<html><head><meta http-equiv="Content-Type" content="text/html; charset=us-ascii"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; line-break: after-white-space;" class=""><div class="">Adding to the discussion:</div><div class=""><br class=""></div>A classic gotcha is the fact that llc/opt target LLVM_DEFAULT_TARGET_TRIPLE by default (which is different for some bots).<div class=""><br class=""></div><div class="">Even using "llc -march" / "opt -march" is not always enough as it only changes the architecture but not the operating system/environment (actually a lot of tests get this wrong, but it usually doesn't make them fail).</div><div class=""><br class=""></div><div class="">The correct way do things is to always provide</div><div class="">-mtriple=...</div><div class="">when invoking llc or opt with any transformations that use TargetTransformInfo.</div><div class=""><br class=""></div><div class="">- Matthias<br class=""><div class=""><div class=""><div><br class=""><blockquote type="cite" class=""><div class="">On Jan 5, 2018, at 3:08 PM, Adrian McCarthy via llvm-dev <<a href="mailto:llvm-dev@lists.llvm.org" class="">llvm-dev@lists.llvm.org</a>> wrote:</div><br class="Apple-interchange-newline"><div class=""><div dir="ltr" class="">Thanks!  It looks like this is sorted out now.</div><div class="gmail_extra"><br class=""><div class="gmail_quote">On Thu, Jan 4, 2018 at 5:07 PM, Friedman, Eli <span dir="ltr" class=""><<a href="mailto:efriedma@codeaurora.org" target="_blank" class="">efriedma@codeaurora.org</a>></span> wrote:<br class=""><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
  
    
  
  <div text="#000000" bgcolor="#FFFFFF" class=""><span class="">
    <div class="m_2587386726027582274moz-cite-prefix">On 1/4/2018 4:51 PM, Adrian McCarthy
      via llvm-dev wrote:<br class="">
    </div>
    <blockquote type="cite" class="">
      <div dir="ltr" class="">I tried to land a small fix before going on
        vacation at the end of the year, but I had to revert it because
        it broke on a few of the build bots, specifically
        clang-cmake-thumbv7-a15, clang-cmake-armv7-a15, and
        llvm-hexagon-elf.
        <div class=""><br class="">
        </div>
        <div class=""><a href="https://reviews.llvm.org/D41264" target="_blank" class="">https://reviews.llvm.org/<wbr class="">D41264</a><br class="">
        </div>
      </div>
    </blockquote>
    <br class=""></span>
    I think you meant to put the test into test/DebugInfo/X86/ ?  Your
    test specifies an x86 target, but those buildbots don't build the
    x86 target.<span class=""><br class="">
    <br class="">
    <blockquote type="cite" class="">
      <div dir="ltr" class="">
        <div class=""><br class="">
        </div>
        <div class="">It seemed to work as expected on the others.</div>
        <div class=""><br class="">
        </div>
        <div class="">Unfortunately, the build logs don't give many clues.  The
          fix included a new test, and that test failed with an exit
          code of 1 but with no messages.  For example:</div>
        <div class="">
          <pre class=""></pre>
          <pre style="text-align:start;text-indent:0px;text-decoration-style:initial;text-decoration-color:initial" class=""><span class="m_2587386726027582274gmail-stdout"><font style="font-family: "Courier New", courier, monotype, monospace; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; text-transform: none; word-spacing: 0px;" size="1" class=""><pre style="font-family: "Courier New", courier, monotype, monospace; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; text-align: start; text-indent: 0px; text-transform: none; word-spacing: 0px;" class=""><span class="m_2587386726027582274gmail-stdout" style="font-family: "Courier New", courier, monotype, monospace;">FAIL: LLVM :: DebugInfo/void-typedef.ll (14416 of 23113)
******************** TEST 'LLVM :: DebugInfo/void-typedef.ll' FAILED ********************
Script:
--
/local/buildbot/slaves/<wbr class="">hexagon-build-02/llvm-hexagon-<wbr class="">elf/llvm.obj/bin/llc /local/buildbot/slaves/<wbr class="">hexagon-build-02/llvm-hexagon-<wbr class="">elf/llvm.src/test/DebugInfo/<wbr class="">void-typedef.ll -o - 2>&1 | /local/buildbot/slaves/<wbr class="">hexagon-build-02/llvm-hexagon-<wbr class="">elf/llvm.obj/bin/FileCheck /local/buildbot/slaves/<wbr class="">hexagon-build-02/llvm-hexagon-<wbr class="">elf/llvm.src/test/DebugInfo/<wbr class="">void-typedef.ll
--
Exit Code: 1


********************</span></pre></font></span></pre>
        </div>
      </div>
    </blockquote>
    <br class=""></span>
    llc is probably producing some error message like "error: unable to
    get target for 'i686', see --version and --triple.", but you're
    redirecting stderr, so it isn't visible.  You should try to avoid
    this when possible.<br class="">
    <br class="">
    Your test doesn't have any positive CHECK lines, so FileCheck
    passes.  You should generally try to include some positive CHECK
    line in regression tests.  And you never need to check for the
    string "Assertion failed"; the test will fail if any command on the
    RUN line fails or crashes.<span class="HOEnZb"><font color="#888888" class=""><br class="">
    <br class="">
    -Eli<br class="">
    <pre class="m_2587386726027582274moz-signature" cols="72">-- 
Employee of Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project</pre>
  </font></span></div>

</blockquote></div><br class=""></div>
_______________________________________________<br class="">LLVM Developers mailing list<br class=""><a href="mailto:llvm-dev@lists.llvm.org" class="">llvm-dev@lists.llvm.org</a><br class="">http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev<br class=""></div></blockquote></div><br class=""></div></div></div></body></html>