Re: [Patch 5/8] ia64: implement crashkernel=auto

[Amerigo Wang <amwang@redhat.com>]

Bernhard Walle wrote:
> * Ingo Molnar <mingo@elte.hu> [2009-08-24 09:43]:
>   
>> * Amerigo Wang <amwang@redhat.com> wrote:
>>     
>>> The reason that I kept 2ULL<<30 instead of 1ULL<<31 is that '1<<30' is  
>>> exactly 1G, so 2ULL<<30 can be easily read as 2G. ;)
>>>       
>> i have no trouble reading 1ULL<<31 as 2G ;-) OTOH, the logic and 
>> pattern of the comparisons (especially without the comment) looked 
>> odd at first sight, until i noticed this.
>>     
>
> Why not just something like
>
>     #define KBYTE(x)  ((x)*1024ULL)
>     #define MBYTE(x)  ((x)*1024ULL*1024)
>     #define GBYTE(x)  ((x)*1024ULL*1024*1024) 
>     #define TBYTE(x)  ((x)*1024ULL*1024*1024*1024)
>
> I find GBYTE(2) much easier to read than 1ULL<<31. Honestly, I would
> add a comment '/* 2G */' if I would write 1ULL<<31 in own code.
>
> But I'm of course not one of that super kernel hackers. ;-)
>
>   

Yeah, great. Here we only need MBYTES() and GBYTES(). ;)

Thanks.