Make the default triple optional by allowing an empty string

[mehdi_amini]

Summary

When building LLVM as a (potentially dynamic) library that can be linked against
by multiple compilers, the default triple is not really meaningful.
We allow to explicitely set it to an empty string when configuring LLVM.
In this case, said "target independent" tests in the test suite that are using
the default triple are disabled by matching the newly available feature
"default_triple".

[jyknight]

I like this idea, but for a different reason than you're proposing it: it shows all of the tests which are depending on a notion of a "default triple", but *ought* to be fixed to run on all targets, instead.

[mehdi_amini]

James, I have another experimental patch that given a list of triple run multiple times these tests with a different triple every time. The implementation is a hack, I need to find a way to clean it.

[probinson]

I also like this tactic of permitting an empty default triple. Definitely an improvement over a silly default triple!

Does llvm-mc always need a triple? That is, would it be better to add test/tools/llvm-mc/lit.local.cfg (like you did for test/DebugInfo) instead of adding REQUIRES to each test/tools/llvm-mc/* file?

It seems weird for target-specific tests to require a default triple, but I guess those are the kinds of tests where you and jyknight are talking about iterating over some list of triples, and that seems like a good future direction.

Have you thought about moving test/DebugInfo/*.ll to a Generic subdirectory? Not saying you should or should not, but it seems possibly sub-optimal to require a default triple for *all* DebugInfo tests, even the target-specific ones. But maybe it's okay the way it is; I haven't done any particular research or thought about it much.

See also inline comments.

include/llvm/Config/config.h.cmake

424

Because this is different from all the other parameters, there should be a comment so nobody will think it's a typo and try to "fix" it.

test/DebugInfo/lit.local.cfg

4

Please fix the missing newline. Also there's an empty first line?

test/Feature/optnone-llc.ll

7

I have generally started associating "test uses llc" with "REQUIRES: default_triple" but here we have a test that doesn't? It seems like the configuration you use (host=X86, default-triple empty, only-target=GPU) should fail here. That is, I'd expect llc to fall back on the host (X86-ish) triple and therefore fail. Doesn't it?

test/lit.cfg

343

It would be good to have a comment briefly stating why we have a default_triple feature. I'm not aware of any other place where these strings are documented.

346

Given that we both misunderstood the purpose of this feature, it would be good to improve the commentary a little ("needed for JIT" or something like that).

[mehdi_amini]

In D12660#242029, @probinson wrote:

Does llvm-mc always need a triple? That is, would it be better to add test/tools/llvm-mc/lit.local.cfg (like you did for test/DebugInfo) instead of adding REQUIRES to each test/tools/llvm-mc/* file?

Created test/tools/llvm-mc/Generic/

It seems weird for target-specific tests to require a default triple, but I guess those are the kinds of tests where you and jyknight are talking about iterating over some list of triples, and that seems like a good future direction.

Do you mean the ones in test/CodeGen/PowerPC, test/CodeGen/X86, etc.?
They really either shouldn't be in these directory, or should specify a triple. This is a bit messy but I wasn't sure which triple to define for these.

Have you thought about moving test/DebugInfo/*.ll to a Generic subdirectory? Not saying you should or should not, but it seems possibly sub-optimal to require a default triple for *all* DebugInfo tests, even the target-specific ones. But maybe it's okay the way it is; I haven't done any particular research or thought about it much.

See also inline comments.

test/Feature/optnone-llc.ll

7

This was a mistake (I ran the tests with a release build and was covered by the "REQUIRES: asserts")

test/lit.cfg

346

What about renaming it then?
I can do that in a separate patch.

[mehdi_amini]

Updated taking comments into account.

[mehdi_amini]

In D12660#243011, @joker.eph wrote:

In D12660#242029, @probinson wrote:

Does llvm-mc always need a triple? That is, would it be better to add test/tools/llvm-mc/lit.local.cfg (like you did for test/DebugInfo) instead of adding REQUIRES to each test/tools/llvm-mc/* file?

Created test/tools/llvm-mc/Generic/

I submitted unintentionally. I added a lit.local.cfg.

It seems weird for target-specific tests to require a default triple, but I guess those are the kinds of tests where you and jyknight are talking about iterating over some list of triples, and that seems like a good future direction.

Do you mean the ones in test/CodeGen/PowerPC, test/CodeGen/X86, etc.?
They really either shouldn't be in these directory, or should specify a triple. This is a bit messy but I wasn't sure which triple to define for these.

Have you thought about moving test/DebugInfo/*.ll to a Generic subdirectory? Not saying you should or should not, but it seems possibly sub-optimal to require a default triple for *all* DebugInfo tests, even the target-specific ones. But maybe it's okay the way it is; I haven't done any particular research or thought about it much.

Moved as suggested.

Thanks for the comments!

[mehdi_amini]

add missing lit.local.cfg

[probinson]

+echristo as debug-info owner, to judge whether moving the top-level DebugInfo tests to a Generic subdirectory is okay.

Two minor comment suggestions, everything else LGTM.

test/lit.cfg

344

Comments should end with a period.

346

I think the name is descriptive, just an example of how it's used would help clarify things.

[echristo]

I'll trust that you're all going in a good direction here. Thanks for checking in!

-eric

[mehdi_amini]

Thanks, r247775.

include/llvm/Config/config.h.cmake

 
 /* Time at which LLVM was configured */
 #cmakedefine LLVM_CONFIGTIME "${LLVM_CONFIGTIME}"
 
 /* Installation directory for data files */
 #cmakedefine LLVM_DATADIR "${LLVM_DATADIR}"
 
 /* Target triple LLVM will generate code for by default */
-#cmakedefine LLVM_DEFAULT_TARGET_TRIPLE "${LLVM_DEFAULT_TARGET_TRIPLE}"
+#define LLVM_DEFAULT_TARGET_TRIPLE "${LLVM_DEFAULT_TARGET_TRIPLE}"
 
 /* Installation directory for documentation */
 #cmakedefine LLVM_DOCSDIR "${LLVM_DOCSDIR}"
 
 /* Define if LLVM is built with asserts and checks that change the layout of
    client-visible data structures.  */
 #cmakedefine LLVM_ENABLE_ABI_BREAKING_CHECKS
 

test/BugPoint/remove_arguments_test.ll

 ; RUN: bugpoint -load %llvmshlibdir/BugpointPasses%shlibext %s -output-prefix %t -bugpoint-crashcalls -silence-passes
 ; RUN: llvm-dis %t-reduced-simplified.bc -o - | FileCheck %s
 ; REQUIRES: loadable_module
 
 ; Test to make sure that arguments are removed from the function if they are
 ; unnecessary. And clean up any types that frees up too.
 
-; CHECK: target triple
+; CHECK: ModuleID
 ; CHECK-NOT: struct.anon
 %struct.anon = type { i32 }
 
 declare i32 @test2()
 
 ; CHECK: define void @test() {
 define i32 @test(i32 %A, %struct.anon* %B, float %C) {
   call i32 @test2()
   ret i32 %1
 }

test/CodeGen/Generic/lit.local.cfg

-if 'native' not in config.available_features:
-    config.unsupported = True
 
+if not config.target_triple:
+    config.unsupported = True
 No newline at end of file

test/CodeGen/PowerPC/2006-01-20-ShiftPartsCrash.ll

 ; RUN: llc < %s
+; REQUIRES: default_triple
 
 define void @iterative_hash_host_wide_int() {
         %zero = alloca i32              ; <i32*> [#uses=2]
         %b = alloca i32         ; <i32*> [#uses=1]
         store i32 0, i32* %zero
         %tmp = load i32, i32* %zero          ; <i32> [#uses=1]
         %tmp5 = bitcast i32 %tmp to i32         ; <i32> [#uses=1]
         %tmp6.u = add i32 %tmp5, 32             ; <i32> [#uses=1]

test/CodeGen/PowerPC/2006-08-15-SelectionCrash.ll

 ; RUN: llc < %s
+; REQUIRES: default_triple
 
   %struct..0anon = type { i32 }
   %struct.rtx_def = type { i16, i8, i8, [1 x %struct..0anon] }
 
 define fastcc void @immed_double_const(i32 %i0, i32 %i1) {
 entry:
   %tmp1 = load i32, i32* null    ; <i32> [#uses=1]
   switch i32 %tmp1, label %bb103 [

test/CodeGen/PowerPC/2006-12-07-LargeAlloca.ll

 ; RUN: llc < %s -march=ppc64
 ; RUN: llc < %s -march=ppc32
 ; RUN: llc < %s 
+; REQUIRES: default_triple
 
 define void @bitap() {
 entry:
         %RMask.i = alloca [256 x i32], align 16         ; <[256 x i32]*> [#uses=1]
         %buffer = alloca [147456 x i8], align 16                ; <[147456 x i8]*> [#uses=0]
         br i1 false, label %bb19, label %bb.preheader
 bb.preheader:           ; preds = %entry
         ret void

test/CodeGen/PowerPC/2006-12-07-SelectCrash.ll

 ; RUN: llc < %s -march=ppc64
 ; RUN: llc < %s -march=ppc32
 ; RUN: llc < %s
+; REQUIRES: default_triple
 
 @qsz.b = external global i1             ; <i1*> [#uses=1]
 
 define fastcc void @qst() {
 entry:
         br i1 true, label %cond_next71, label %cond_true
 cond_true:              ; preds = %entry
         ret void

test/CodeGen/PowerPC/2007-11-19-VectorSplitting.ll

 ; RUN: llc < %s 
 ; RUN: llc < %s -march=ppc32 -mcpu=g3
 ; RUN: llc < %s -march=ppc32 -mcpu=g5
 ; PR1811
+; REQUIRES: default_triple
 
 define void @execute_shader(<4 x float>* %OUT, <4 x float>* %IN, <4 x float>*
 %CONST) {
 entry:
         %input2 = load <4 x float>, <4 x float>* null, align 16               ; <<4 x float>>
          %shuffle7 = shufflevector <4 x float> %input2, <4 x float> < float 0.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00 >, <4 x i32> < i32 2, i32 2, i32 2, i32 2 >    ; <<4 x float>> [#uses=1]
 
         %mul1 = fmul <4 x float> %shuffle7, zeroinitializer              ; <<4 x
         %add2 = fadd <4 x float> %mul1, %input2          ; <<4 x float>>
         store <4 x float> %add2, <4 x float>* null, align 16
         ret void
 }

test/CodeGen/PowerPC/load-shift-combine.ll

 ; RUN: llc < %s
+; REQUIRES: default_triple
 
 ; This used to cause a crash.  A standard load is converted to a pre-increment
 ; load.  Later the pre-increment load is combined with a subsequent SRL to
 ; produce a smaller load.  This transform invalidly created a standard load
 ; and propagated the produced value into uses of both produced values of the
 ; pre-increment load.  The result was a crash when attempting to process an
 ; add with a token-chain operand.
 

test/CodeGen/X86/2006-10-02-BoolRetCrash.ll

 ; RUN: llc < %s 
 ; PR933
+; REQUIRES: default_triple
 
 define fastcc i1 @test() {
         ret i1 true
 }
 

test/CodeGen/X86/2009-06-05-VariableIndexInsert.ll

 ; RUN: llc < %s
+; REQUIRES: default_triple
 
 define <2 x i64> @_mm_insert_epi16(<2 x i64> %a, i32 %b, i32 %imm) nounwind readnone {
 entry:
   %conv = bitcast <2 x i64> %a to <8 x i16>    ; <<8 x i16>> [#uses=1]
   %conv2 = trunc i32 %b to i16    ; <i16> [#uses=1]
   %and = and i32 %imm, 7    ; <i32> [#uses=1]
   %vecins = insertelement <8 x i16> %conv, i16 %conv2, i32 %and    ; <<8 x i16>> [#uses=1]
   %conv6 = bitcast <8 x i16> %vecins to <2 x i64>    ; <<2 x i64>> [#uses=1]
   ret <2 x i64> %conv6
 }

test/CodeGen/X86/2009-06-06-ConcatVectors.ll

 ; RUN: llc < %s
+; REQUIRES: default_triple
 
 define <2 x i64> @_mm_movpi64_pi64(<1 x i64> %a, <1 x i64> %b) nounwind readnone {
 entry:
   %0 = shufflevector <1 x i64> %a, <1 x i64> %b, <2 x i32> <i32 0, i32 1>
   ret <2 x i64> %0
 }
 

test/CodeGen/X86/2010-02-01-DbgValueCrash.ll

 ; RUN: llc -O1 < %s
+; REQUIRES: default_triple
+
 ; ModuleID = 'pr6157.bc'
 ; formerly crashed in SelectionDAGBuilder
 
 %tart.reflect.ComplexType = type { double, double }
 
 @.type.SwitchStmtTest = constant %tart.reflect.ComplexType { double 3.0, double 2.0 }
 
 define i32 @"main(tart.core.String[])->int32"(i32 %args) {

test/CodeGen/X86/2010-07-06-DbgCrash.ll

 ; RUN: llc -O0 -relocation-model pic < %s -o /dev/null
+; REQUIRES: default_triple
 ; PR7545
 @.str = private constant [4 x i8] c"one\00", align 1 ; <[4 x i8]*> [#uses=1]
 @.str1 = private constant [4 x i8] c"two\00", align 1 ; <[5 x i8]*> [#uses=1]
 @C.9.2167 = internal constant [2 x i8*] [i8* getelementptr inbounds ([4 x i8], [4 x i8]* @.str, i64 0, i64 0), i8* getelementptr inbounds ([4 x i8], [4 x i8]* @.str1, i64 0, i64 0)]
 !38 = !DIFile(filename: "pbmsrch.c", directory: "/Users/grawp/LLVM/test-suite/MultiSource/Benchmarks/MiBench/office-stringsearch")
 !39 = distinct !DICompileUnit(language: DW_LANG_C89, producer: "4.2.1 (Based on Apple Inc. build 5658) (LLVM build 9999)", isOptimized: true, emissionKind: 0, file: !109, enums: !108, retainedTypes: !108)
 !46 = !DIDerivedType(tag: DW_TAG_pointer_type, size: 64, align: 64, file: !109, baseType: !47)
 !47 = !DIBasicType(tag: DW_TAG_base_type, name: "char", size: 8, align: 8, encoding: DW_ATE_signed_char)

test/CodeGen/X86/GC/alloc_loop.ll

 ; RUN: llc < %s
+; REQUIRES: default_triple
 
 
 declare i8* @llvm_gc_allocate(i32)
 declare void @llvm_gc_initialize(i32)
 
 declare void @llvm.gcroot(i8**, i8*)
 declare void @llvm.gcwrite(i8*, i8*, i8**)
 

test/CodeGen/X86/GC/cg-O0.ll

 ; RUN: llc < %s -O0
+; REQUIRES: default_triple
 
 define i32 @main() {
 entry:
   call void @f()
   ret i32 0
 }
 
 define void @f() gc "ocaml" {
 entry:
   %ptr.stackref = alloca i8*
   %gcroot = bitcast i8** %ptr.stackref to i8**
   call void @llvm.gcroot(i8** %gcroot, i8* null)
   ret void
 }
 
 declare void @llvm.gcroot(i8**, i8*) nounwind

test/CodeGen/X86/GC/lower_gcroot.ll

 ; RUN: llc < %s
+; REQUIRES: default_triple
 
   %Env = type i8*
 
 define void @.main(%Env) gc "shadow-stack" {
   %Root = alloca %Env
   call void @llvm.gcroot( %Env* %Root, %Env null )
   unreachable
 }
 
 declare void @llvm.gcroot(%Env*, %Env)

test/CodeGen/X86/extractelement-shuffle.ll

 ; RUN: llc < %s
+; REQUIRES: default_triple
 
 ; Examples that exhibits a bug in DAGCombine.  The case is triggered by the
 ; following program.  The bug is DAGCombine assumes that the bit convert
 ; preserves the number of elements so the optimization code tries to read
 ; through the 3rd mask element, which doesn't exist.
 define i32 @update(<2 x i64> %val1, <2 x i64> %val2) nounwind readnone {
 entry:
   %shuf = shufflevector <2 x i64> %val1, <2 x i64> %val2, <2 x i32> <i32 0, i32 3>
   %bit  = bitcast <2 x i64> %shuf to <4 x i32>
   %res =  extractelement <4 x i32> %bit, i32 3
   ret i32 %res
 }

test/CodeGen/X86/vshift_scalar.ll

 ; RUN: llc < %s
+; REQUIRES: default_triple
 
 ; Legalization test that requires scalarizing a vector.
 
 define void @update(<1 x i32> %val, <1 x i32>* %dst) nounwind {
 entry:
   %shl = shl <1 x i32> %val, < i32 2>
   %shr = ashr <1 x i32> %val, < i32 4>
   store <1 x i32> %shr, <1 x i32>* %dst
   ret void
 }

test/DebugInfo/lit.local.cfg

-if 'native' not in config.available_features:
-    config.unsupported = True
 
+if not config.target_triple:
+    config.unsupported = True
 No newline at end of file

test/Feature/optnone-llc.ll

 ; RUN: llc -O0 -debug %s -o /dev/null 2>&1 | FileCheck %s --check-prefix=LLC-O0
 ; RUN: llc -O1 -debug %s -o /dev/null 2>&1 | FileCheck %s --check-prefix=LLC-Ox
 ; RUN: llc -O2 -debug %s -o /dev/null 2>&1 | FileCheck %s --check-prefix=LLC-Ox
 ; RUN: llc -O3 -debug %s -o /dev/null 2>&1 | FileCheck %s --check-prefix=LLC-Ox
 ; RUN: llc -misched-postra -debug %s -o /dev/null 2>&1 | FileCheck %s --check-prefix=LLC-MORE
 
-; REQUIRES: asserts, native
+; REQUIRES: asserts
 
 ; This test verifies that we don't run Machine Function optimizations
 ; on optnone functions.
 
 ; Function Attrs: noinline optnone
 define i32 @_Z3fooi(i32 %x) #0 {
 entry:
   %x.addr = alloca i32, align 4

test/Feature/optnone.ll

 ; RUN: llvm-as < %s | llvm-dis | FileCheck %s
-; REQUIRE: native
 
 ; Check for the presence of attribute optnone in the disassembly.
 
 ; CHECK: @foo() #0
 define void @foo() #0 {
   ret void
 }
 
 ; CHECK: attributes #0 = { noinline optnone }
 attributes #0 = { optnone noinline }
 

test/LTO/X86/bcsection.ll

 ; RUN: llvm-lto -exported-symbol=main -exported-symbol=_main -o %T/bcsection.elf.o %T/bcsection.elf.bco
 ; RUN: llvm-nm %T/bcsection.elf.o | FileCheck %s
 
 ; RUN: llvm-mc -I=%T -filetype=obj -triple=x86_64-apple-darwin11 -o %T/bcsection.macho.bco %p/Inputs/bcsection.macho.s
 ; RUN: llvm-nm %T/bcsection.macho.bco | FileCheck %s
 ; RUN: llvm-lto -exported-symbol=main -exported-symbol=_main -o %T/bcsection.macho.o %T/bcsection.macho.bco
 ; RUN: llvm-nm %T/bcsection.macho.o | FileCheck %s
 
+; REQUIRES: default_triple
+
 ; CHECK: main
 define i32 @main() {
   ret i32 0
 }

test/LTO/X86/current-section.ll

 ; RUN: llvm-as < %s >%t1
 ; RUN: llvm-lto -o %t2 %t1
+; REQUIRES: default_triple
 
 module asm ".align 4"

test/LTO/X86/list-symbols.ll

 ; RUN: llvm-as -o %T/1.bc %s
 ; RUN: llvm-as -o %T/2.bc %S/Inputs/list-symbols.ll
 ; RUN: llvm-lto -list-symbols-only %T/1.bc %T/2.bc | FileCheck %s
+; REQUIRES: default_triple
 
 ; CHECK-LABEL: 1.bc:
 ; CHECK-DAG: foo
 ; CHECK-DAG: glob
 ; CHECK-LABEL: 2.bc:
 ; CHECK-DAG: glob
 ; CHECK-DAG: bar
 
 @glob = global i32 0
 define void @foo() {
   ret void
 }

test/Linker/subprogram-linkonce-weak.ll

-; REQUIRES: native
 ; RUN: llvm-link %s %S/Inputs/subprogram-linkonce-weak.ll -S -o %t1
 ; RUN: FileCheck %s -check-prefix=LW -check-prefix=CHECK <%t1
 ; RUN: llvm-link %S/Inputs/subprogram-linkonce-weak.ll %s -S -o %t2
 ; RUN: FileCheck %s -check-prefix=WL -check-prefix=CHECK <%t2
+; REQUIRES: default_triple
 
 ; This testcase tests the following flow:
 ;  - File A defines a linkonce version of @foo which has inlined into @bar.
 ;  - File B defines a weak version of @foo (different definition).
 ;  - Linkage rules state File B version of @foo wins.
 ;  - @bar still has inlined debug info related to the linkonce @foo.
 ;
 ; This should fix PR22792, although the testcase was hand-written.  There's a

test/Linker/type-unique-odr-a.ll

-; REQUIRES: object-emission, native
+; REQUIRES: default_triple
+; REQUIRES: object-emission
 ;
 ; RUN: llvm-link %s %p/type-unique-odr-b.ll -S -o - | %llc_dwarf -dwarf-linkage-names=Enable -filetype=obj -O0 | llvm-dwarfdump -debug-dump=info - | FileCheck %s
 ;
 ; Test ODR-based type uniquing for C++ class members.
 ; rdar://problem/15851313.
 ;
 ; $ cat -n type-unique-odr-a.cpp
 ;     1 class A {

test/Linker/type-unique-simple-a.ll

-; REQUIRES: object-emission, native
+; REQUIRES: object-emission
+; REQUIRES: default_triple
 
 ; RUN: llvm-link %s %p/type-unique-simple-b.ll -S -o %t
 ; RUN: cat %t | FileCheck %s -check-prefix=LINK
 ; RUN: %llc_dwarf -filetype=obj -O0 < %t > %t2
 ; RUN: llvm-dwarfdump -debug-dump=info %t2 | FileCheck %s
 
 ; Make sure the backend generates a single DIE and uses ref_addr.
 ; CHECK: 0x[[BASE:.*]]: DW_TAG_structure_type

test/Linker/type-unique-simple2-a.ll

-; REQUIRES: object-emission, native
+; REQUIRES: object-emission
+; REQUIRES: default_triple
 ;
 ; RUN: llvm-link %s %p/type-unique-simple2-b.ll -S -o - | %llc_dwarf -filetype=obj -O0 | llvm-dwarfdump -debug-dump=info - | FileCheck %s
 ;
 ; Tests for a merge error where attributes are inserted twice into the same DIE.
 ;
 ; $ cat ab.h
 ; typedef int foo_t;
 ; class A {

test/Linker/type-unique-simple2.ll

-; REQUIRES: object-emission; native
+; REQUIRES: object-emission
+; REQUIRES: default_triple
 
 ; RUN: llvm-link %S/Inputs/type-unique-simple2-a.ll %S/Inputs/type-unique-simple2-b.ll -S -o %t
 ; RUN: cat %t | FileCheck %S/Inputs/type-unique-simple2-a.ll -check-prefix=LINK
 ; RUN: %llc_dwarf -filetype=obj -O0 < %t > %t2
 ; RUN: llvm-dwarfdump -debug-dump=info %t2 | FileCheck %S/Inputs/type-unique-simple2-a.ll

test/Linker/type-unique-type-array-a.ll

-; REQUIRES: object-emission, native
+; REQUIRES: object-emission
+; REQUIRES: default_triple
 ;
 ; RUN: llvm-link %s %p/type-unique-type-array-b.ll -S -o - | %llc_dwarf -filetype=obj -O0 | llvm-dwarfdump -debug-dump=info - | FileCheck %s
 ;
 ; rdar://problem/17628609
 ;
 ; cat -n a.cpp
 ;     1 struct SA {
 ;     2   int a;

test/MC/ELF/cfi-version.ll

 ; RUN: %llc_dwarf %s -o - -dwarf-version 2 -filetype=obj | llvm-dwarfdump - | FileCheck %s --check-prefix=DWARF2
 ; RUN: %llc_dwarf %s -o - -dwarf-version 3 -filetype=obj | llvm-dwarfdump - | FileCheck %s --check-prefix=DWARF3
 ; RUN: %llc_dwarf %s -o - -dwarf-version 4 -filetype=obj | llvm-dwarfdump - | FileCheck %s --check-prefix=DWARF4
 
 ; .debug_frame is not emitted for targeting Windows x64.
 ; REQUIRES: debug_frame
+; REQUIRES: default_triple
 
 ; Function Attrs: nounwind
 define i32 @foo() #0 {
 entry:
   %call = call i32 bitcast (i32 (...)* @bar to i32 ()*)(), !dbg !12
   %add = add nsw i32 %call, 1, !dbg !12
   ret i32 %add, !dbg !12
 }

test/Other/2010-05-06-Printer.ll

-; REQUIRES: native
 ; RUN: llc -O2 -print-after-all < %s 2>/dev/null
+; REQUIRES: default_triple
 
 define void @tester(){
   ret void
 }
 

test/Transforms/BranchFolding/2007-10-19-InlineAsmDirectives.ll

 ; RUN: opt < %s -O3 | llc -no-integrated-as | FileCheck %s
-; REQUIRES: X86
+; REQUIRES: default_triple
+
 ;; We don't want branch folding to fold asm directives.
 
 ; CHECK: bork_directive
 ; CHECK: bork_directive
 ; CHECK-NOT: bork_directive
 
 define void @bork(i32 %param) {
 entry:
   %tmp = icmp eq i32 %param, 0
         br i1 %tmp, label %cond_true, label %cond_false
 
-cond_true:   
+cond_true:
         call void asm sideeffect ".bork_directive /* ${0:c}:${1:c} */", "i,i,~{dirflag},~{fpsr},~{flags}"( i32 37, i32 927 )
         ret void
 
 cond_false:
   call void asm sideeffect ".foo_directive ${0:c}:${1:c}", "i,i,~{dirflag},~{fpsr},~{flags}"( i32 37, i32 927 )
         call void asm sideeffect ".bork_directive /* ${0:c}:${1:c} */", "i,i,~{dirflag},~{fpsr},~{flags}"( i32 37, i32 927 )
         ret void
 }

test/lit.cfg

 if not 'hexagon' in config.target_triple and not re.match(r'^x86_64.*-(mingw32|windows-gnu|win32)', config.target_triple):
     config.available_features.add("object-emission")
 
 if config.have_zlib == "1":
     config.available_features.add("zlib")
 else:
     config.available_features.add("nozlib")
 
-# Native compilation: host arch == target arch and native backend built-in
+if config.target_triple:
+    config.available_features.add("default_triple")
+
+# Native compilation: host arch == default triple arch
 # FIXME: Consider cases that target can be executed
 # even if host_triple were different from target_triple.
-if (config.host_triple == config.target_triple and
-    config.native_target in config.targets_to_build):
+if config.host_triple == config.target_triple:
     config.available_features.add("native")
 
 import subprocess
 
 def have_ld_plugin_support():
     if not os.path.exists(os.path.join(config.llvm_shlib_dir, 'LLVMgold.so')):
         return False
 

test/tools/llvm-mc/fatal_warnings.test

 # RUN: not llvm-mc --fatal-warnings %s 2>&1 | FileCheck %s
 # XFAIL: hexagon
-# REQUIRES: native
+; REQUIRES: default_triple
 
 # CHECK: error: .warning directive invoked in source file
 .warning

test/tools/llvm-mc/line_end_with_space.test

 RUN: llvm-mc -disassemble %s
-# REQUIRES: native
+; REQUIRES: default_triple
+ 

test/tools/llvm-mc/no_warnings.test

 # RUN: llvm-mc --no-warn %s 2>&1 | FileCheck %s
 # XFAIL: hexagon
-# REQUIRES: native
+; REQUIRES: default_triple
 
 # CHECK-NOT: warning:
 .warning