Add TargetLowering::prepareVolatileOrAtomicLoad
One unusual feature of the z architecture is that the result of a
previous load can be reused indefinitely for subsequent loads, even if
a cache-coherent store to that location is performed by another CPU.
A special serializing instruction must be used if you want to force
a load to be reattempted.
Since volatile loads are not supposed to be omitted in this way,
we should insert a serializing instruction before each such load.
The same goes for atomic loads.
The patch implements this at the IR->DAG boundary, in a similar way
to atomic fences. It is a no-op for targets other than SystemZ.
llvm-svn: 196905
diff --git a/llvm/test/CodeGen/SystemZ/Large/branch-range-05.py b/llvm/test/CodeGen/SystemZ/Large/branch-range-05.py
index 6928b8f..52f4a96 100644
--- a/llvm/test/CodeGen/SystemZ/Large/branch-range-05.py
+++ b/llvm/test/CodeGen/SystemZ/Large/branch-range-05.py
@@ -82,7 +82,7 @@
for i in xrange(branch_blocks):
next = 'before%d' % (i + 1) if i + 1 < branch_blocks else 'main'
print 'before%d:' % i
- print ' %%bcur%d = load volatile i8 *%%stop' % i
+ print ' %%bcur%d = load i8 *%%stop' % i
print ' %%bext%d = sext i8 %%bcur%d to i32' % (i, i)
print ' %%btest%d = icmp slt i32 %%bext%d, %d' % (i, i, i + 50)
print ' br i1 %%btest%d, label %%after0, label %%%s' % (i, next)
@@ -98,7 +98,7 @@
print ' store volatile i8 %d, i8 *%%ptr%d' % (value, i)
for i in xrange(branch_blocks):
- print ' %%acur%d = load volatile i8 *%%stop' % i
+ print ' %%acur%d = load i8 *%%stop' % i
print ' %%aext%d = sext i8 %%acur%d to i32' % (i, i)
print ' %%atest%d = icmp slt i32 %%aext%d, %d' % (i, i, i + 100)
print ' br i1 %%atest%d, label %%main, label %%after%d' % (i, i)