svn commit: r287016 - in stable/10: . share/man/man4 sys/arm/mv sys/boot/forth sys/conf sys/dev/ata sys/dev/ata/chipsets sys/modules/ata/atapci/chipsets sys/modules/ata/atapci/chipsets/ataadaptec s...

[Alexander Motin]

On 25.08.2015 13:37, Gleb Smirnoff wrote:
> On Sat, Aug 22, 2015 at 07:32:48AM +0000, Alexander Motin wrote:
> A> Author: mav
> A> Date: Sat Aug 22 07:32:47 2015
> A> New Revision: 287016
> A> URL: https://svnweb.freebsd.org/changeset/base/287016
> A> 
> A> Log:
> A>   MFC r280451:
> A>   Remove from legacy ata(4) driver support for hardware, supported by newer
> A>   and more functional drivers ahci(4), siis(4) and mvs(4).
> A>   
> A>   This removes about 3400 lines of code, unused since FreeBSD 9.0 release.
> 
> What is the reason to break POLA in stable branch? 

Reducing diff from head to make later merges easier.

> Don't get me wrong. I'm all for removing stale code from head. But is it
> worth to hurt someone who runs custom built kernel with 'device ad' and
> tracks stable/10?

There never was legacy ATA stack in stable/10, so there is no such
users. The only POLA breakage here is that people with custom kernels
for some reason having only ata driver may now require ahci, siis or
mvs. If such cases exist, they are most likely a configuration bugs, as
alternative gives NCQ, hot-plug, etc. for free.

-- 
Alexander Motin