Re: git: 7be29291845a - stable/14 - UPDATING: fix a vestigial reference to -CURRENT

26.08.2023 7:32, Glen Barber wrote:

Moving to stable@ and net@ as requested.

> On Sat, Aug 26, 2023 at 05:13:50AM +0700, Eugene Grosbein wrote:
>> 26.08.2023 3:55, Glen Barber wrote:
>>
>>>> Please note the following commit:
>>>> https://cgit.freebsd.org/src/commit/libexec/rc/rc.conf?id=ccc806a049383e5611b3752e6f384cf03a208039
>>>>
>>>> I mean, we need to comment out dumpdev="AUTO" for stable/14, too.
>>>>
>>>
>>> On stable/14, dumpdev is set to "NO", per the documented changes to make
>>> on a new stable branch.
>>>
>>> https://docs.freebsd.org/en/articles/freebsd-releng/#releng-head-branching
>>>
>>> As it is, it matches libexec/rc/rc.conf from stable/13.
>>
>> dumpdev="NO" in /etc/defaults/rc.conf overrides dumpdev in /boot/loader.conf
>> That should not happen.
>>
> 
> This seems backwards to me.  The implementation has always been handled
> in rc.conf, so as an end-user that explicitly set something(tm) in
> rc.conf, I do not expect to have to change loader.conf as well.

Right. However, dumpdev is special one. A panic in STABLE could occur before /etc/rc.conf is consulted.
Therefore, one have to specify dumpdev in the loader.conf instead of rc.conf in such a case.
Sadly, dumpdev in the loader.conf is later overridden by dumpdev="NO" in /etc/defaults/rc.conf
unless it is duplicated in /etc/rc.conf OR we comment out dumpdev="NO" in /etc/defaults/rc.conf for STABLE.

I suggest we do the latter for STABLE.

> Having said that, since both stable/14 and stable/13 are in sync in this
> regard (dumpdev="NO" in rc.conf), I will not object to commenting the
> entry in releng/14.0 when it is branched.
> 
> I would also like this taken to a more appropriate list, such as
> -current or -stable.
> 
> Glen
> 

Eugene