Rebased PNaCl localmods in LLVM to 223109

test/MC/X86/AlignedBundling/nacl-call-auto-align-to-end.s

+# RUN: llvm-mc -filetype=obj -triple i686-unknown-nacl %s -o - \
+# RUN:   | llvm-objdump -disassemble -no-show-raw-insn - | FileCheck %s
+
+# RUN: llvm-mc -filetype=obj -triple x86_64-unknown-nacl \
+# RUN: -sfi-hide-sandbox-base=false %s -o - \
+# RUN:   | llvm-objdump -disassemble -no-show-raw-insn - | FileCheck %s
+
+# RUN: llvm-mc -filetype=obj -triple x86_64-unknown-nacl \
+# RUN: -sfi-hide-sandbox-base=true %s -o - \
+# RUN:   | llvm-objdump -disassemble -no-show-raw-insn - \
+# RUN:   | FileCheck --check-prefix=HIDE %s
+
+# Test that bare call instructions in NaCl are automatically aligned to the end
+# of a bundle without the need for .bundle_align directives, to match gas's
+# behavior.
+  .text
+foo:
+# Each of these mov instructions is 4 bytes long.
+  movsd %xmm1,%xmm2
+  movsd %xmm1,%xmm2
+  movsd %xmm1,%xmm2
+  movsd %xmm1,%xmm2
+# Each of these movs is 5 bytes long.
+  movl $1, %eax
+  movl $1, %eax
+  call   bar
+# To align this call to a bundle end, we need a 1-byte NOP.
+# CHECK:        1a:  nop
+# CHECK-NEXT:   1b: call
+
+  movsd %xmm1,%xmm2
+  movsd %xmm1,%xmm2
+  movsd %xmm1,%xmm2
+  movsd %xmm1,%xmm2
+  movl $1, %eax
+  movl $1, %eax
+  movl $1, %eax
+  call   bar
+# Here we have to pad until the end of the *next* boundary because
+# otherwise the call crosses a boundary.
+# The last byte of the bundle has to be a 1-byte nop so it doesn't
+# cross the boundary itself.
+# CHECK:      3f: nop
+# The remaining nops can be implemented any way the compiler wants.
+# CHECK:      5b: call
+
+# HIDE-NOT: call