Re: [RESEND PATCH v3 1/2] sched/deadline: Add cpudl_maximum_dl() for clean-up

From: Byungchul Park
Date: Fri Jun 01 2018 - 02:30:57 EST




On 2018-06-01 15:02, Joel Fernandes wrote:
On Fri, Jun 01, 2018 at 12:07:48PM +0900, Byungchul Park wrote:


On 2018-05-25 14:13, Byungchul Park wrote:


On 2018-05-09 15:33, Byungchul Park wrote:
On Thu, Jan 11, 2018 at 10:07:16AM +0100, Peter Zijlstra wrote:


Sorry for the huge delay on this, but I'll have to postpone further.
Still busy with meltdown/spectre stuff.

Please consider this. Even though it's not a big bug, anyway leading
mis-behavior in certain situaions.

Could you see this patches, it's been too long since the start tho?

Please, any opinion.

Just my opinion: this patch [1] is just a cosmetic change. I would argue that
there's no readability improvement by wrapping up elements[0].dl. Infact I
even feel that the elements[0].cpu should directly be accessed since both
.cpu and .dl for the 0th element are directly accessed only from one place
(cpudl_find) and only one time, and explicitly accessing index 0 makes it
more clear that this is the root of the max-heap.

IOW I don't see any benefit in hiding it behind a wrapper which hides the
fact that we're accessing the root of the max heap, but I don't terribly hate
this patch and I'm Ok if maintainers and other reviewers think its worth it.

Hi Joel,

Talking about the *1st patch*, no matter whether denied or not, even
though I think it looks weird to abstract only p->elements[0].cpu with
a function, but not cp->elements[0].dl.

thanks,

- Joel

[1] https://patchwork.kernel.org/patch/10149099/



--
Thanks,
Byungchul