This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: obvious patch to unroll.c


Jan Hubicka wrote:
Hi,
unroll does not copy CONST_OR_PURE flag when duplicating insns causing
BB boundary in the middle of libcall and eventually makes us to mess up
everything.  Regtested, bootstrapped with -funroll-loops and installed
as obvious.
Mon Dec  2 20:25:56 CET 2002  Jan Hubicka  <jh@suse.cz>
	* unroll.c (copy_loop_body): Copy CONST_OR_PURE_CALL_P.
Index: unroll.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/unroll.c,v
retrieving revision 1.183
diff -c -3 -p -r1.183 unroll.c
*** unroll.c	27 Oct 2002 14:47:55 -0000	1.183
--- unroll.c	2 Dec 2002 19:25:53 -0000
*************** copy_loop_body (loop, copy_start, copy_e
*** 2214,2219 ****
--- 2214,2220 ----
  	  REG_NOTES (copy) = initial_reg_note_copy (REG_NOTES (insn), map);
  	  INSN_SCOPE (copy) = INSN_SCOPE (insn);
  	  SIBLING_CALL_P (copy) = SIBLING_CALL_P (insn);
+ 	  CONST_OR_PURE_CALL_P (new) = CONST_OR_PURE_CALL_P (insn);
                                ^^^^
Looks a bit odd given that the preceding stmts use copy?

/* Because the USAGE information potentially contains objects other
than hard registers, we need to copy it. */





Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]