernel/shstk.c

@@ -71,6 +71,53 @@ int shstk_setup(void)
         return 0;
  }

+int shstk_alloc_thread_stack(struct task_struct *tsk, unsigned long clone_flags,
+                            unsigned long stack_size)
+{

...

quoted

+       state = get_xsave_addr(&tsk->thread.fpu.state.xsave, XFEATURE_CET_USER);
+       if (!state)
+               return -EINVAL;
+

The get_xsave_addr() API is horrible, and we already have one
egregiously buggy instance in the kernel.  Let's not add another
dubious use case.

If state == NULL, this means that CET_USER is in its init state.
CET_USER in the init state should behave identically regardless of
whether XINUSE[CET_USER] is set.  Can you please adjust this code
accordingly?

I will work on that.

Thanks,
Yu-cheng