<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;">Looks like both the ppc-64 and mips bots are still failing after I reverted this change.<div><br></div><div>Since:</div><div><br></div><div>1. The mips bot always performs a clean build,</div><div>2. The ppc64-bot failed again after a clean build (I asked the ppc-64 maintainers to clean the bot which they did... Thanks Will!),</div><div><br></div><div>I think it is safe to assume that this change was not the cause of the failures that said builders were seeing. Thus I am going to recommit this patch.</div><div><br></div><div>Michael</div><div><br><div><div>On Jun 27, 2013, at 1:40 PM, Michael Gottesman <<a href="mailto:mgottesman@apple.com">mgottesman@apple.com</a>> wrote:</div><br class="Apple-interchange-newline"><blockquote type="cite"><div style="letter-spacing: normal; orphans: auto; text-align: start; text-indent: 0px; text-transform: none; white-space: normal; widows: auto; word-spacing: 0px; -webkit-text-stroke-width: 0px;">Author: mgottesman<br>Date: Thu Jun 27 15:40:11 2013<br>New Revision: 185099<br><br>URL:<span class="Apple-converted-space"> </span><a href="http://llvm.org/viewvc/llvm-project?rev=185099&view=rev">http://llvm.org/viewvc/llvm-project?rev=185099&view=rev</a><br>Log:<br>Revert "[APFloat] Removed APFloat constructor which initialized to either zero/NaN but allowed you to arbitrarily set the category of the float."<br><br>This reverts commit r185095. This is causing a FileCheck failure on<br>the 3dnow intrinsics on at least the mips/ppc bots but not on the x86<br>bots.<br><br>Reverting while I figure out what is going on.<br><br>Modified:<br> llvm/trunk/include/llvm/ADT/APFloat.h<br> llvm/trunk/lib/Support/APFloat.cpp<br> llvm/trunk/lib/Transforms/InstCombine/InstCombineCompares.cpp<br><br>Modified: llvm/trunk/include/llvm/ADT/APFloat.h<br>URL:<span class="Apple-converted-space"> </span><a href="http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/APFloat.h?rev=185099&r1=185098&r2=185099&view=diff">http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/APFloat.h?rev=185099&r1=185098&r2=185099&view=diff</a><br>==============================================================================<br>--- llvm/trunk/include/llvm/ADT/APFloat.h (original)<br>+++ llvm/trunk/include/llvm/ADT/APFloat.h Thu Jun 27 15:40:11 2013<br>@@ -191,6 +191,7 @@ public:<br> APFloat(const fltSemantics &); // Default construct to 0.0<br> APFloat(const fltSemantics &, StringRef);<br> APFloat(const fltSemantics &, integerPart);<br>+ APFloat(const fltSemantics &, fltCategory, bool negative);<br> APFloat(const fltSemantics &, uninitializedTag);<br> APFloat(const fltSemantics &, const APInt &);<br> explicit APFloat(double d);<br><br>Modified: llvm/trunk/lib/Support/APFloat.cpp<br>URL:<span class="Apple-converted-space"> </span><a href="http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/APFloat.cpp?rev=185099&r1=185098&r2=185099&view=diff">http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/APFloat.cpp?rev=185099&r1=185098&r2=185099&view=diff</a><br>==============================================================================<br>--- llvm/trunk/lib/Support/APFloat.cpp (original)<br>+++ llvm/trunk/lib/Support/APFloat.cpp Thu Jun 27 15:40:11 2013<br>@@ -795,6 +795,17 @@ APFloat::APFloat(const fltSemantics &our<br> initialize(&ourSemantics);<br>}<br><br>+APFloat::APFloat(const fltSemantics &ourSemantics,<br>+ fltCategory ourCategory, bool negative) {<br>+ initialize(&ourSemantics);<br>+ category = ourCategory;<br>+ sign = negative;<br>+ if (isFiniteNonZero())<br>+ category = fcZero;<br>+ else if (ourCategory == fcNaN)<br>+ makeNaN();<br>+}<br>+<br>APFloat::APFloat(const fltSemantics &ourSemantics, StringRef text) {<br> initialize(&ourSemantics);<br> convertFromString(text, rmNearestTiesToEven);<br>@@ -2395,8 +2406,8 @@ APFloat::roundSignificandWithExponent(co<br> excessPrecision = calcSemantics.precision - semantics->precision;<br> truncatedBits = excessPrecision;<br><br>- APFloat decSig = APFloat::getZero(calcSemantics, sign);<br>- APFloat pow5(calcSemantics);<br>+ APFloat decSig(calcSemantics, fcZero, sign);<br>+ APFloat pow5(calcSemantics, fcZero, false);<br><br> sigStatus = decSig.convertFromUnsignedParts(decSigParts, sigPartCount,<br> rmNearestTiesToEven);<br>@@ -3377,16 +3388,15 @@ APFloat APFloat::getSmallest(const fltSe<br>}<br><br>APFloat APFloat::getSmallestNormalized(const fltSemantics &Sem, bool Negative) {<br>- APFloat Val(Sem, uninitialized);<br>+ APFloat Val(Sem, fcNormal, Negative);<br><br> // We want (in interchange format):<br> // sign = {Negative}<br> // exponent = 0..0<br> // significand = 10..0<br><br>- Val.zeroSignificand();<br>- Val.sign = Negative;<br> Val.exponent = Sem.minExponent;<br>+ Val.zeroSignificand();<br> Val.significandParts()[partCountForBits(Sem.precision)-1] |=<br> (((integerPart) 1) << ((Sem.precision - 1) % integerPartWidth));<br><br><br>Modified: llvm/trunk/lib/Transforms/InstCombine/InstCombineCompares.cpp<br>URL:<span class="Apple-converted-space"> </span><a href="http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/InstCombine/InstCombineCompares.cpp?rev=185099&r1=185098&r2=185099&view=diff">http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/InstCombine/InstCombineCompares.cpp?rev=185099&r1=185098&r2=185099&view=diff</a><br>==============================================================================<br>--- llvm/trunk/lib/Transforms/InstCombine/InstCombineCompares.cpp (original)<br>+++ llvm/trunk/lib/Transforms/InstCombine/InstCombineCompares.cpp Thu Jun 27 15:40:11 2013<br>@@ -2888,7 +2888,7 @@ Instruction *InstCombiner::FoldFCmp_IntT<br> if (!LHSUnsigned) {<br> // If the RHS value is > SignedMax, fold the comparison. This handles +INF<br> // and large values.<br>- APFloat SMax(RHS.getSemantics());<br>+ APFloat SMax(RHS.getSemantics(), APFloat::fcZero, false);<br> SMax.convertFromAPInt(APInt::getSignedMaxValue(IntWidth), true,<br> APFloat::rmNearestTiesToEven);<br> if (SMax.compare(RHS) == APFloat::cmpLessThan) { // smax < 13123.0<br>@@ -2900,7 +2900,7 @@ Instruction *InstCombiner::FoldFCmp_IntT<br> } else {<br> // If the RHS value is > UnsignedMax, fold the comparison. This handles<br> // +INF and large values.<br>- APFloat UMax(RHS.getSemantics());<br>+ APFloat UMax(RHS.getSemantics(), APFloat::fcZero, false);<br> UMax.convertFromAPInt(APInt::getMaxValue(IntWidth), false,<br> APFloat::rmNearestTiesToEven);<br> if (UMax.compare(RHS) == APFloat::cmpLessThan) { // umax < 13123.0<br>@@ -2913,7 +2913,7 @@ Instruction *InstCombiner::FoldFCmp_IntT<br><br> if (!LHSUnsigned) {<br> // See if the RHS value is < SignedMin.<br>- APFloat SMin(RHS.getSemantics());<br>+ APFloat SMin(RHS.getSemantics(), APFloat::fcZero, false);<br> SMin.convertFromAPInt(APInt::getSignedMinValue(IntWidth), true,<br> APFloat::rmNearestTiesToEven);<br> if (SMin.compare(RHS) == APFloat::cmpGreaterThan) { // smin > 12312.0<br>@@ -2924,7 +2924,7 @@ Instruction *InstCombiner::FoldFCmp_IntT<br> }<br> } else {<br> // See if the RHS value is < UnsignedMin.<br>- APFloat SMin(RHS.getSemantics());<br>+ APFloat SMin(RHS.getSemantics(), APFloat::fcZero, false);<br> SMin.convertFromAPInt(APInt::getMinValue(IntWidth), true,<br> APFloat::rmNearestTiesToEven);<br> if (SMin.compare(RHS) == APFloat::cmpGreaterThan) { // umin > 12312.0<br><br><br>_______________________________________________<br>llvm-commits mailing list<br><a href="mailto:llvm-commits@cs.uiuc.edu">llvm-commits@cs.uiuc.edu</a><br><a href="http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits">http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits</a></div></blockquote></div><br></div></body></html>