Re: [PATCH 06/14] block: lift setting the readahead size into the block layer

[Christoph Hellwig]

On Wed, Jul 22, 2020 at 07:13:54AM +0000, Johannes Thumshirn wrote:
> On 22/07/2020 08:27, Christoph Hellwig wrote:
> > +	q->backing_dev_info->ra_pages =
> > +		max(queue_io_opt(q) * 2 / PAGE_SIZE, VM_READAHEAD_PAGES);
> 
> Dumb question, wouldn't a '>> PAGE_SHIFT' be better instead of a potentially 
> costly division?
> 
> Or aren't we caring at all as it's a) not in the fast-path and b) compilers 
> can optimize it to a shift?

That's my thinking.  If anyone has a strong preference I can change
it.