soft-fp: Fix used without set warning in _FP_MUL and _FP_DIV

Seen in, e.g. ports/sysdeps/alpha/soft-fp/ots_mul.c.

The problem here is we have a switch on the class of X*Y, followed by
a switch on the class of R.  The exponent field of R really is not set
by the first switch for NaN outputs, because we know it's not going to
be used.  The compiler is not smart enough to see through this.

By pulling the exponent computation out of the NORMAL*NORMAL path, we
will not be pessimizing the most common case, while still providing an
arbitrary value to satisfy the compiler.
This commit is contained in:
Richard Henderson 2012-12-07 13:17:08 -06:00
parent 7514f09e1e
commit 573cd4843a
2 changed files with 8 additions and 2 deletions

View File

@ -1,3 +1,9 @@
2012-12-07 Richard Henderson <rth@redhat.com>
* soft-fp/op-common.h (_FP_MUL): Pull computation of R_e from the
normal/normal case to before the switch.
(_FP_DIV): Likewise.
2012-12-06 Carlos O'Donell <carlos@systemhalted.org>
Mike Frysinger <vapier@gentoo.org>

View File

@ -748,11 +748,11 @@ do { \
#define _FP_MUL(fs, wc, R, X, Y) \
do { \
R##_s = X##_s ^ Y##_s; \
R##_e = X##_e + Y##_e + 1; \
switch (_FP_CLS_COMBINE(X##_c, Y##_c)) \
{ \
case _FP_CLS_COMBINE(FP_CLS_NORMAL,FP_CLS_NORMAL): \
R##_c = FP_CLS_NORMAL; \
R##_e = X##_e + Y##_e + 1; \
\
_FP_MUL_MEAT_##fs(R,X,Y); \
\
@ -811,11 +811,11 @@ do { \
#define _FP_DIV(fs, wc, R, X, Y) \
do { \
R##_s = X##_s ^ Y##_s; \
R##_e = X##_e - Y##_e; \
switch (_FP_CLS_COMBINE(X##_c, Y##_c)) \
{ \
case _FP_CLS_COMBINE(FP_CLS_NORMAL,FP_CLS_NORMAL): \
R##_c = FP_CLS_NORMAL; \
R##_e = X##_e - Y##_e; \
\
_FP_DIV_MEAT_##fs(R,X,Y); \
break; \