Re: [PATCH] fs/fat: calculate checksum in a loop instead of directly calculating

[Heinrich Schuchardt]

On 24.02.2015 03:42, OGAWA Hirofumi wrote:
> Heinrich Schuchardt <xypron.glpk@xxxxxx> writes:
> 
>> You have to put
>> __attribute__((optimize("unroll-loops")))
>> here to unroll the loop inside the function:
>>
>>
>> static __attribute__ ((noinline))
>> __attribute__((optimize("unroll-loops")))
>> int test(unsigned char *name)
>>
>> $ time ./c.inline aaaaaaaaaaa
>> 14
>>
>> real    0m0.743s
>> user    0m0.740s
>> sys     0m0.000s
>>
>> Without __attribute__((optimize("unroll-loops"))) :
>> $ time ./c.loop aaaaaaaaaaa
>> 14
>>
>> real    0m1.482s
>> user    0m1.472s
>> sys     0m0.004s
>>
>> With __attribute__((optimize("unroll-loops"))) :
>>
>> $ time ./c.loop aaaaaaaaaaa
>> 14
>>
>> real    0m0.742s
>> user    0m0.740s
>> sys     0m0.000s
> 
> This attribute has to be added to the caller, not fat_checksum()?  I.e.,
> we has to add it to all callers of fat_checksum()?

The attribute is applied to the declaration of the function in which you
need loop unrolling.
https://gcc.gnu.org/onlinedocs/gcc-4.9.1/gcc/Attribute-Syntax.html

> 
> Well, this is interesting gcc optimize option though, maybe not worth to
> introduce this to kernel only for fatfs.

There is an effort to compile the Linux kernel with Clang. We should
avoid new GCC specific items.

Best regards

Heinrich
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/