lkml.org 
[lkml]   [2019]   [Sep]   [18]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: [PATCH 2/2] block, bfq: delete "bfq" prefix from cgroup filenames
From
Date


> Il giorno 17 set 2019, alle ore 23:32, Tejun Heo <tj@kernel.org> ha scritto:
>
> Hello,
>
> On Tue, Sep 17, 2019 at 06:51:48PM +0200, Paolo Valente wrote:
>> When bfq was merged into mainline, there were two I/O schedulers that
>> implemented the proportional-share policy: bfq for blk-mq and cfq for
>> legacy blk. bfq's interface files in the blkio/io controller have the
>> same names as cfq. But the cgroups interface doesn't allow two
>> entities to use the same name for their files, so for bfq we had to
>> prepend the "bfq" prefix to each of its files. However no legacy code
>> uses these modified file names. This naming also causes confusion, as,
>> e.g., in [1].
>>
>> Now cfq has gone with legacy blk, so there is no need any longer for
>> these prefixes in (the never used) bfq names. In view of this fact, this
>> commit removes these prefixes, thereby enabling legacy code to truly
>> use the proportional share policy in blk-mq.
>
> So, I wrote the iocost switching patch and don't have a strong
> interest in whether bfq prefix should get dropped or not. However, I
> gotta point out that flipping interface this way is way out of the
> norm.
>
> In the previous release cycle, the right thing to do was dropping the
> bfq prefix but that wasn't possible because bfq's interface wasn't
> compatible at that point and didn't made to be compatible in time.
> Non-obviously different interface with the same name is a lot worse
> than giving it a new name, so the only acceptable course of action at
> that point was keeping the bfq prefix.
>
> Now that the interface has already been published in a released
> kernel, dropping the prefix would be something extremely unusual as
> there would already be users who will be affected by the interface
> flip-flop. We sometimes do change interfaces but I'm having a
> difficult time seeing the overriding rationales in this case.
>

This issue is a nightmare :)

Userspace wants the weight to be called weight (I'm not reporting
links to threads again). *Any* solution that gets to this is ok for me.

A solution that both fulfills userspace request and doesn't break
anything for hypothetical users of the current interface already made
it to mainline, and Linus liked it too. It is:
19e9da9e86c4 ("block, bfq: add weight symlink to the bfq.weight cgroup parameter")

But it was then reverted on Tejun's request to do exactly what we
don't want do any longer now:
cf8929885de3 ("cgroup/bfq: revert bfq.weight symlink change")

So, Jens, Tejun, can we please just revert that revert?

Thanks,
Paolo

> Thanks.
>
> --
> tejun

\
 
 \ /
  Last update: 2019-09-18 07:19    [W:0.143 / U:0.820 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site