<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<div class="moz-cite-prefix">I will take care of this right. I was
already in the process of doing as you suggest.<br>
Some code I wanted as a record in the source control so I can see
where I need to fix things.<br>
<br>
Reed<br>
<br>
On 11/04/2013 02:33 PM, Eric Christopher wrote:<br>
</div>
<blockquote
cite="mid:CALehDX4cVCZjm9nbriLTppVS=tR3W9eRFfOU4hOTS8LLrDwjaQ@mail.gmail.com"
type="cite">
<meta http-equiv="Content-Type" content="text/html;
charset=ISO-8859-1">
<div dir="ltr">
<div class="gmail_extra">
<div class="gmail_quote">
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex"><br>
+ STI(&TM.getSubtarget<MipsSubtarget>()),
MF(0), MCP(0), PrescannedForConstants(false){}<br>
<br>
</blockquote>
<div><br>
</div>
<div>80-col.</div>
<div><br>
</div>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex"><br>
+#ifdef IN_PROGRESS<br>
</blockquote>
<div><br>
</div>
<div>No. Please remove this sort of code.</div>
<div><br>
</div>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex"><br>
+#ifdef IN_PROGRESS<br>
+ case ARM::t2BR_JT:<br>
</blockquote>
<div><br>
</div>
<div>Especially not when you've just copied code out of the
ARM port.</div>
<div> </div>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
+<br>
+ }<br>
+<br>
+<br>
+ if (Opc == Mips::CONSTPOOL_ENTRY)<br>
+ continue;<br>
+<br>
+<br>
</blockquote>
<div><br>
</div>
<div>I'm a fan of vertical whitespace, but this is a bit
much.</div>
<div> </div>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex"><br>
+#ifdef IN_PROGRESS<br>
+ // Taking the address of a CP entry.<br>
+ case ARM::LEApcrel:<br>
</blockquote>
<div><br>
</div>
<div>Again.</div>
<div> </div>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
+<br>
+ // On Thumb, offsets==2 mod 4 are rounded down by the
hardware for<br>
+ // purposes of the displacement computation; compensate
for that here.<br>
+ // For unknown alignments, getMaxDisp() constrains the
range instead.<br>
+#ifdef IN_PROGRESS<br>
+ if (isThumb && U.KnownAlignment)<br>
+ UserOffset &= ~3u;<br>
+#endif<br>
</blockquote>
<div><br>
</div>
<div>Again. There are many more cases in the rest of the
code.</div>
<div> </div>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
+/// isCPEntryInRange - Returns true if the distance
between specific MI and<br>
+/// specific ConstPool entry instruction can fit in MI's
displacement field.<br>
+bool MipsConstantIslands::isCPEntryInRange(MachineInstr
*MI, unsigned UserOffset,<br>
+ MachineInstr
*CPEMI, unsigned MaxDisp,<br>
+ bool NegOk, bool
DoDump) {<br>
</blockquote>
<div><br>
</div>
<div>Why a DoDump variable?</div>
<div> </div>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
+ // Avoid splitting an IT block.<br>
</blockquote>
<div><br>
</div>
<div>You have IT blocks? This is why we don't just copy
code.</div>
<div><br>
</div>
<div>Again, please remove all of the non-mips code out of
this immediately.</div>
<div><br>
</div>
<div>-eric</div>
</div>
</div>
</div>
</blockquote>
<br>
</body>
</html>