Re: [PATCH 2/3] x86/PV: replace assertions in '0' debug key stack dumping

[Roger Pau Monné <roger.pau@xxxxxxxxxx>]

On Wed, Sep 29, 2021 at 11:42:54AM +0200, Jan Beulich wrote:
> While it was me to add them, I'm afraid I don't see justification for
> the assertions: A vCPU may very well have got preempted while in user
> mode. Limit compat guest user mode stack dumps to the containing page
> (like is done when using do_page_walk()), and suppress their dumping
> altogether for 64-bit Dom0.

I'm slightly lost by this last sentence...

> Fixes: cc0de53a903c ("x86: improve output resulting from sending '0' over 
> serial")
> Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>
> ---
> An alternative to suppressing the dump for 64-bit would be to make
> do_page_fault() guest-user-mode aware.
> 
> --- a/xen/arch/x86/traps.c
> +++ b/xen/arch/x86/traps.c
> @@ -254,7 +254,6 @@ static void compat_show_guest_stack(stru
>          struct vcpu *vcpu;
>          unsigned long mfn;
>  
> -        ASSERT(guest_kernel_mode(v, regs));
>          mfn = read_cr3() >> PAGE_SHIFT;
>          for_each_vcpu( v->domain, vcpu )
>              if ( pagetable_get_pfn(vcpu->arch.guest_table) == mfn )
> @@ -269,6 +268,8 @@ static void compat_show_guest_stack(stru
>              }
>              mask = PAGE_SIZE;
>          }
> +        else if ( !guest_kernel_mode(v, regs) )
> +            mask = PAGE_SIZE;
>      }
>  
>      for ( i = 0; i < debug_stack_lines * 8; i++ )
> @@ -328,7 +329,12 @@ static void show_guest_stack(struct vcpu
>      {
>          struct vcpu *vcpu;
>  
> -        ASSERT(guest_kernel_mode(v, regs));
> +        if ( !guest_kernel_mode(v, regs) )
> +        {
> +            printk("User mode stack\n");
> +            return;
> +        }

...as you seem to unconditionally prevent the dump regardless of
whether it's dom0 or domU as long as it's not a kernel stack?

I assume when running in PV 64bit mode user-space could be executing a
32bit program and hence Xen could then misprint the stack as a 64bit
one?

Thanks, Roger.