<html><head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
  </head>
  <body>
    <p><br>
    </p>
    <div class="moz-cite-prefix">On 7/15/20 3:33 AM, Serge Guelton via
      llvm-dev wrote:<br>
    </div>
    <blockquote type="cite" cite="mid:CALqhe3-Kior=1pFhkgyGQq1VwzW+aohzSHWcucryTrxhJRqnsw@mail.gmail.com">
      
      <div dir="ltr">Hi folks,
        <div><br>
        </div>
        <div>some more information on this feature -  as a reminder I
          started one month ago to work on an expensive check that would
          verify that pass return status is correctly reported by
          passes, i.e. no pass return « IR not modified » while actually
          modifying it.</div>
        <div>It took ~20 pass fixes to achieve that goal, as many passes
          were not respectful of that contract, but as
          of 3667d87a33d3c8d4072a41fd84bb880c59347dc0,  <a href="https://reviews.llvm.org/D80916" rel="noreferrer" target="_blank" moz-do-not-send="true">https://reviews.<span class="gmail-il">llvm</span>.org/D80916</a> has been
          merged in master and the check is active, which should prevent
          further regression on that topic.</div>
        <div><br>
        </div>
        <div>Thanks a lot to @foad, @jdoerfert, @fhahn, @calixte (and
          others I'm sorry to forgot) for their help during the reviews.</div>
      </div>
    </blockquote>
    <p><br>
    </p>
    <p>This is great news!</p>
    <p>Some years ago, we did some experiments on whether we could
      develop more fixed-point optimization within LLVM's pipeline. This
      was not the only impediment we identified, but it was a major one.</p>
    <p>Out of curiosity, does change here include changes to names, and
      other semantically-irrelevant changes (e.g., changing the order of
      operands in a PHI)?</p>
    <p> -Hal<br>
    </p>
    <p><br>
    </p>
    <blockquote type="cite" cite="mid:CALqhe3-Kior=1pFhkgyGQq1VwzW+aohzSHWcucryTrxhJRqnsw@mail.gmail.com">
      <div dir="ltr">
        <table class="gmail-phui-status-list-view">
          <tbody>
            <tr class="gmail-">
              <td class="gmail-phui-status-item-target"><br>
              </td>
              <td class="gmail-phui-status-item-note"><br>
              </td>
            </tr>
          </tbody>
        </table>
      </div>
      <br>
      <div class="gmail_quote">
        <div dir="ltr" class="gmail_attr">On Fri, Jun 12, 2020 at 11:24
          PM Mehdi AMINI <<a href="mailto:joker.eph@gmail.com" moz-do-not-send="true">joker.eph@gmail.com</a>> wrote:<br>
        </div>
        <blockquote class="gmail_quote" style="margin:0px 0px 0px
          0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
          <div dir="ltr">
            <div dir="ltr"><br>
            </div>
            <br>
            <div class="gmail_quote">
              <div dir="ltr" class="gmail_attr">On Thu, Jun 11, 2020 at
                8:42 AM Serge Guelton via llvm-dev <<a href="mailto:llvm-dev@lists.llvm.org" target="_blank" moz-do-not-send="true">llvm-dev@lists.llvm.org</a>>
                wrote:<br>
              </div>
              <blockquote class="gmail_quote" style="margin:0px 0px 0px
                0.8ex;border-left:1px solid
                rgb(204,204,204);padding-left:1ex">Hi folks,<br>
                <br>
                Per the documentation[0], whenever an LLVM pass doesn't
                modify the IR it's run on, it<br>
                should return `false`--it's okay to return `true` if no
                change happen, just less<br>
                optimal. In the New PM area, this is generally
                translated into a `PreservedAnalyses::all()`.<br>
                <br>
                <a href="https://reviews.llvm.org/D80916" rel="noreferrer" target="_blank" moz-do-not-send="true">https://reviews.llvm.org/D80916</a>
                provides an `EXPENSIVE_CHECK` that computes a<br>
                hash of the IR before and after the pass, and checks
                that any change is<br>
                correctly reported. The hash is currently incomplete (on
                purpose, let's start<br>
                small), but it turns out a dozen of passes do not
                satisfy that<br>
                requirement.<br>
                <br>
                This could lead to various category of bugs (dangling
                references, inconsistent<br>
                state, etc). This affects both New and Legacy PM, as
                passes tend to wrap functions<br>
                that report their status.<br>
                <br>
                I wrote a bunch of patches for all failure detected by
                this check, as I cannot land the<br>
                check now, it would break the buildbots :-) Any help to
                review the remaining<br>
                ones [1] is appreciated.<br>
                <br>
                Once that check lands and we're relatively confident on
                the quality of the<br>
                return status, some more optimizations could be
                triggered, like<br>
                <a href="https://reviews.llvm.org/D80707" rel="noreferrer" target="_blank" moz-do-not-send="true">https://reviews.llvm.org/D80707</a>.<br>
              </blockquote>
              <div><br>
              </div>
              <div>Awesome feature! I am really fond of these pieces of
                infrastructure that can defend against human mistakes
                and save countless hours of debugging when subtle issues
                arise.</div>
              <div><br>
              </div>
              <div>Thanks Serge,</div>
              <div><br>
              </div>
              <div>-- </div>
              <div>Mehdi</div>
              <div><br>
              </div>
              <div> </div>
              <blockquote class="gmail_quote" style="margin:0px 0px 0px
                0.8ex;border-left:1px solid
                rgb(204,204,204);padding-left:1ex">
                <br>
                <br>
                [0] <a href="https://llvm.org/docs/WritingAnLLVMPass.html#the-runonmodule-method" rel="noreferrer" target="_blank" moz-do-not-send="true">https://llvm.org/docs/WritingAnLLVMPass.html#the-runonmodule-method</a><br>
                [1] <a href="https://reviews.llvm.org/D81230" rel="noreferrer" target="_blank" moz-do-not-send="true">https://reviews.llvm.org/D81230</a><br>
                    <a href="https://reviews.llvm.org/D81236" rel="noreferrer" target="_blank" moz-do-not-send="true">https://reviews.llvm.org/D81236</a><br>
                    <a href="https://reviews.llvm.org/D81256" rel="noreferrer" target="_blank" moz-do-not-send="true">https://reviews.llvm.org/D81256</a><br>
                    <a href="https://reviews.llvm.org/D81238" rel="noreferrer" target="_blank" moz-do-not-send="true">https://reviews.llvm.org/D81238</a><br>
                    <a href="https://reviews.llvm.org/D81225" rel="noreferrer" target="_blank" moz-do-not-send="true">https://reviews.llvm.org/D81225</a><br>
                <br>
                _______________________________________________<br>
                LLVM Developers mailing list<br>
                <a href="mailto:llvm-dev@lists.llvm.org" target="_blank" moz-do-not-send="true">llvm-dev@lists.llvm.org</a><br>
                <a href="https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev" rel="noreferrer" target="_blank" moz-do-not-send="true">https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev</a><br>
              </blockquote>
            </div>
          </div>
        </blockquote>
      </div>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <pre class="moz-quote-pre" wrap="">_______________________________________________
LLVM Developers mailing list
<a class="moz-txt-link-abbreviated" href="mailto:llvm-dev@lists.llvm.org">llvm-dev@lists.llvm.org</a>
<a class="moz-txt-link-freetext" href="https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev">https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev</a>
</pre>
    </blockquote>
    <pre class="moz-signature" cols="72">-- 
Hal Finkel
Lead, Compiler Technology and Programming Languages
Leadership Computing Facility
Argonne National Laboratory</pre>
  </body>
</html>