This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
RS6000 fix pr16480
- From: Alan Modra <amodra at bigpond dot net dot au>
- To: gcc-patches at gcc dot gnu dot org
- Date: Wed, 25 Aug 2004 21:28:53 +0930
- Subject: RS6000 fix pr16480
This fixes an abort on a memory address that rs6000_split_multireg_move
wasn't expecting, a (mem/i:DI (symbol_ref ..)), which is allowed by
legitimate_small_data_p. In looking over other addresses allowed by
rs6000_legitimate_address, I think we also need to look inside a LO_SUM
for a clash with the destination reg.
PR target/16480
* config/rs6000/rs6000.c (rs6000_split_multireg_move): Don't abort
on "(mem (symbol_ref ..))" rtl. Look at LO_SUM base regs as well
as PLUS base regs.
Bootstrap and regression test on powerpc-linux in progress. To fix this
properly on gcc-3.4, I think we also should back-port the
offsettable_memref_p change to rs6000_split_multireg_move.
diff -urp -xCVS -x'*~' gcc-virgin/gcc/config/rs6000/rs6000.c gcc-current/gcc/config/rs6000/rs6000.c
--- gcc-virgin/gcc/config/rs6000/rs6000.c 2004-08-25 13:08:14.909251153 +0930
+++ gcc-current/gcc/config/rs6000/rs6000.c 2004-08-25 21:13:33.405845840 +0930
@@ -11352,18 +11353,14 @@ rs6000_split_multireg_move (rtx dst, rtx
src = newsrc;
}
- /* We have now address involving an base register only.
- If we use one of the registers to address memory,
- we have change that register last. */
-
- breg = (GET_CODE (XEXP (src, 0)) == PLUS
- ? XEXP (XEXP (src, 0), 0)
- : XEXP (src, 0));
-
- if (!REG_P (breg))
- abort();
-
- if (REGNO (breg) >= REGNO (dst)
+ breg = XEXP (src, 0);
+ if (GET_CODE (breg) == PLUS || GET_CODE (breg) == LO_SUM)
+ breg = XEXP (breg, 0);
+
+ /* If the base register we are using to address memory is
+ also a destination reg, then change that register last. */
+ if (REG_P (breg)
+ && REGNO (breg) >= REGNO (dst)
&& REGNO (breg) < REGNO (dst) + nregs)
j = REGNO (breg) - REGNO (dst);
}
--
Alan Modra
IBM OzLabs - Linux Technology Centre