gmp-4.2.1 on cygwin

[Sisyphus]

Hi,

I built (without any hassle, iirc) gmp-4.2.1 on cygwin on my old 32-bit 
pentium-3 (Windows 2000) box. To build GMP on that box I ran 
'./configure --disable-shared --enable-static', followed by 'make', 'make 
check' (which passed all tests), and 'make install'.

I have just acquired a 64-bit AthlonX2 (4600) box, running Windows XP 
(32-bit). This new box also contains the exact same version of cygwin as was 
installed on the old P-III. Again I started with 
'./configure --disable-shared --enable-static' .... but this time, when 
running 'make', I get the errors:

 gcc -DHAVE_CONFIG_H -I. -I. -I.. -D__GMP_WITHIN_GMP -I.. -DOPERATION_divis  
-m32 -O2 -fomit-frame-pointer -mtune=k8 -march=k8 -c divis.c -o divis.o
divis.c: In function `__gmpn_divisible_p':
divis.c:76: error: `TMP_DECL' undeclared (first use in this function)
divis.c:76: error: (Each undeclared identifier is reported only once
divis.c:76: error: for each function it appears in.)
divis.c:141: error: `TMP_MARK' undeclared (first use in this function)
divis.c:154: error: `TMP_FREE' undeclared (first use in this function)
make[2]: *** [divis.lo] Error 1
make[2]: Leaving directory `/cygdrive/c/downloads/cygwin/gmp-4.2.1/mpn'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/cygdrive/c/downloads/cygwin/gmp-4.2.1'
make: *** [all] Error 2

I'm a little nonplussed by this (which is a common experience for me wrt 
cygwin).

Is this something that warrants a bug report ? Or have I most likely done 
something stupid ?

Cheers,
Rob