On 12/05/2019 03:06 PM, Matthew Wilcox wrote:
On Thu, Dec 05, 2019 at 08:56:07AM +0800, Tiezhu Yang wrote:And, as I asked twice in the last round of review, did you benchmark this change?Before sending this v2 patch, I have done the test used with your test program and already pointed out the following implementation is better:I didn't mean "have you run the test program i wrote". I meant "have you booted a kernel with this change and done some performance measurements to see if you've changed anything".
Oh, no, it is hard to measure the performance influence with this patch. Based on the above analysis, I think the performance influence is very small due to is_dot_dotdot() is a such short static inline function. Thanks, Tiezhu Yang
|