<html><head><meta http-equiv="Content-Type" content="text/html charset=us-ascii"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space;" class=""><br class=""><div><blockquote type="cite" class=""><div class="">On Mar 17, 2017, at 9:12 AM, David Blaikie <<a href="mailto:dblaikie@gmail.com" class="">dblaikie@gmail.com</a>> wrote:</div><br class="Apple-interchange-newline"><div class=""><div dir="ltr" class="">Hmm, one wrinkle... <br class=""><br class="">the "if (auto Fragment ...)" case doesn't run in any existing test cases. (I put an "assert(false)" there & ran check-llvm).<br class=""><br class="">Any ideas if that's still reachable? Perhaps this only happens for direct bool parameters, never fragments.</div></div></blockquote><div><br class=""></div><div>I just finished building the swift stdlib in 64 and 32 bit (which is typically a good source of these) and also didn't hit the assertion. We'd need to hand-craft a testcase for opt -mem2reg.</div><div><br class=""></div><div>-- adrian</div><br class=""><blockquote type="cite" class=""><div class=""><br class=""><div class="gmail_quote"><div dir="ltr" class="">On Fri, Mar 17, 2017 at 8:28 AM Adrian Prantl via Phabricator <<a href="mailto:reviews@reviews.llvm.org" class="">reviews@reviews.llvm.org</a>> wrote:<br class=""></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">aprantl accepted this revision.<br class="gmail_msg">
aprantl added a comment.<br class="gmail_msg">
This revision is now accepted and ready to land.<br class="gmail_msg">
<br class="gmail_msg">
Took me a while to find the functional change amidst the refactoring :-)<br class="gmail_msg">
As discussed in the PR, this LGTM, thanks!<br class="gmail_msg">
One inline comment about a comment.<br class="gmail_msg">
<br class="gmail_msg">
<br class="gmail_msg">
<br class="gmail_msg">
================<br class="gmail_msg">
Comment at: lib/Transforms/Utils/Local.cpp:1118<br class="gmail_msg">
// describing will always be smaller than the variable size, because<br class="gmail_msg">
// VariableSize == Size of Alloca described by DDI. Since SI stores<br class="gmail_msg">
// to the alloca described by DDI, if it's first operand is an extend,<br class="gmail_msg">
----------------<br class="gmail_msg">
Does this comment still match our understanding?<br class="gmail_msg">
Particularly the `The fragment we're describing will always be smaller than the variable size` bit.<br class="gmail_msg">
<br class="gmail_msg">
<br class="gmail_msg">
<a href="https://reviews.llvm.org/D31062" rel="noreferrer" class="gmail_msg" target="_blank">https://reviews.llvm.org/D31062</a><br class="gmail_msg">
<br class="gmail_msg">
<br class="gmail_msg">
<br class="gmail_msg">
</blockquote></div>
</div></blockquote></div><br class=""></body></html>