Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 20 Mar 2014 22:12:31 -0400
From:      "Littlefield, Tyler" <tyler@tysdomain.com>
To:        kpneal@pobox.com
Cc:        "freebsd-questions@freebsd.org" <freebsd-questions@freebsd.org>
Subject:   Re: kernel config:advice
Message-ID:  <532BA00F.1050401@tysdomain.com>
In-Reply-To: <20140321002111.GB18540@neutralgood.org>
References:  <532B6E81.60604@tysdomain.com> <20140321002111.GB18540@neutralgood.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On 3/20/2014 8:21 PM, kpneal@pobox.com wrote:
> On Thu, Mar 20, 2014 at 06:41:05PM -0400, Littlefield, Tyler wrote:
>> hello:
>> My kernel that is running currently is a bit large and has a lot of
>> support for hardware I do not use. Here is my kernel config:
> What, is your machine memory constrained? The generic kernel is, what 12MB
> in size?  It's a bother using a custom kernel, and I wonder if it is really
> worth the trouble. As someone who used to compile custom kernels for fun
> I have to say I wouldn't bother with it today.
>
> As far as I know the biggest reason to have a custom kernel is if you
> want to use ipsec. But don't hold me to that.
>
>> #device    mps
> You commented out the mps driver.
>
>> #device    scbus
> This is part of the CAM subsystem. Since you are getting compilation errors
> from that subsystem I suggest putting this device back into your config.
> This goes double when the commented out device is actually mentioned in
> your dmesg log.
>
>> errors:
>> tws_cam.c:(.text+0x1093): undefined reference to `xpt_done'
>> tws_cam.o: In function `tws_scsi_complete':
>> tws_cam.c:(.text+0x196d): undefined reference to `xpt_done'
>> tws_cam.o: In function `tws_timeout':
>> tws_cam.c:(.text+0x1afb): undefined reference to `xpt_freeze_simq'
>> tws_cam.c:(.text+0x1b92): undefined reference to `xpt_release_simq'
> Come to think of it, it makes more sense to start with a full GENERIC
> config and then gradually remove devices. That way if you get an error
> compiling/linking you know where to look. Haphazard scattershot changes
> just leaves you with a mess.
>   
>> (probe2:mps0:0:2:0): INQUIRY. CDB: 12 00 00 00 24 00
>> (probe2:mps0:0:2:0): CAM status: Invalid Target ID
>> (probe2:mps0:0:2:0): Error 22, Unretryable error
> Seems like if you have an mps controller you will need that mps device
> that you removed. Otherwise the kernel won't be able to talk to your disks.
>
>> da0 at mps0 bus 0 scbus0 target 0 lun 0
> Here's that mention of scbus. Removing this from your config is probably
> the problem.
>
>> Does anyone see any issues with this? I am unable to compile. While I am
>> not trying to trim out everything, I'd really just like to have this
>> customized for my hardware (this is a dedicated server). I'm going to be
>> putting a lot of load on this server and wanted to do what I could to
>> minimize unwanted overhead (if any really).
> I really doubt you will see any real benefit from this. The kernel just
> isn't that large by today's standards. If it did eliminate any overhead,
> and that overhead mattered, then your machine is waaay too close to the
> limit already and should be upgraded now.
>

Hello:
Thanks a lot for the help--I feel a bit dim after looking at the reply. 
The machine is rather large (32 gb ram), etc. I just wanted to cut some 
of this out. I started generic and was trying to remove unwanted devices 
and options, then got a bit lost with it when I started seeing 
compilation errors. thanks again for the help, I really appreciate it.

-- 
Take care,
Ty
http://tds-solutions.net
He that will not reason is a bigot; he that cannot reason is a fool; he that dares not reason is a slave.




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?532BA00F.1050401>