test-float failure on m68k

[David Huggins-Daines <dhd at eradicator dot org>]

Hi,

I get a few failures in test-float when building libc 2.1.2 (possibly
somewhat patched - this is the Debian 2.1.2-10 package) for Linux/m68k.

I've attached the test result.

Cheers

-- 
< james>  overfiend: I am not an autobuilder, I am a human being
testing float (without inline functions)
Fail: yn (10, 0.7) = -4.244...*10^9
Result:
 is:         -4.24471731200000000000e+09  -0x1.fa027e00000000000000p+31
 should be:  -4.24471936000000000000e+09  -0x1.fa028e00000000000000p+31
 difference:  2.04800000000000000000e+03   0x1.00000000000000000000p+11
Fail: yn (10, 1.0) = -1.216...*10^8
Result:
 is:         -1.21617984000000000000e+08  -0x1.cfef9000000000000000p+26
 should be:  -1.21618016000000000000e+08  -0x1.cfef9800000000000000p+26
 difference:  3.20000000000000000000e+01   0x1.00000000000000000000p+5
Fail: yn (10, 2.0) = -129184.5...
Result:
 is:         -1.29184484375000000000e+05  -0x1.f8a07c00000000000000p+16
 should be:  -1.29184539062500000000e+05  -0x1.f8a08a00000000000000p+16
 difference:  5.46875000000000000000e-02   0x1.c0000000000000000000p-5
Fail: yn (10, 10.0) = -0.35981...
Result:
 is:         -3.59814077615737915039e-01  -0x1.70731a00000000000000p-2
 should be:  -3.59814167022705078125e-01  -0x1.70732000000000000000p-2
 difference:  8.94069671630859375000e-08   0x1.80000000000000000000p-24

Test suite completed:
  2278 test cases plus 2266 tests for exception flags executed.
  4 errors occured.