projects / mit-scheme.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: a961596) | patch
i386 flonums: Fix substitution of FLDZ,FUCOMP for FTST in 521ffb3.
authorMatt Birkholz <puck@birchwood-abbey.net>
Fri, 10 Oct 2014 23:39:22 +0000 (16:39 -0700)
committerMatt Birkholz <puck@birchwood-abbey.net>
Sat, 11 Oct 2014 20:27:21 +0000 (13:27 -0700)
commita7e78243edfa7dd632b3bbc6bd25525309b7476f
treeacb3738b9a2cc07f3a9d3efdf312de8738b0e792tree | snapshot
parenta96159685cfd488544c24a011464c6ba2afbe55ccommit | diff
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,
thus float->integers to overflow when LIAR/SVM compiled runtime/
arith.scm.
src/microcode/cmpauxmd/i386.m4 diff | blob | history
MIT/GNU Scheme, Pucked