[AArch64] Fix shrinkwrapping interactions with atomics (PR92692)

The separate shrinkwrapping pass may insert stores in the middle
of atomics loops which can cause issues on some implementations.
Avoid this by delaying splitting atomics patterns until after
prolog/epilog generation.

gcc/
	PR target/92692
	* config/aarch64/aarch64.c (aarch64_split_compare_and_swap)
	Add assert to ensure prolog has been emitted.
	(aarch64_split_atomic_op): Likewise.
	* config/aarch64/atomics.md (aarch64_compare_and_swap<mode>)
	Use epilogue_completed rather than reload_completed.
	(aarch64_atomic_exchange<mode>): Likewise.
	(aarch64_atomic_<atomic_optab><mode>): Likewise.
	(atomic_nand<mode>): Likewise.
	(aarch64_atomic_fetch_<atomic_optab><mode>): Likewise.
	(atomic_fetch_nand<mode>): Likewise.
	(aarch64_atomic_<atomic_optab>_fetch<mode>): Likewise.
	(atomic_nand_fetch<mode>): Likewise.

gcc/ChangeLog

+2020-01-17  Wilco Dijkstra  <wdijkstr@arm.com>
+
+	PR target/92692
+	* config/aarch64/aarch64.c (aarch64_split_compare_and_swap)
+	Add assert to ensure prolog has been emitted.
+	(aarch64_split_atomic_op): Likewise.
+	* config/aarch64/atomics.md (aarch64_compare_and_swap<mode>)
+	Use epilogue_completed rather than reload_completed.
+	(aarch64_atomic_exchange<mode>): Likewise.
+	(aarch64_atomic_<atomic_optab><mode>): Likewise.
+	(atomic_nand<mode>): Likewise.
+	(aarch64_atomic_fetch_<atomic_optab><mode>): Likewise.
+	(atomic_fetch_nand<mode>): Likewise.
+	(aarch64_atomic_<atomic_optab>_fetch<mode>): Likewise.
+	(atomic_nand_fetch<mode>): Likewise.
+
 2020-01-17  Richard Sandiford  <richard.sandiford@arm.com>
 
 	PR target/93133

gcc/config/aarch64/aarch64.c

@@ -18375,6 +18375,9 @@ aarch64_emit_post_barrier (enum memmodel model)
 void
 aarch64_split_compare_and_swap (rtx operands[])
 {
+  /* Split after prolog/epilog to avoid interactions with shrinkwrapping.  */
+  gcc_assert (epilogue_completed);
+
   rtx rval, mem, oldval, newval, scratch, x, model_rtx;
   machine_mode mode;
   bool is_weak;
@@ -18469,6 +18472,9 @@ void
 aarch64_split_atomic_op (enum rtx_code code, rtx old_out, rtx new_out, rtx mem,
 			 rtx value, rtx model_rtx, rtx cond)
 {
+  /* Split after prolog/epilog to avoid interactions with shrinkwrapping.  */
+  gcc_assert (epilogue_completed);
+
   machine_mode mode = GET_MODE (mem);
   machine_mode wmode = (mode == DImode ? DImode : SImode);
   const enum memmodel model = memmodel_from_int (INTVAL (model_rtx));

gcc/config/aarch64/atomics.md

@@ -56,7 +56,7 @@
    (clobber (match_scratch:SI 7 "=&r"))]
   ""
   "#"
-  "&& reload_completed"
+  "&& epilogue_completed"
   [(const_int 0)]
   {
     aarch64_split_compare_and_swap (operands);
@@ -80,7 +80,7 @@
    (clobber (match_scratch:SI 7 "=&r"))]
   ""
   "#"
-  "&& reload_completed"
+  "&& epilogue_completed"
   [(const_int 0)]
   {
     aarch64_split_compare_and_swap (operands);
@@ -104,7 +104,7 @@
    (clobber (match_scratch:SI 7 "=&r"))]
   ""
   "#"
-  "&& reload_completed"
+  "&& epilogue_completed"
   [(const_int 0)]
   {
     aarch64_split_compare_and_swap (operands);
@@ -223,7 +223,7 @@
    (clobber (match_scratch:SI 4 "=&r"))]
   ""
   "#"
-  "&& reload_completed"
+  "&& epilogue_completed"
   [(const_int 0)]
   {
     aarch64_split_atomic_op (SET, operands[0], NULL, operands[1],
@@ -344,7 +344,7 @@
   (clobber (match_scratch:SI 4 "=&r"))]
   ""
   "#"
-  "&& reload_completed"
+  "&& epilogue_completed"
   [(const_int 0)]
   {
     aarch64_split_atomic_op (<CODE>, NULL, operands[3], operands[0],
@@ -400,7 +400,7 @@
    (clobber (match_scratch:SI 4 "=&r"))]
   ""
   "#"
-  "&& reload_completed"
+  "&& epilogue_completed"
   [(const_int 0)]
   {
      aarch64_split_atomic_op (NOT, NULL, operands[3], operands[0],
@@ -504,7 +504,7 @@
    (clobber (match_scratch:SI 5 "=&r"))]
   ""
   "#"
-  "&& reload_completed"
+  "&& epilogue_completed"
   [(const_int 0)]
   {
     aarch64_split_atomic_op (<CODE>, operands[0], operands[4], operands[1],
@@ -551,7 +551,7 @@
    (clobber (match_scratch:SI 5 "=&r"))]
   ""
   "#"
-  "&& reload_completed"
+  "&& epilogue_completed"
   [(const_int 0)]
   {
     aarch64_split_atomic_op (NOT, operands[0], operands[4], operands[1],
@@ -604,7 +604,7 @@
    (clobber (match_scratch:SI 4 "=&r"))]
   ""
   "#"
-  "&& reload_completed"
+  "&& epilogue_completed"
   [(const_int 0)]
   {
     aarch64_split_atomic_op (<CODE>, NULL, operands[0], operands[1],
@@ -628,7 +628,7 @@
    (clobber (match_scratch:SI 4 "=&r"))]
   ""
   "#"
-  "&& reload_completed"
+  "&& epilogue_completed"
   [(const_int 0)]
   {
     aarch64_split_atomic_op (NOT, NULL, operands[0], operands[1],