<div class="gmail_extra"><div class="gmail_quote">On Mon, Aug 27, 2012 at 2:03 AM, Anitha Boyapati <span dir="ltr"><<a href="mailto:anitha.boyapati@gmail.com" target="_blank" class="cremed">anitha.boyapati@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><br><br><div class="gmail_quote"><div class="im">On 27 August 2012 12:39, Chandler Carruth <span dir="ltr"><<a href="mailto:chandlerc@google.com" target="_blank" class="cremed">chandlerc@google.com</a>></span> wrote:<br>
<blockquote style="margin:0px 0px 0px 0.8ex;padding-left:1ex;border-left-color:rgb(204,204,204);border-left-width:1px;border-left-style:solid" class="gmail_quote">
FYI, it seems like a build bot failed:<div><br></div><div><a href="http://bb.pgr.jp/builders/cmake-clang-i686-mingw32/builds/15" target="_blank" class="cremed">http://bb.pgr.jp/builders/cmake-clang-i686-mingw32/builds/15</a><br>
</div></blockquote>
<div> </div></div><div>Strange. The test fma_patterns.ll passes for me with rev 162655.</div><div> </div><div>build/test> clang -v<br>clang version 3.2 (trunk 162661) (llvm/trunk 162654:162655M)<br>Target: x86_64-unknown-linux-gnu<br>
Thread model: posix<br></div><div>build/test> llvm-lit CodeGen/X86/fma_patterns.ll<br>-- Testing: 1 tests, 1 threads --<br>PASS: LLVM :: CodeGen/X86/fma_patterns.ll (1 of 1)<br>Testing Time: 0.08s<br> Expected Passes : 1<br>
</div><div> </div><div>Moreover, for that particular function "def _test_x86_fnmadd_ss" ( reported as failed), I could see that the generated assembly is same as what is expected in fma_patterns.ll</div><div> </div>
<div>>>>>>>>>></div><div>build/test> llc -mcpu=bdver2 -mattr=-fma4 -fp-contract=fast test_x86_fnmadd.ll<br>build/test> more test_x86_fnmadd.s<br> .file "test_x86_fnmadd.ll"<br>
.text<br> .globl test_x86_fnmadd_ss<br> .align 16, 0x90<br> .type test_x86_fnmadd_ss,@function<br>test_x86_fnmadd_ss: # @test_x86_fnmadd_ss<br> .cfi_startproc<br>
# BB#0:<br> vfnmadd213ss %xmm2, %xmm0, %xmm1<br> vmovaps %xmm1, %xmm0<br> ret<br>.Ltmp0:<br> .size test_x86_fnmadd_ss, .Ltmp0-test_x86_fnmadd_ss<br> .cfi_endproc<br></div><div><<<<<<</div>
<div> </div><div>Looks like given report has memory form of fma. Am I missing something here? Can mingw32 make a difference?</div></div></blockquote><div><br></div><div>Perhaps calling convention? I don't really know anything about mingw32, but it certainly seems to be repeatably failing.</div>
<div><br></div><div>The RUN line immediately before yours has an explicit triple in it, you might also need one. </div></div></div>