[Mono-dev] Failing make check
Paul Melis
mono-devel at assumetheposition.nl
Fri May 1 04:39:56 EDT 2009
Hi,
After successfully building mono 2.4 on a freshly installed ubuntu
laptop I get a SIGILL somewhere in the basic tests:
[...]
Test run: image=/home/melis/c/mono-2.4/mono/mini/basic-float.exe,
opts=peephole
Results: total tests: 29, all pass
Elapsed time: 0.004071 secs (0.000318, 0.003753), Code size: 7100
Test run: image=/home/melis/c/mono-2.4/mono/mini/basic-float.exe,
opts=branch
Results: total tests: 29, all pass
Elapsed time: 0.003926 secs (0.000315, 0.003611), Code size: 7299
Test run: image=/home/melis/c/mono-2.4/mono/mini/basic-float.exe, opts=cfold
Results: total tests: 29, all pass
Elapsed time: 0.003806 secs (0.000313, 0.003493), Code size: 7843
Test run: image=/home/melis/c/mono-2.4/mono/mini/basic-float.exe, opts=fcmov
Results: total tests: 29, all pass
Elapsed time: 0.003790 secs (0.000309, 0.003481), Code size: 7254
Test run: image=/home/melis/c/mono-2.4/mono/mini/basic-float.exe, opts=simd
Results: total tests: 29, all pass
Elapsed time: 0.003811 secs (0.000310, 0.003501), Code size: 7843
Test run: image=/home/melis/c/mono-2.4/mono/mini/basic-float.exe, opts=sse2
Unhandled Exception: System.ExecutionEngineException: SIGILL
at Tests.test_0_conv_i () [0x00000]
make[4]: *** [rcheck] Error 1
make[4]: Leaving directory `/home/melis/c/mono-2.4/mono/mini'
make[3]: *** [check-am] Error 2
make[3]: Leaving directory `/home/melis/c/mono-2.4/mono/mini'
make[2]: *** [check] Error 2
make[2]: Leaving directory `/home/melis/c/mono-2.4/mono/mini'
make[1]: *** [check-recursive] Error 1
make[1]: Leaving directory `/home/melis/c/mono-2.4/mono'
make: *** [check-recursive] Error 1
It seems that an attempt is made to run code using optimizations that
use SSE2 instructions, however, the laptop's CPU (an old P3) does not
have SSE2, only SSE. I don't see any switch to ./configure to tell it to
not assume SSE2 is available. So is this a bug in runtime CPU
identification, or perhaps simply in the way the test cases are run? Is
my mono installation now botched?
Regards,
Paul
More information about the Mono-devel-list
mailing list