<p dir="ltr">Good point.  Will do. </p>
<div class="gmail_quote">On Mar 18, 2015 7:54 AM, "Eli Bendersky" <<a href="mailto:eliben@google.com">eliben@google.com</a>> wrote:<br type="attribution"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">================<br>
Comment at: test/SemaCUDA/<a href="http://asm-constraints.cu:7" target="_blank">asm-constraints.cu:7</a><br>
@@ +6,3 @@<br>
+  short h;<br>
+  // asm with PTX constraints. Some of them are PTX-specific.<br>
+  __asm__("dont care" : "=h"(h): "f"(0.0), "d"(0.0), "h"(0), "r"(0), "l"(0));<br>
----------------<br>
Are there checks anywhere that asm diagnostics are validated within device?<br>
<br>
Perhaps a positive test would be helpful (make sure that an invalid constraint produces a diagnostic within a kernel)<br>
<br>
<a href="http://reviews.llvm.org/D8392" target="_blank">http://reviews.llvm.org/D8392</a><br>
<br>
EMAIL PREFERENCES<br>
  <a href="http://reviews.llvm.org/settings/panel/emailpreferences/" target="_blank">http://reviews.llvm.org/settings/panel/emailpreferences/</a><br>
<br>
<br>
</blockquote></div>