Re: Kernel config for Sun 3/50

David Brownlee <[email protected]> Mon, 30 Jan 2023 18:34:53 +0000
Newsgroups gmane.os.netbsd.ports.sun3
Message-ID <CAGN_6pbf1RDRan79uWQpB1nB-XXjaTfEK9Xj865u2nScC0D8kQ@mail.gmail.com>
--0000000000004caa7505f37f7978
Content-Type: text/plain; charset="UTF-8"

On Thu, 26 Jan 2023 at 01:32, Chris Hanson <[email protected]>
wrote:

> On Jan 20, 2023, at 9:11 PM, Trent McNair <[email protected]> wrote:
> >
> > I commented out the kernel size check code and built a minimal kernel,
> > it does boot and tries to go into single-user mode but only after a
> > very long time . After that, things start crashing, complaining about
> > swap space. I'm not sure at what point in the boot sequence the swap
> > gets activated, but I don't see any indication on my NFS server that
> > the swap file is mounted or accessed. I tried swapctl -A from the
> > shell prompt but that crashed as well. I've attached a boot log below.
>
> One thing I've done when booting minimal-memory systems (such as an HP
> 9000-332 with 8MB) is put attaching a swap file at the very top of /etc/rc,
> because I was seeing the attempt to execute rc failing due to lack of
> memory. Normally swap is attached towards the middle or end of /etc/rc,
> from what I recall anyway.
>
>
That's a very good point

% cd /etc/rc.d && rcorder * | head -20 | fmt
wdogctl fsck_root modules resize_root root devpubd-early devpubd
zfs ccd rndctl cgd lvm raidframe sysctl DISKS dkstrat swap1 fsck
mountcritlocal clearcritlocal

>From a quick look at the rc startup order for NetBSD-10 the system runs a
few things before swap1 gets called. swap2 comes much later (swap1 mounts
swap devices, while swap2 mounts on files).

There is possibly an argument for a "swap0", to come directly after
fsck_root, which would attempt to mount available swap devices, and swap1
could be tweaked to ignore existing device mounts

David

--0000000000004caa7505f37f7978
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_quote"><div dir=3D"ltr" class=3D"gmail=
_attr">On Thu, 26 Jan 2023 at 01:32, Chris Hanson &lt;<a href=3D"mailto:cmh=
[email protected]">[email protected]</a>&gt; wrote:<br></div=
><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border=
-left:1px solid rgb(204,204,204);padding-left:1ex">On Jan 20, 2023, at 9:11=
 PM, Trent McNair &lt;<a href=3D"mailto:[email protected]" target=3D"_blank=
">[email protected]</a>&gt; wrote:<br>
&gt; <br>
&gt; I commented out the kernel size check code and built a minimal kernel,=
<br>
&gt; it does boot and tries to go into single-user mode but only after a<br=
>
&gt; very long time . After that, things start crashing, complaining about<=
br>
&gt; swap space. I&#39;m not sure at what point in the boot sequence the sw=
ap<br>
&gt; gets activated, but I don&#39;t see any indication on my NFS server th=
at<br>
&gt; the swap file is mounted or accessed. I tried swapctl -A from the<br>
&gt; shell prompt but that crashed as well. I&#39;ve attached a boot log be=
low.<br>
<br>
One thing I&#39;ve done when booting minimal-memory systems (such as an HP =
9000-332 with 8MB) is put attaching a swap file at the very top of /etc/rc,=
 because I was seeing the attempt to execute rc failing due to lack of memo=
ry. Normally swap is attached towards the middle or end of /etc/rc, from wh=
at I recall anyway.<br>
<br></blockquote><div><br></div><div>That&#39;s a very good point</div><div=
><br></div><div>% cd /etc/rc.d &amp;&amp; rcorder * | head -20 | fmt<br>wdo=
gctl fsck_root modules resize_root root devpubd-early devpubd<br>zfs ccd rn=
dctl cgd lvm raidframe sysctl DISKS dkstrat swap1 fsck<br>mountcritlocal cl=
earcritlocal<br></div></div><div class=3D"gmail_quote"><br></div><div class=
=3D"gmail_quote">From a quick look at the rc startup order for NetBSD-10 th=
e system runs a few things before swap1 gets called. swap2 comes much later=
 (swap1 mounts swap devices, while swap2 mounts on files).</div><div class=
=3D"gmail_quote"><br></div><div class=3D"gmail_quote">There is possibly an =
argument for a &quot;swap0&quot;, to come directly after fsck_root, which w=
ould attempt to mount available swap devices, and swap1 could be tweaked to=
 ignore existing device mounts</div><div class=3D"gmail_quote"><br></div><d=
iv class=3D"gmail_quote">David<br></div></div>

--0000000000004caa7505f37f7978--