Patch for PR rtl-optimization/22472 (Critical HPPA bug)

[Steve Ellcey]

> I wasn't thinking clearly last night (too much beer at the OLS party).
> I believe the above should be:
> 
> #define HARD_REGNO_RENAME_OK(OLD_REG, NEW_REG) \
>   ((NEW_REG) != 2 || regs_ever_live[2] || current_function_calls_eh_return)

David,

I tested your modification of my patch and it seems to work fine for me.
I had no regressions and it fixes the bug.  Is it OK to check it in?

Steve Ellcey
sje@cup.hp.com


2005-07-25  Steve Ellcey  <sje@cup.hp.com>

	PR rtl-optimization/22472
	* config/pa/pa.h (HARD_REGNO_RENAME_OK): Define.


*** gcc.orig/gcc/config/pa/pa.h	Wed Jul 20 10:37:07 2005
--- gcc/gcc/config/pa/pa.h	Mon Jul 25 08:33:52 2005
*************** typedef struct machine_function GTY(())
*** 358,363 ****
--- 358,369 ----
  #define FRAME_POINTER_REQUIRED \
    (current_function_calls_alloca)
  
+ /* Don't allow hard registers to be renamed into r2 unless r2
+    is already live or already being saved (due to eh).  */
+ 
+ #define HARD_REGNO_RENAME_OK(OLD_REG, NEW_REG) \
+   ((NEW_REG) != 2 || regs_ever_live[2] || current_function_calls_eh_return)
+ 
  /* C statement to store the difference between the frame pointer
     and the stack pointer values immediately after the function prologue.