the "proper" way to identify the bitness of your kernel and CPU

[Sharpe, Sam J]

2009/4/3 Robert P. J. Day <rpjday at crashcourse.ca>:
> On Fri, 3 Apr 2009, Sharpe, Sam J wrote:
>
>> 2009/4/3 Robert P. J. Day <rpjday at crashcourse.ca>:
>> >
>> >  what is the fedora-approved way to identify the wordsize of both
>> > your running kernel and your CPU?  for the kernel, i'm used to running
>> >
>> >  $ uname -r
>> >
>> > and just looking at the suffix, which in my case would be either
>> > "i686" or "x86_64".  is there a simpler way?
>>
>> # man arch
>>    NAME
>>        arch - print machine hardware name (same as uname -m)
>>
>> # arch
>> x86_64
>>
>> # uname -m
>> x86_64
>
>  hang on ... if i have a 32-bit kernel running on a 64-bit CPU, what
> will the above print?  i'd want it to print, well, i686 or something
> that represents 32 bits.  is that what it would do?

Yes, it's the right way to run "uname -r" and look at the suffix...
it's the same thing, only it just prints out the suffix.

e.g. A 64bit capable CPU with a 32 bit (RHEL5) OS:

[sam at machine ~]$ arch
i686 (this represents the installed kernel architecture)

[sam at machine ~]$ [[ $(grep lm /proc/cpuinfo) ]] && echo x86_64 || echo i686
x86_64 (this represents the processor architecture)

-- 
Sam