Can a gate keeper help review this patch
The compiler build with --with-build-optimize=DEBUG
See the bug description at https://bugs.open64.net/show_bug.cgi?id=823
The problem is that we were calling:
GRA_LIVE_Recalc_Liveness(region ? REGION_get_rid( rwn) : NULL);
GRA_LIVE_Rename_TNs();
at +O1, when we should not. As a result, LRA encountered an
unexpected GTN and asserted.
The fix is to guard this code with:
if (!CG_localize_tns) {
CG_localize_tns is set when the optimization level is <= 1.
here is the patch
Index: osprey/be/cg/cg.cxx
===================================================================
--- osprey/be/cg/cg.cxx (revision 3663)
+++ osprey/be/cg/cg.cxx (working copy)
@@ -1487,8 +1487,10 @@
Check_for_Dump_ALL ( TP_CGEXP, NULL, "Pre LIS" );
#else
- GRA_LIVE_Recalc_Liveness(region ? REGION_get_rid( rwn) : NULL);
- GRA_LIVE_Rename_TNs();
+ if (!CG_localize_tns) {
+ GRA_LIVE_Recalc_Liveness(region ? REGION_get_rid( rwn) : NULL);
+ GRA_LIVE_Rename_TNs();
+ }
#if !defined(TARG_PPC32) // PPC IGLS_Schedule_Region bugs
IGLS_Schedule_Region (TRUE /* before register allocation */);
#ifdef TARG_X8664
yongchong
------------------------------------------------------------------------------
All of the data generated in your IT infrastructure is seriously valuable.
Why? It contains a definitive record of application performance, security
threats, fraudulent activity, and more. Splunk takes this data and makes
sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-d2d-c2
_______________________________________________
Open64-devel mailing list
Open64-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/open64-devel
