I think I must have messed up my previous testing of MIPS16 __sync stuff.
This patch fixes one bit of fallout, as seen in ia64-sync-1.c. If the
result of sync_lock_test_and_set is unused, the optab helper routine
maybe_emit_sync_lock_test_and_set gets called with a "target" of const0_rtx.
If the operation expands to a library call, maybe_emit_sync_lock_test_and_set
in turn passes const0_rtx down as the required target of the call.
We then get an ICE when moving the return register into const0_rtx.
The same problem applies to the atomic_ version.
Everywhere else in optabs.c just passes a null call target, so it seemed
simplest to do the same here.
Tested on mips64-linux-gnu. OK to install?
Richard
gcc/
* optabs.c (maybe_emit_sync_lock_test_and_set): Pass a null target
to emit_library_call_value.
(expand_atomic_compare_and_swap): Likewise.
Index: gcc/optabs.c
===================================================================
--- gcc/optabs.c 2011-12-03 16:19:48.000000000 +0000
+++ gcc/optabs.c 2011-12-04 08:52:12.000000000 +0000
@@ -7400,7 +7400,7 @@ maybe_emit_sync_lock_test_and_set (rtx t
rtx addr;
addr = convert_memory_address (ptr_mode, XEXP (mem, 0));
- return emit_library_call_value (libfunc, target, LCT_NORMAL,
+ return emit_library_call_value (libfunc, NULL_RTX, LCT_NORMAL,
mode, 2, addr, ptr_mode,
val, mode);
}
@@ -7637,7 +7637,7 @@ expand_atomic_compare_and_swap (rtx *pta
if (libfunc != NULL)
{
rtx addr = convert_memory_address (ptr_mode, XEXP (mem, 0));
- target_oval = emit_library_call_value (libfunc, target_oval, LCT_NORMAL,
+ target_oval = emit_library_call_value (libfunc, NULL_RTX, LCT_NORMAL,
mode, 3, addr, ptr_mode,
expected, mode, desired, mode);
