[cfe-dev] make check failures

[Jay Foad]

On 29 November 2010 23:33, dalej <dalej at apple.com> wrote:
>
> On Nov 26, 2010, at 2:11 AM, Jay Foad wrote:
>
>> I've just tried "make check" on a fresh checkout of llvm and clang. I get:
>
> These tests are actually testing llvm-gcc, not llvm.  You can almost certainly get them to pass by building llvm-gcc and keeping it up to date.

When I configure LLVM --with-llvmgcc --with-llvmgxx, make check-all
reports loads of stuff ending with:

    [...]
    LLVM :: Transforms/PartialSpecialize/heuristics.ll
    LLVM :: Transforms/SCCP/ipsccp-addr-taken.ll
    LLVM :: Transforms/ScalarRepl/memcpy-from-global.ll
    LLVM :: Transforms/ScalarRepl/vector_promote.ll
    LLVM :: Transforms/SimplifyCFG/basictest.ll
    LLVM :: Transforms/SimplifyCFG/indirectbr.ll
    LLVM :: Transforms/SimplifyCFG/speculate-with-offset.ll
    LLVM :: Transforms/SimplifyLibCalls/Puts.ll
    LLVM :: Transforms/SimplifyLibCalls/StrPBrk.ll
    LLVM :: Transforms/SimplifyLibCalls/StrRChr.ll
    LLVM :: Transforms/SimplifyLibCalls/StrSpn.ll
    LLVM :: Verifier/2010-08-07-PointerIntrinsic.ll

  Expected Passes    : 7684
  Expected Failures  : 69
  Unsupported Tests  : 69
  Unexpected Passes  : 3
  Unexpected Failures: 571
make[1]: *** [check-local-all] Error 1


Here's one of the failures, picked at random:


********************
FAIL: LLVM :: Transforms/ScalarRepl/vector_promote.ll (7914 of 8396)
******************** TEST 'LLVM ::
Transforms/ScalarRepl/vector_promote.ll' FAILED ********************
Script:
--
opt < /home/jay/svn/llvm-project/llvm/trunk/test/Transforms/ScalarRepl/vector_promote.ll
-scalarrepl -S | FileCheck
/home/jay/svn/llvm-project/llvm/trunk/test/Transforms/ScalarRepl/vector_promote.ll
--
Exit Code: 1
Command Output (stderr):
--
/home/jay/svn/llvm-project/llvm/trunk/test/Transforms/ScalarRepl/vector_promote.ll:99:15:
error: expected string not found in input
; CHECK-NEXT: bitcast <2 x float> %X to i64
              ^
<stdin>:48:18: note: scanning from here
define i64 @test6(<2 x float> %X) {
                 ^
<stdin>:49:9: note: possible intended match here
 %tmp = bitcast <2 x float> %X to <1 x i64> ; <<1 x i64>> [#uses=1]
        ^
--


Is this the "disaster" of which Daniel spake?

Is there an easy way to run some regression tests on the GCC front end?

Thanks,
Jay.