>>>the attached patch fixed compilation error of CVS head on Linux/AMD64 >>>with GCC 3.3.3. The problem is that instruction 'bswapl' need 32b >>>registers, but the inline assembler in HOST_c2l() led to using 64b ones, >> >>??? l argument passed to HOST_c2l macro should be 32-bit unsigned int >>even on Linux/AMD64... At least I get that variable declared as 32-bit >>unsigned int and gcc [3.3.2] compiles the whole thing and the test >>passes... Not that something is inherently wrong with proposed patch, >>but I'd like to figure out how you ended up in such situation. Can you >>complement report with 'apps/openssl version -a' output? A. > > Oops, I used a wrong compilation options and omitted -DMD32_REG_T=int so > it failed. With the default config it certainly works. > Sorry for the noise. ^^^^^ Don't be! -DMD32_REG_T is a hook to influence behaviour of particular compiler and the fact that you didn't use default options does *not* disqualifies your original post as bug report. I mean it should compile even without -DMD32_REG_T=int and I'm commiting the fix. Cheers. A.
______________________________________________________________________ OpenSSL Project http://www.openssl.org Development Mailing List [EMAIL PROTECTED] Automated List Manager [EMAIL PROTECTED]
