Weight variables by their number of uses for register allocation.

tests_lit/assembler/x86/jump_encodings.ll

Index: tests_lit/assembler/x86/jump_encodings.ll
diff --git a/tests_lit/assembler/x86/jump_encodings.ll b/tests_lit/assembler/x86/jump_encodings.ll
index d94be0a22212e74df02d22ae15657cfc3f55334c..22587219e29fc8e482937b7c2278f590a7653e33 100644
--- a/tests_lit/assembler/x86/jump_encodings.ll
+++ b/tests_lit/assembler/x86/jump_encodings.ll
@@ -169,13 +169,13 @@ next2:
   br i1 %cmp, label %next, label %next2
 }
 ; CHECK-LABEL: test_local_forward_then_back
-; CHECK:      14: {{.*}} mov DWORD PTR
-; CHECK-NEXT: 16: {{.*}} mfence
-; CHECK-NEXT: 19: {{.*}} mov DWORD PTR {{.*}},0x1
-; CHECK-NEXT: 20: {{.*}} cmp
-; CHECK-NEXT: 23: {{.*}} jb 33
-; CHECK:      37: {{.*}} jne 14
-; CHECK:      39: {{.*}} jmp 19
+; CHECK:      {{.*}} mov DWORD PTR
+; CHECK-NEXT: {{.*}} mfence
+; CHECK-NEXT: {{.*}} mov {{.*}},0x1

[jvoung (off chromium), 2015/08/28 20:31:45]

Hmm yes, this test is pretty brittle and annoying to update. However, removing
the instruction addresses and jmp targets mean it's not testing anything new
(the "make sure that the instruction size accounting for the forward branches
are correct").

On possible middle-ground is to only check the jmp to %next2 and the address of
%next2.

That is, keep the address of mov {{.*}},0x1 (the first instruction for %next2),
and keep the target of the last jmp (which goes to %next2). For me, the address
is 0x16.

[ascull, 2015/08/28 21:22:36]

Done.

+; CHECK-NEXT: {{.*}} cmp
+; CHECK-NEXT: {{.*}} jb
+; CHECK:      {{.*}} jne
+; CHECK:      {{.*}} jmp
 
 
 ; Test that backward local branches also work and are small.