svn commit: r292788 - in head/sys: geom/part sys

Ravi Pokala rpokala at mac.com
Mon Jan 11 04:54:56 UTC 2016


Hi folks,

I have a Panasas-proprietary GPT UUID that I'd like to commit upstream. There's ~zero chance anyone outside of Panasas or our customers would ever see it, but we're trying to reduce gratuitous differences against upstream, and this is a good candidate. When I mentioned this at the November 2015 DevSummit, no one there objected.

There was a recent change in this area, which brought up some questions. See below:

-----Original Message-----


From: <owner-src-committers at freebsd.org> on behalf of Allan Jude <allanjude at FreeBSD.org>
Date: 2015-12-27, Sunday at 10:12
To: <src-committers at freebsd.org>, <svn-src-all at freebsd.org>, <svn-src-head at freebsd.org>
Subject: svn commit: r292788 - in head/sys: geom/part sys

>Author: allanjude
>Date: Sun Dec 27 18:12:13 2015
>New Revision: 292788
>URL: https://svnweb.freebsd.org/changeset/base/292788
>
>Log:
>  Add some additional GPT partition types
>  
>...
>  
>  Differential Revision:	https://reviews.freebsd.org/D3841
>
>...
>

In that review, there was discussion about the sorting of aliases in (enum g_part_alias). It was originally thought that the aliases needed to be sorted, due to binary searching in g_part_alias_name(). It turned out that g_part_alias_name() did a linear search, so the list didn't strictly need to be sorted. But, there was talk of potentially switching to a binary search later, so Allan went with the sorted version.

ae@ and kib@ voiced some concerns:

From: <owner-src-committers at freebsd.org> on behalf of "Andrey V. Elsukov" <bu7cher at yandex.ru>
Date: 2015-12-28, Monday at 00:32
To: Allan Jude <allanjude at FreeBSD.org>, <src-committers at freebsd.org>, <svn-src-all at freebsd.org>, <svn-src-head at freebsd.org>
Subject: Re: svn commit: r292788 - in head/sys: geom/part sys


> This enum could be used by third-party kernel module and in general it
> is not good to change the order of elements here. This can break `gpart
> show` output for such modules. Just for the future. :)


From: <owner-src-committers at freebsd.org> on behalf of Konstantin Belousov <kostikbel at gmail.com>
Date: 2015-12-28, Monday at 02:49
To: "Andrey V. Elsukov" <bu7cher at yandex.ru>
Cc: Allan Jude <allanjude at FreeBSD.org>, <src-committers at freebsd.org>, <svn-src-all at freebsd.org>, <svn-src-head at freebsd.org>
Subject: Re: svn commit: r292788 - in head/sys: geom/part sys


> If the enum values are considered part of KBI, then explicit assignment
> of the enum values is much preferred.


As best as I can tell, nothing in base/head or ports/head looks at the GPT UUIDs or aliases (outside of GEOM_PART itself, of course), so ae@ and kib@'s concerns are not an issue at this time. Therefore, my plan is to insert the new alias and UUID entries into the various enums and lists in-order, rather than just appending. Any objections?

Thanks,

Ravi (rpokala@)



More information about the svn-src-head mailing list