Re: 7.1 on a IBM 43P, 7248-132

[T <bobs%thelibertytree.org@localhost>]

I was able to get rid of of most of the remaining issues by changing a 
few kernel conf options.
By switching out the Intel 10/100/1000 NIC with a Netgear 10/100 NIC, 
the following errors went away:

PCI bus 0: Warning: Total bandwidth exceeded!? (4091954)

cd0(siop0:0:6:0): Check Condition on CDB: 0x46 00 00 00 00 00 00 00 08 
00 SENSE KEY: Illegal Request ASC/ASCQ: Invalid Command Operation Code

and by enabling universal PHY module "ukphy*  at mii? phy 
?                   # generic unknown PHYs":

OUI 0x005500 model 0x0038 rev 0 at wm0 phy 1 not configured wm0: Assumed 
PHY access function (type = 0) might be incorrect. Use BM and retry. OUI 
0x005500 model 0x0038 rev 0 at wm0 phy 1 not configured

Only issue left when loading the kernel is that no default boot device 
is detected, but I can get around this by being specific in the config 
"config netbsd root on sd0 type ffs". This will allow the boot to 
continue, even though the message still displays:

boot device: <unknown>

The integrated pcn0 still locks up the machine if I try to bring it up 
with "ifconfig pcn0 up" so not sure about that one. Has anyone else with 
a hardware machine have this issue with 7.1 or newer?

On 07/22/2017 03:55 AM, T wrote:
> Finally got this to work:
> http://dmesgd.nycbug.org/index.cgi?do=view&id=3228
>
> I had to recompile the kernel to add in support for a PCI NIC and PCI 
> USB card and took the opportunity to trim out some things that were 
> not needed or planned on being used (WD).
>
> I was wrong about the 2MB limit but ended up making a small kernel to 
> put into the PReP partion based on the INSTALL_SMALL kernel config.
>
> I can post the kernel config files if you would find it useful.
>
> On 07/01/2017 01:34 AM, T wrote:
> > I'm using the integrated ncr53c810 SCSI-2 controller so it may not be 
> > the same issue you are having.
> >
> > Using sysinst_small.fs in the PReP partition allowed loading of the 
> > boot loader. The boot process will proceed to a point where it 
> > requires user input to specify a boot device because it does not find 
> > one by default. Providing it "sd0" will allow for the generic kernel 
> > to load (when going multi-user?) and provide the login prompt.
> >
> > If I enter "boot -s" at the boot loader prompt (single user), it will 
> > boot into sysinst.
> >
> > sysinst.fs will boot when I network boot with BOOTP/TFTP. I have not 
> > tried writing that into the PReP partition (as I did with the small 
> > version) but I suspect it would work because it is less than 2MB in 
> > size.
> >
> > sysinst_com0.fs used in the same way mentioned above will not work 
> > for me. I have not tried sysinst_small_com0.fs in this manner.
> >
> >
> > On 06/30/2017 03:35 AM, Artyom Tarasenko wrote:
> > > Interesting. At a first glance the log says that the scsi driver is
> > > broken in the NetBSD 7+. I hit that problem in this thread:
> > > http://mail-index.netbsd.org/port-prep/2017/01/23/msg000108.html
> > >
> > > Are you saying the very same NetBSD version works for you if you use a
> > > smaller sysinit image?
> > >
> > > On Fri, Jun 30, 2017 at 7:23 AM, T <bobs%thelibertytree.org@localhost> wrote:
> > > > The PReP partition was set to use 5MB (as recommended in the 
> > > > documentation)
> > > > and located in the first BIOS partition. Only the kernels 2MB or 
> > > > less seem
> > > > to make it to the boot loader for me. The <2MB serial console 
> > > > sysinit image
> > > > will make it to the boot loader if I put it into the PReP 
> > > > partition, but
> > > > will not boot past that point as mentioned in this thread:
> > > > http://mail-index.netbsd.org/port-prep/2017/06/27/msg000119.html
> > > >
> > > >
> > > >
> > > > On 06/29/2017 03:32 AM, Artyom Tarasenko wrote:
> > > > > Is it a general issue, or was the PReP partition just too small in 
> > > > > your
> > > > > setup?
> > > > >
> > > > > On Thu, Jun 29, 2017 at 10:15 AM, T <bobs%thelibertytree.org@localhost> wrote:
> > > > > > Found that the kernel size is the issue by using mkbootimage with
> > > > > > install_small.fs and dding that into the PreP partition and 
> > > > > > getting that
> > > > > > to
> > > > > > boot. Will end up doing the small kernel work around mentioned in 
> > > > > > the
> > > > > > referenced email thread.
> > > > > >
> > > > > >
> > > > > >
> > > > > > On 06/24/2017 06:37 PM, t wrote:
> > > > > > > Hello,
> > > > > > > I'm trying to get NetBSD 7.1 working on a 7248-132 but cannot 
> > > > > > > get the
> > > > > > > computer to boot from the SCSI disk after installation. Booting the
> > > > > > > install kernel over the network seems to work with BOOTP/TFTP. 
> > > > > > > After
> > > > > > > completing the installation process and restarting the machine, 
> > > > > > > I load
> > > > > > > into the SMS and change the boot priority to boot from the SCSI 
> > > > > > > disk
> > > > > > > instead of the network. Upon restarting and completing the boot 
> > > > > > > screen,
> > > > > > > the computer ends up at a black screen and blinking cursor.
> > > > > > > I believe this issue may be related to this thread:
> > > > > > > https://mail-index.netbsd.org/port-prep/2014/07/thread1.html#000101
> > > > > > >
> > > > > > > This is my first time installing NetBSD so it could also be 
> > > > > > > something
> > > > > > > I'm
> > > > > > > doing.
> > > > > > >
> > > > > > > I've tried re-writing the generic kernel to the prep partition with
> > > > > > > "mkbootimage" but that did not seem to work.
> > > > > > > Any help would be appreciated.
> > > > > > >
> > > > > > >
> > > > > > > One other thing I've noticed with this machine is interactions 
> > > > > > > with the
> > > > > > > pcn0 device seem to occasionally lockup the machine. I 
> > > > > > > originally was
> > > > > > > going to install the NetBSD binary packages from a NFS mount, 
> > > > > > > but the
> > > > > > > system would lockup halfway through the installation of the second
> > > > > > > package. I bypassed this issue by using the CDROM instead.