[llvm] r222720 - Add an interesting test that we already get right. NFC.

[Rafael Espíndola]

On 24 November 2014 at 23:57, Sean Silva <chisophugis at gmail.com> wrote:
> What is the meaning of "unrelated" in this context?

The linker has not seen two values it wants to link where each has one
of those types. That only happens when it gets to

define %t @g() {
 %x = call %t @f()
 ret %t %x
}

Where it becomes clear that it is advantageous for @g and @f to return
the same type.

> -- Sean Silva
>
> On Mon, Nov 24, 2014 at 7:47 PM, Rafael Espindola
> <rafael.espindola at gmail.com> wrote:
>>
>> Author: rafael
>> Date: Mon Nov 24 21:47:57 2014
>> New Revision: 222720
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=222720&view=rev
>> Log:
>> Add an interesting test that we already get right. NFC.
>>
>> Added:
>>     llvm/trunk/test/Linker/Inputs/type-unique-unrelated2.ll
>>     llvm/trunk/test/Linker/Inputs/type-unique-unrelated3.ll
>>     llvm/trunk/test/Linker/type-unique-unrelated.ll
>>
>> Added: llvm/trunk/test/Linker/Inputs/type-unique-unrelated2.ll
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Linker/Inputs/type-unique-unrelated2.ll?rev=222720&view=auto
>>
>> ==============================================================================
>> --- llvm/trunk/test/Linker/Inputs/type-unique-unrelated2.ll (added)
>> +++ llvm/trunk/test/Linker/Inputs/type-unique-unrelated2.ll Mon Nov 24
>> 21:47:57 2014
>> @@ -0,0 +1,7 @@
>> +%t = type { i8* }
>> +declare %t @g()
>> +
>> +define %t @g2() {
>> + %x = call %t @g()
>> + ret %t %x
>> +}
>>
>> Added: llvm/trunk/test/Linker/Inputs/type-unique-unrelated3.ll
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Linker/Inputs/type-unique-unrelated3.ll?rev=222720&view=auto
>>
>> ==============================================================================
>> --- llvm/trunk/test/Linker/Inputs/type-unique-unrelated3.ll (added)
>> +++ llvm/trunk/test/Linker/Inputs/type-unique-unrelated3.ll Mon Nov 24
>> 21:47:57 2014
>> @@ -0,0 +1,7 @@
>> +%t = type { i8* }
>> +declare %t @f()
>> +
>> +define %t @g() {
>> + %x = call %t @f()
>> + ret %t %x
>> +}
>>
>> Added: llvm/trunk/test/Linker/type-unique-unrelated.ll
>> URL:
>> http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Linker/type-unique-unrelated.ll?rev=222720&view=auto
>>
>> ==============================================================================
>> --- llvm/trunk/test/Linker/type-unique-unrelated.ll (added)
>> +++ llvm/trunk/test/Linker/type-unique-unrelated.ll Mon Nov 24 21:47:57
>> 2014
>> @@ -0,0 +1,31 @@
>> +; RUN: llvm-link -S %s %p/Inputs/type-unique-unrelated2.ll
>> %p/Inputs/type-unique-unrelated3.ll | FileCheck %s
>> +
>> +; CHECK: %t = type { i8* }
>> +
>> +; CHECK: define %t @f2() {
>> +; CHECK-NEXT:   %x = call %t @f2()
>> +; CHECK-NEXT:   ret %t %x
>> +; CHECK-NEXT: }
>> +
>> +; CHECK: define %t @g2() {
>> +; CHECK-NEXT:   %x = call %t @g()
>> +; CHECK-NEXT:   ret %t %x
>> +; CHECK-NEXT: }
>> +
>> +; CHECK: define %t @g() {
>> +; CHECK-NEXT:  %x = call %t @f()
>> +; CHECK-NEXT:  ret %t %x
>> +; CHECK-NEXT: }
>> +
>> +; The idea of this test is that the %t in this file and the one in
>> +; type-unique-unrelated2.ll look unrelated until
>> type-unique-unrelated3.ll
>> +; is merged in.
>> +
>> +%t = type { i8* }
>> +declare %t @f()
>> +
>> +define %t @f2() {
>> + %x = call %t @f2()
>> + ret %t %x
>> +}
>> +
>>
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
>