https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122275
Bug ID: 122275
Summary: `DFP_SET_ROUNDMODE` etc. pass wrong operand size on
i686-w64-mingw32
Product: gcc
Version: 16.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: libgcc
Assignee: unassigned at gcc dot gnu.org
Reporter: lh_mouse at 126 dot com
Target Milestone: ---
The x87 control word is `unsigned short` but DFP macros pass `unsigned int`:
```
#define DFP_SET_ROUNDMODE(round) \
do \
{ \
unsigned int _cw; \
__asm__ __volatile__ ("fnstcw\t%0" : "=m" (_cw)); \
_cw &= ~FP_RND_MASK; \
_cw |= round; \
__asm__ __volatile__ ("fldcw\t%0" :: "m" (_cw)); \
```
This is seen when building libgcc on i686-w64-mingw32 with `-masm=intel`:
```
make[3]: Entering directory
'/home/lh_mouse/GitHub/MINGW-packages/mingw-w64-gcc/src/build-MINGW32/i686-w64-mingw32/libgcc'
/home/lh_mouse/GitHub/MINGW-packages/mingw-w64-gcc/src/build-MINGW32/./gcc/xgcc
-B/home/lh_mouse/GitHub/MINGW-packages/mingw-w64-gcc/src/build-MINGW32/./gcc/
-L/mingw32/i686-w64-mingw32/lib -L/mingw32/lib -isystem
/mingw32/i686-w64-mingw32/include -isystem /mingw32/include
-B/mingw32/i686-w64-mingw32/bin/ -B/mingw32/i686-w64-mingw32/lib/ -isystem
/mingw32/i686-w64-mingw32/include -isystem
/mingw32/i686-w64-mingw32/sys-include -fno-checking -g -march=pentium4
-mtune=generic -masm=intel -O2 -pipe -Wp,-D_FORTIFY_SOURCE=2
-fstack-protector-strong -Wp,-D__USE_MINGW_ANSI_STDIO=1 -O2
-I../../../gcc/libgcc/../winsup/w32api/include -g -march=pentium4
-mtune=generic -O2 -pipe -Wp,-D_FORTIFY_SOURCE=2 -fstack-protector-strong
-Wp,-D__USE_MINGW_ANSI_STDIO=1 -DIN_GCC -W -Wall -Wno-error=narrowing
-Wwrite-strings -Wcast-qual -Wno-format -Wstrict-prototypes
-Wmissing-prototypes -Wold-style-definition -isystem ./include -g
-DIN_LIBGCC2 -fbuilding-libgcc -fno-stack-protector -I. -I. -I../.././gcc
-I../../../gcc/libgcc -I../../../gcc/libgcc/. -I../../../gcc/libgcc/../gcc
-I../../../gcc/libgcc/../include -I../../../gcc/libgcc/config/i386
-I../../../gcc/libgcc/config/libbid -DENABLE_DECIMAL_BID_FORMAT -DHAVE_CC_TLS
-o bid128_div.o -MT bid128_div.o -MD -MP -MF bid128_div.dep -c
../../../gcc/libgcc/config/libbid/bid128_div.c
{standard input}: Assembler messages:
{standard input}:7137: Error: operand size mismatch for `fnstcw'
{standard input}:7153: Error: operand size mismatch for `fnstcw'
{standard input}:7162: Error: operand size mismatch for `fldcw'
{standard input}:7375: Error: operand size mismatch for `fnstcw'
{standard input}:7387: Error: operand size mismatch for `fldcw'
{standard input}:7704: Error: operand size mismatch for `fnstcw'
{standard input}:7716: Error: operand size mismatch for `fldcw'
{standard input}:9210: Error: operand size mismatch for `fnstcw'
{standard input}:9222: Error: operand size mismatch for `fldcw'
{standard input}:9584: Error: operand size mismatch for `fnstcw'
{standard input}:9596: Error: operand size mismatch for `fldcw'
{standard input}:9684: Error: operand size mismatch for `fnstcw'
{standard input}:9696: Error: operand size mismatch for `fldcw'
{standard input}:9759: Error: operand size mismatch for `fnstcw'
{standard input}:9771: Error: operand size mismatch for `fldcw'
{standard input}:11142: Error: operand size mismatch for `fnstcw'
```
