From: Matt Birkholz <puck@birchwood-abbey.net>
Date: Fri, 10 Oct 2014 23:39:22 +0000 (-0700)
Subject: i386 flonums: Fix substitution of FLDZ,FUCOMP for FTST in 521ffb3.
X-Git-Tag: mit-scheme-pucked-9.2.12~398^2~2
X-Git-Url: https://birchwood-abbey.net/git?a=commitdiff_plain;h=4ca9dbd244fa3d158a779a6ebf17e73a9b7f7e22;p=mit-scheme.git

i386 flonums: Fix substitution of FLDZ,FUCOMP for FTST in 521ffb3.

The branch instructions needed to be swapped.

This caused compiled generic arithmetic to get e.g. (> x 0.) wrong,
causing an overflow in float->integers when compiling
runtime/arith.scm for the SVM.
---

diff --git a/src/microcode/cmpauxmd/i386.m4 b/src/microcode/cmpauxmd/i386.m4
index ae640bfb9..bf538d6d3 100644
--- a/src/microcode/cmpauxmd/i386.m4
+++ b/src/microcode/cmpauxmd/i386.m4
@@ -1141,8 +1141,8 @@ asm_generic_$1_fail:
 IF387(`define_unary_operation(decrement,22,sub,fsubr)
 define_unary_operation(increment,26,add,fadd)
 
-define_unary_predicate(negative,2a,jl,jb)
-define_unary_predicate(positive,2c,jg,ja)
+define_unary_predicate(negative,2a,jl,ja)
+define_unary_predicate(positive,2c,jg,jb)
 define_unary_predicate(zero,2d,je,je)
 
 # define_binary_operation(name,index,fix*fix,flo*flo)