Re: EmulatorPkg Unix Host Segmentation fault.

[Andrew Fish <afish@apple.com>]

> On Nov 19, 2018, at 2:12 PM, Laszlo Ersek <lersek@redhat.com> wrote:
> 
> Jordan wrote:
> 
>>>> So, is it safe to adjust rbp? Unknown. It may not be if rbp is not
>>>> used as a frame pointer. Is it safe to *not* adjust rbp and
>>>> potentially allow the old temp ram stack to be used? Unknown.
> 
> Andrew wrote:
> 
>> Looks like OvmfPkg uses SetJump()/LongJump() to change the stack. 
>> 
>>  //
>>  // Use SetJump()/LongJump() to switch to a new stack.
>>  // 
>>  if (SetJump (&JumpBuffer) == 0) {
>> #if defined (MDE_CPU_IA32)
>>    JumpBuffer.Esp = JumpBuffer.Esp + DebugAgentContext.StackMigrateOffset;
>>    JumpBuffer.Ebp = JumpBuffer.Ebp + DebugAgentContext.StackMigrateOffset;
>> #endif    
>> #if defined (MDE_CPU_X64)
>>    JumpBuffer.Rsp = JumpBuffer.Rsp + DebugAgentContext.StackMigrateOffset;
>>    JumpBuffer.Rbp = JumpBuffer.Rbp + DebugAgentContext.StackMigrateOffset;
>> #endif    
>>    LongJump (&JumpBuffer, (UINTN)-1);
>>  }
>> 
>>  SaveAndSetDebugTimerInterrupt (OldStatus);
>> 
>>  return EFI_SUCCESS;
>> }
>> 
>> But given the above code is C code RBP is going to be restored on return. This would seem to imply that the adjusting of the callers RBP is not required? [...]
> 
> The Ebp/Rbp assignments were added in a separate bugfix, namely
> 
>  https://github.com/tianocore/edk2/commit/89796c69d9fd
> 

Laszlo,

This makes sense since as the post-amble when using frame pointers is either:

	addq	$288, %rsp              ## imm = 0x120
	popq	%rbp
	retq

or 

>    0x00000000fffcd42f <+403>:	c9	leaveq
>    0x00000000fffcd430 <+404>:	c3	retq

I've noticed that clang does not seem to be a big fan of the leave function and it adjusts the stack pointer using math rather than using %rbp. 

The stack unwind algorithm implies that the frame point, %rbp in our case, is valid while the function is running. Even for the clang code gen not switching the frame point would break the stack unwind. 

Thanks,

Andrew Fish 



> Laszlo