<html><head><meta http-equiv="Content-Type" content="text/html charset=utf-8"></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 Jan 26, 2015, at 6:15 PM, Chandler Carruth <<a href="mailto:chandlerc@google.com" class="">chandlerc@google.com</a>> wrote:</div><br class="Apple-interchange-newline"><div class=""><div dir="ltr" class=""><div class="gmail_extra"><br class=""><div class="gmail_quote">On Mon, Jan 26, 2015 at 3:09 PM, Jim Grosbach <span dir="ltr" class=""><<a href="mailto:grosbach@apple.com" target="_blank" class="">grosbach@apple.com</a>></span> wrote:<br class=""><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">There should be exactly zero DAG combines that have to run, ever. They are by definition not required for correctness. Anything else is a bug, and a bad one.</blockquote></div><br class="">The X86 backend at least heavily uses target-specific DAG combines to "lower" DAG nodes that it doesn't have any actual lowering for.</div><div class="gmail_extra"><br class=""></div><div class="gmail_extra">Yes, it's a bug. But I don't even know how much work would be required to fix it.</div></div>
</div></blockquote></div><br class=""><div class="">Yeah, I certainly don’t claim our backends are bug free, especially with regards to things like this where we’re not actively testing for it. I would just like to be actively moving things forwards when we have the opportunity.</div></body></html>