[lkml]   [2012]   [Feb]   [22]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: [PATCH 00/10] jump label: introduce very_[un]likely + cleanups + docs
The bike shed is getting really dark.

On Wed, 2012-02-22 at 09:20 -0500, Mathieu Desnoyers wrote:
> [...]
> One possible naming that might be a good fit:
> - read_always_likely()
> - read_always_unlikely()
> - read_always_branch()
> I think it is important to convey both that it is expected to be always
> read, pretty much never updated, and the bias, or absence of bias.

That actually looks even more confusing. "read_always"? What the hell is

> I also _like_ to have a relatively long name here, because the update
> cost is so high that someone should really think before using this
> facility. In my opinion, it's not "just" a stronger likely/unlikely.

Then make it what it is (with a long name...)

if (jump_label_likely())
if (jump_label_unlikely())
That's probably the least confusing of the names. And for the cases we
don't care:

if (jump_label_branch())
The above is the most descriptive and I would say the least confusing.
Someone on IRC said that they wish it had jump_label in the name. As
they see there's a "CONFIG_JUMP_LABEL" it tells us where those jump
labels are used.

Need to go and knock down the bike shed now.

-- Steve

 \ /
  Last update: 2012-02-22 15:39    [from the cache]
©2003-2014 Jasper Spaans. hosted at Digital Ocean