Updated patch.
I've verified that I'm changing the only relevant place in
tree-ssa-loop-ivopts.c that creates a POINTER_PLUS_EXPR, so I do think
that this is the only changed to fix the problem for ivopts.
OK for mainline?
Ian
gcc/ChangeLog:
2016-01-22 Ian Lance Taylor <i...@google.com>
* common.opt (fkeep-gc-roots-live): New option.
* tree-ssa-loop-ivopts.c (add_autoinc_candidates): If
-fkeep-gc-roots-live, skip pointers.
* doc/invoke.texi (Optimize Options): Document
-fkeep-gc-roots-live.
gcc/testsuite/ChangeLog:
2016-01-22 Ian Lance Taylor <i...@google.com>
* gcc.dg/tree-ssa/ivopt_5.c: New test.
Index: common.opt
===================================================================
--- common.opt (revision 232580)
+++ common.opt (working copy)
@@ -1380,6 +1380,10 @@
Enable hoisting adjacent loads to encourage generating conditional move
instructions.
+fkeep-gc-roots-live
+Common Report Var(flag_keep_gc_roots_live) Optimization
+Always keep a pointer to a live memory block
+
floop-parallelize-all
Common Report Var(flag_loop_parallelize_all) Optimization
Mark all loops as parallel.
Index: doc/invoke.texi
===================================================================
--- doc/invoke.texi (revision 232580)
+++ doc/invoke.texi (working copy)
@@ -359,7 +359,7 @@
-fno-ira-share-spill-slots @gol
-fisolate-erroneous-paths-dereference -fisolate-erroneous-paths-attribute @gol
-fivopts -fkeep-inline-functions -fkeep-static-functions @gol
--fkeep-static-consts -flive-range-shrinkage @gol
+-fkeep-static-consts -fkeep-gc-roots-live -flive-range-shrinkage @gol
-floop-block -floop-interchange -floop-strip-mine @gol
-floop-unroll-and-jam -floop-nest-optimize @gol
-floop-parallelize-all -flra-remat -flto -flto-compression-level @gol
@@ -6621,6 +6621,17 @@
If you use @option{-Wunsafe-loop-optimizations}, the compiler warns you
if it finds this kind of loop.
+@item -fkeep-gc-roots-live
+@opindex fkeep-gc-roots-live
+This option tells the compiler that a garbage collector will be used,
+and that therefore the compiled code must retain a live pointer into
+all memory blocks. The compiler is permitted to construct a pointer
+that is outside the bounds of a memory block, but it must ensure that
+given a pointer into memory, some pointer into that memory remains
+live in the compiled code whenever it is live in the source code.
+This option is disabled by default for most languages, enabled by
+default for languages that use garbage collection.
+
@item -fcrossjumping
@opindex fcrossjumping
Perform cross-jumping transformation.
Index: testsuite/gcc.dg/tree-ssa/ivopt_5.c
===================================================================
--- testsuite/gcc.dg/tree-ssa/ivopt_5.c (revision 0)
+++ testsuite/gcc.dg/tree-ssa/ivopt_5.c (working copy)
@@ -0,0 +1,23 @@
+/* { dg-options "-O2 -fdump-tree-ivopts -fkeep-gc-roots-live" } */
+
+/* No ivopts here when using -fkeep-gc-roots-live. */
+
+void foo (char *pstart, int n)
+{
+ char *p;
+ char *pend = pstart + n;
+
+ for (p = pstart; p < pend; p++)
+ *p = 1;
+}
+
+void foo1 (char *pstart, int n)
+{
+ char *p;
+ char *pend = pstart + n;
+
+ for (p = pstart; p != pend; p++)
+ *p = 1;
+}
+
+/* { dg-final { scan-tree-dump-times "ivtmp.\[0-9_\]* = PHI <" 0 "ivopts"} } */
Index: tree-ssa-loop-ivopts.c
===================================================================
--- tree-ssa-loop-ivopts.c (revision 232580)
+++ tree-ssa-loop-ivopts.c (working copy)
@@ -2956,6 +2956,16 @@
|| !cst_and_fits_in_hwi (step))
return;
+ /* -fkeep-gc-roots-live means that we have to keep a real pointer
+ live, but the ivopts code may replace a real pointer with one
+ pointing before or after the memory block that is then adjusted
+ into the memory block during the loop. FIXME: It would likely be
+ better to actually force the pointer live and still use ivopts;
+ for example, it would be enough to write the pointer into memory
+ and keep it there until after the loop. */
+ if (flag_keep_gc_roots_live && POINTER_TYPE_P (TREE_TYPE (base)))
+ return;
+
cstepi = int_cst_value (step);
mem_mode = TYPE_MODE (TREE_TYPE (*use->op_p));
