Re: ERLITE-3 boot problems

On 18/08/2020 18:02, Brook Milligan wrote:
>
>> On Aug 17, 2020, at 7:55 PM, Simon Burge <[email protected]> wrote:
>>
>> If you don't mind using -current instead of NetBSD 9 Jared has made an
>> image that you can dd(1) on to your USB then it magically resizes the
>> rootfs up to the USB drive size on first boot:
>>
>>     https://nam01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fnycdn.netbsd.org%2Fpub%2FNetBSD-daily%2FHEAD%2F202008170740Z%2Fevbmips-mips64eb%2Fbinary%2Fgzimg%2Focteon.img.gz&amp;data=01%7C01%7Cbrook%40biology.nmsu.edu%7C6425c61e24fc4fc1254c08d84319e09e%7Ca3ec87a89fb84158ba8ff11bace1ebaa%7C1&amp;sdata=XtN2yAnIEJ6d6tudiu36ydNu2V7Ys0p7pVMhTZu%2BhsU%3D&amp;reserved=0
>
> I have tried this with no luck.
>
> I copied the image to my USB drive with dd.  The ERLITE did not detect the kernel (see script below).  Putting it back into my NetBSD-9/amd64 machine, it was detected as umass0 not sd0, so I had to reboot.  (How can the busses be rescanned to detect this properly?)
>
> Looking more carefully at the image, it seems to have an NTFS partition and a NetBSD partition defined in a GPT (see below).  However, I could not mount either to look at their contents.
>
> In any case, this is not the solution for the ERLITE, which brings me back to the disklabel questions.  Should the kernel have to fabricate a geometry and is that a warning sign?  How can a proper disklabel be written so that the ERLITE (or now OCTEON) kernel can detect it as the root filesystem?
>
> Thanks for any help.
>
> Cheers,
> Brook



I have

bootcmd=fatload usb 0 $loadaddr vmlinux.64;bootoctlinux $loadaddr
coremask=0x3 root=wedge:octeon-root

but the image uses "netbsd" for kernel file name on the FAT partition

Nick