This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
[PATCH] Fix (?) load_kills_store
- From: Zdenek Dvorak <rakdver at atrey dot karlin dot mff dot cuni dot cz>
- To: gcc-patches at gcc dot gnu dot org
- Cc: rth at redhat dot com
- Date: Sun, 29 Jun 2003 15:26:26 +0200
- Subject: [PATCH] Fix (?) load_kills_store
Hello,
considering description of load_kills_store, it seems to me that
arguments to true_dependence are passed in reversed order, which
could cause problems when one of the mems is unchanging (actually
switching the order seems to fix a misscompilation with my dce of
mems patch).
Zdenek
Changelog:
* gcse.c (load_kills_store): Pass arguments to true_dependence
in correct order.
Index: gcse.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/gcse.c,v
retrieving revision 1.253
diff -c -3 -p -r1.253 gcse.c
*** gcse.c 28 Jun 2003 14:56:23 -0000 1.253
--- gcse.c 29 Jun 2003 13:18:19 -0000
*************** static bool
*** 7397,7403 ****
load_kills_store (x, store_pattern)
rtx x, store_pattern;
{
! if (true_dependence (x, GET_MODE (x), store_pattern, rtx_addr_varies_p))
return true;
return false;
}
--- 7397,7404 ----
load_kills_store (x, store_pattern)
rtx x, store_pattern;
{
! if (true_dependence (store_pattern, GET_MODE (store_pattern), x,
! rtx_addr_varies_p))
return true;
return false;
}