<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;"><br><div><div>On Dec 20, 2013, at 9:33 AM, Paweł Bylica <<a href="mailto:pawel.bylica@ibs.org.pl">pawel.bylica@ibs.org.pl</a>> wrote:</div><br class="Apple-interchange-newline"><blockquote type="cite"><div dir="ltr">Hello there,<div><br></div><div>I have a high level code which would look like that in C++:</div><div><br></div><div><div><font face="courier new, monospace">enum E { A, B, C };</font></div><div><font face="courier new, monospace"><br>
</font></div><div><font face="courier new, monospace">E int2e(long i) {</font></div><div><font face="courier new, monospace"> switch(i) {</font></div><div><font face="courier new, monospace"> case 0: return A;</font></div>
<div><font face="courier new, monospace"> case 1: return B;</font></div><div><font face="courier new, monospace"> case 2: return C;</font></div><div><font face="courier new, monospace"> default: return A;</font></div>
<div><font face="courier new, monospace"> }</font></div><div><font face="courier new, monospace">}</font></div></div><div><br></div><div>It is compiled to this IR with O3 optimization:</div><div><br></div><div><div><font face="courier new, monospace">define i64 @int2e(i64 %i_arg) #0 {</font></div>
<div><font face="courier new, monospace">entry:</font></div><div><font face="courier new, monospace"> switch i64 %i_arg, label %label_case1 [</font></div><div><font face="courier new, monospace"> i64 2, label %label_case3</font></div>
<div><font face="courier new, monospace"> i64 1, label %label_case2</font></div><div><font face="courier new, monospace"> ]</font></div><div><font face="courier new, monospace"><br></font></div><div><font face="courier new, monospace">label_case1: ; preds = %entry, %label_case3, %label_case2</font></div>
<div><font face="courier new, monospace"> %merge = phi i64 [ %i_arg, %label_case2 ], [ %i_arg, %label_case3 ], [ 0, %entry ]</font></div><div><font face="courier new, monospace"> ret i64 %merge</font></div><div><font face="courier new, monospace"><br>
</font></div><div><font face="courier new, monospace">label_case2: ; preds = %entry</font></div><div><font face="courier new, monospace"> br label %label_case1</font></div><div><font face="courier new, monospace"><br>
</font></div><div><font face="courier new, monospace">label_case3: ; preds = %entry</font></div><div><font face="courier new, monospace"> br label %label_case1</font></div><div><font face="courier new, monospace">}</font></div>
</div><div><br></div><div>In the result IR `phi` instruction has 3 branches, but two first of them returns the same value. Shouln't it be optimized?</div><div><br></div><div>- Paweł</div></div></blockquote><br></div><div>I ran into almost the same case a few months ago, and half fixed it but never finished the patch. I found that it does the right thing if you manually ran some combination of simplifycfg and instcombine on the unoptimized IR it would do the right thing, but the standard pass order doesn’t do it.</div><br></body></html>