Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > comp.lang.c > #237840
| From | Tim Rentsch <tr.17687@z991.linuxsc.com> |
|---|---|
| Newsgroups | comp.lang.c |
| Subject | Re: portable way to get highest bit set? |
| Date | 2023-10-12 07:58 -0700 |
| Organization | A noiseless patient Spider |
| Message-ID | <86zg0nanw5.fsf@linuxsc.com> (permalink) |
| References | (4 earlier) <20231012111100.272c96b3209baad26a150e55@g{oogle}mail.com> <86cyxkb2ka.fsf@linuxsc.com> <20231012141719.99f5a10ec921db3ee6f7d948@g{oogle}mail.com> <864jiwaqic.fsf@linuxsc.com> <ffTVM.60121$8fO.5299@fx15.iad> |
scott@slp53.sl.home (Scott Lurndal) writes:
> Tim Rentsch <tr.17687@z991.linuxsc.com> writes:
>
>> Anton Shepelev <anton.txt@g{oogle}mail.com> writes:
>>
>>> Tim Rentsch:
>>>
>>>> But the approach you are using requires knowing the size
>>>> of the type involved. The challenge is to write code that
>>>> works without having to know the size in advance.
>>>
>>> Well, I could use macros, or the largest integer type, or:
>>
>> Neither necessary nor sufficient. The code should work for
>> the type T being __uint128_t, which is larger than uintmax_t.
>>
>>> #include <stddef.h>
>>> #include <limits.h>
>>>
>>> [..examine value character-by-character, including
>>> determining if little-endian encoding is used..]
>>
>> You're making things more complicated than they need to be.
>> Looking at character values rather than just values of the type
>> involved is almost guaranteed to have potential representation
>> issues. For example, little endian and big endian are not the
>> only choices possible for endianness.
>>
>> Below is a proposed solution that mostly works. What problems
>> does it have, and how can they be fixed? And can we find
>> something better?
>>
>> /* The type T has been supplied somewhere upstream. It */
>> /* is an unsigned type, and possibly outside the realm */
>> /* of 'integer types' as defined by the C standard, as */
>> /* for example __uint128_t. */
>>
>> T
>> highest_bit_set( T u ){
>> T r = 1;
>> while( u>>1 > r ) r <<= 1;
>> return r;
>> }
>
> Given a reasonable compiler (gcc or clang), I'd use
>
> /**
> * Return the 1-relative bit number of the
> * highest set bit. If no bits are set,
> * return zero.
> */
> size_t
> highest_bit_set(unsigned long long u)
> {
> return u ? 64ul - __builtin_clzll(u) : 0ul;
> }
That gives a solution to a different problem.
When you have code that solves the problem being
discussed, I'm interested to see it.
Back to comp.lang.c | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread
Re: portable way to get highest bit set? Anton Shepelev <anton.txt@g{oogle}mail.com> - 2023-10-12 11:11 +0300
Re: portable way to get highest bit set? Michael S <already5chosen@yahoo.com> - 2023-10-12 02:27 -0700
Re: portable way to get highest bit set? Tim Rentsch <tr.17687@z991.linuxsc.com> - 2023-10-12 02:41 -0700
Re: portable way to get highest bit set? Anton Shepelev <anton.txt@g{oogle}mail.com> - 2023-10-12 14:17 +0300
Re: portable way to get highest bit set? Tim Rentsch <tr.17687@z991.linuxsc.com> - 2023-10-12 07:01 -0700
Re: portable way to get highest bit set? scott@slp53.sl.home (Scott Lurndal) - 2023-10-12 14:18 +0000
Re: portable way to get highest bit set? Tim Rentsch <tr.17687@z991.linuxsc.com> - 2023-10-12 07:58 -0700
Re: portable way to get highest bit set? Michael S <already5chosen@yahoo.com> - 2023-10-12 17:30 +0300
Re: portable way to get highest bit set? Tim Rentsch <tr.17687@z991.linuxsc.com> - 2023-10-12 08:05 -0700
Re: portable way to get highest bit set? Michael S <already5chosen@yahoo.com> - 2023-10-12 19:16 +0300
Re: portable way to get highest bit set? Anton Shepelev <anton.txt@g{oogle}mail.com> - 2023-10-12 18:31 +0300
csiph-web