Re: [TRIVIAL] Re: UP went into unexpected trashing

[Petr Vandrovec]

On  8 Nov 02 at 19:33, Rusty Trivial Russell wrote:
> --- trivial-2.5-bk/include/asm-i386/bitops.h.orig   2002-11-08 18:47:20.000000000 +1100
> +++ trivial-2.5-bk/include/asm-i386/bitops.h    2002-11-08 18:47:20.000000000 +1100
> @@ -311,12 +311,13 @@
>         "repe; scasl\n\t"
>         "jz 1f\n\t"
>         "leal -4(%%edi),%%edi\n\t"
> -       "bsfl (%%edi),%%eax\n"
> -       "1:\tsubl %%ebx,%%edi\n\t"
> +       "bsfl (%%edi),%%edx\n"
> +       "subl %%ebx,%%edi\n\t"
>         "shll $3,%%edi\n\t"
> -       "addl %%edi,%%eax"
> +       "addl %%edi,%%edx\n\t"
> +       "1:\tmovl %%edx,%%eax\n\t"
>         :"=a" (res), "=&c" (d0), "=&D" (d1)
> -       :"1" ((size + 31) >> 5), "2" (addr), "b" (addr));
> +       :"1" ((size + 31) >> 5), "2" (addr), "b" (addr), "d" (size));
>     return res;

EDX is modified, should not you list "=d" as output, with new variable (d2)?

Or better, drop last line of assembly code, and say that (res) is in
"d", and list "a" as clobbered or dummy output register.

And BTW, if you'll do 

unsigned long b = 0x8000;
find_first_bit(&b, 1);

return value will be 15 even with patched function. So either more
fixing is needed, or code which calls find_first_bit() with value which
is not multiple of 32 should take special care that last dword does
not contain set bits after last interesting bit.

So maybe callers should just treat any return value >= size as "not found",
leaving older smaller code in place.
                                            Best regards,
                                                Petr Vandrovec
                                                [e-mail]