On Sun, Aug 17, 2014 at 04:26:53PM +0200, Andy Polyakov via RT wrote: > Hi, > > > 1.0.2 beta2 has a testsuite failure on sparc: > > Testing cipher CAMELLIA-192-OFB(encrypt) > > Key > > 0000 8e 73 b0 f7 da 0e 64 52 c8 10 f3 2b 80 90 79 e5 > > 0010 62 f8 ea d2 52 2c 6b 7b > > IV > > 0000 52 ef 01 da 52 60 2f e0 97 5f 78 ac 84 bf 8a 50 > > Plaintext > > 0000 30 c8 1Makefile:149: recipe for target 'test_evp' failed > > make[2]: Leaving directory '/«PKGBUILDDIR»/test' > > Makefile:455: recipe for target 'tests' failed > > > > I didn't see that problem with beta 1. > > I assume we are talking about Linux. But what is your hardware (T4 or > earlier)? Compiler? Configuration parameters? Thing is that I can't > reproduce this on T4 (with or without hardware-assisted Camellia), not > in 32- nor 64-bit builds. System is equipped with gcc 4.6.3. What's > strange is that fails in OFB, which means that earlier Camellia ECB, > CBC, CFB tests pass. Which in turn means that there is nothing what with > Camellia code itself, and it has to be ofb128.o. But it was tested with > AES earlier in this test...
I was suspecting a compiler bug, I've looked at the assembler before, it looked wrong to me, but don't know enough about sparc. So I actually retried this a few days ago and it worked now. It was on a different machine, but with the same type of CPU: cpu : TI UltraSparc IIIi (Jalapeno) fpu : UltraSparc IIIi integrated FPU pmu : ultra3i prom : OBP 4.11.4 2003/07/23 08:04 type : sun4u It failed with Debian's gcc 4.9.0-9, worked with 4.9.1-5. It looked to me like it generated code for memcpy() wrong. I'm just going to assume it's a compiler bug, feel free to close it. But if you really want I can try to look at the differences in assembler. Kurt ______________________________________________________________________ OpenSSL Project http://www.openssl.org Development Mailing List [email protected] Automated List Manager [email protected]
