Re: Intermediate void* casts
Geoff Wing <[email protected]>
| Newsgroups | gmane.os.netbsd.devel.general |
|---|---|
| Organization | PrimeNet Computer Consultants |
| Message-ID | <[email protected]> |
Martin Husemann <[email protected]> typed: : On Sun, Aug 10, 2003 at 05:15:43PM -0700, Bill Studenmund wrote: :> I'm not so sure about that. Can/will the compiler notice that you're only :> using half of a 64-bit value? Otherwise, by making this change, you're : If gcc does miss the obvious, the 32bit code might be slightly slower : than the old code. I can't test that right now, anyone with i386 gcc 3.3.1 : around? If the value is spread accross two 32 bit registers, and the code Works OK. On i386: compiled -O2 with compiled -O2 with gcc 2.95.3 (NetBSD nb4) gcc 3.3.1 (from NetBSD src yesterday) 00000000 <bswap64>: 00000000 <bswap64>: 0: 55 push %ebp 0: 55 push %ebp 1: 89 e5 mov %esp,%ebp 1: 89 e5 mov %esp,%ebp 3: 53 push %ebx | 3: 8b 45 08 mov 0x8(%ebp),%e 4: 8b 45 08 mov 0x8(%ebp),%e | 6: 53 push %ebx 7: 8b 55 0c mov 0xc(%ebp),%e 7: 8b 55 0c mov 0xc(%ebp),%e a: 89 c1 mov %eax,%ecx a: 89 c1 mov %eax,%ecx c: 66 c1 c9 08 ror $0x8,%cx c: 66 c1 c9 08 ror $0x8,%cx 10: c1 c9 10 ror $0x10,%ecx 10: c1 c9 10 ror $0x10,%ecx 13: 66 c1 c9 08 ror $0x8,%cx 13: 66 c1 c9 08 ror $0x8,%cx 17: 89 d0 mov %edx,%eax | 17: 89 d3 mov %edx,%ebx 19: 31 d2 xor %edx,%edx | 19: 66 c1 cb 08 ror $0x8,%bx 1b: 66 c1 c8 08 ror $0x8,%ax | 1d: c1 cb 10 ror $0x10,%ebx 1f: c1 c8 10 ror $0x10,%eax | 20: 66 c1 cb 08 ror $0x8,%bx 22: 66 c1 c8 08 ror $0x8,%ax | 24: 31 c0 xor %eax,%eax 26: 31 db xor %ebx,%ebx | 26: 89 ca mov %ecx,%edx 28: 89 cb mov %ecx,%ebx | 28: 89 d9 mov %ebx,%ecx 2a: 31 c9 xor %ecx,%ecx | 2a: 31 db xor %ebx,%ebx 2c: 31 d2 xor %edx,%edx | 2c: 09 da or %ebx,%edx 2e: 09 c1 or %eax,%ecx | 2e: 09 c8 or %ecx,%eax 30: 09 d3 or %edx,%ebx | 30: 8b 1c 24 mov (%esp,1),%eb 32: 89 c8 mov %ecx,%eax | 33: c9 leave 34: 89 da mov %ebx,%edx | 34: c3 ret 36: 5b pop %ebx < 37: c9 leave < 38: c3 ret < Following is how gcc 3.4 20030806 (experimental) with -O2 makes it: 00000000 <bswap64>: 0: 55 push %ebp 1: 89 e5 mov %esp,%ebp 3: 8b 45 08 mov 0x8(%ebp),%eax 6: 53 push %ebx 7: 8b 55 0c mov 0xc(%ebp),%edx a: 89 c1 mov %eax,%ecx c: 66 c1 c9 08 ror $0x8,%cx 10: c1 c9 10 ror $0x10,%ecx 13: 66 c1 c9 08 ror $0x8,%cx 17: 89 d3 mov %edx,%ebx 19: 66 c1 cb 08 ror $0x8,%bx 1d: c1 cb 10 ror $0x10,%ebx 20: 66 c1 cb 08 ror $0x8,%bx 24: 31 c0 xor %eax,%eax 26: 89 ca mov %ecx,%edx 28: 89 d9 mov %ebx,%ecx 2a: 31 db xor %ebx,%ebx 2c: 09 da or %ebx,%edx 2e: 09 c8 or %ecx,%eax 30: 5b pop %ebx 31: c9 leave 32: c3 ret Regards, -- Geoff Wing : <[email protected]> Rxvt Stuff : <[email protected]> Zsh Stuff : <[email protected]>