Hi Naruaki, hi Izumi, On 08/27/2018 09:15 PM, Naruaki.Etomi wrote:
[ 9.0068690] warning: no /dev/console
I think following is a need. (I wonder if it was failure ) # cd /target_hdd/dev # ./MAKEDEV all
Yeah. I think I hadn't done this yet at this point.
/etc/rc.conf is not configured. Multiuser boot aborted.
Please rewrite "rc_configured=YES"
Dito.
The result is working:
I'm so glad. ...apart from the MAC address patch that is not yet included in the ELF image. It slipped my mind！ Try this kerne. (I think I'm OK) https://github.com/nullnilaki/NetBSD_irisboot/blob/master/IndigoR3K/re_netbsd_elf
Thanks for the fix, I'll check it out soon.
SGI machines are roughly 2 categories. ARCS machine: Crimson, Indigo R4K, Indy, Indigo2, O2, Octane... NO ARCS machine: Indigo R3K, Personal IRIS, IRIS Power series... https://en.wikipedia.org/wiki/ARCS_(computing) ARC is a standard and open. I know how to use ARCS routine. For example: https://nxr.netbsd.org/xref/src/sys/arch/sgimips/stand/common/boot.c#177 Therefore, I recognize OSLoadPartition environment variable. https://nxr.netbsd.org/xref/src/sys/arch/sgimips/stand/common/boot.c#190 Therefore, I recognize OSLoadFilename environment variable. On the contrary, NO ARCS machine's PROM(NVRAM) routine is proprietary and close. I don’t know how to use NO ARCS machine's PROM(NVRAM) routine. No one knows it besides SGI. Disassembling sash binary is very very hard. (Please remember Indigo R3K MAC address promblem)
I understand now.
As a result, It is necessary to input target disk and kernel name. I know it sucks, but please give me a break...
Makes perfect sense then. But maybe you could include `dksc(0,1,0)/netbsd` as the default path to the kernel in irisboot, which could be used if no argument is given to irisboot. This could work for the major part of installations where the kernel is on the partiton for "/" on the first disk. And users could still overwrite it with their preferred settings.
In addition one could also install irisboot with "sash" as name in the SGI volume header and the machine could hopefully work with the default settings then.
What do you think about that? I'll give that irisboot-named-sash a try.
Say, which PR refers to the kernel patch(es) (incl. the MAC address fix for early Indigo R3K)?
It would be cool to be able to build that kernel for future NetBSD releases.
I hadn't sent Indigo R3000 patch yet. Give me a sec to explain... As the reason, My Indigo ran out of battery a few days ago! https://twitter.com/nullnilaki/status/1031949233857880064
Arrggghhh. What a pity. And also sad that the TOD clock won't start working on the Indigo as long as no new battery is attached. On other machines (e.g. Indy) it is enough to fix the date in the PROM monitor and reboot to have a working machine for as long as it is powered.
On 09/01/2018 05:18 PM, Izumi Tsutsui wrote:
Today I had a meeting with Etomi-san and reviewed his patches in PRs. At least all kernel changes (PR 53378, 53518, 53519, 53520, 53521, 53522) are okay for me so I'll commit them in a few days. I also replaced dead batteries on all his Indigo R3K and R4000 boards with CR2032 and battery holders, so he can continue his development again.I hope this will help remaining sgimips users
Fortunately Izumi could help you out in the meantime. @Izumi: Cool, that you were around. I'm sadly too far away for a solder job. :-) @Naruaki, Izumi:BTW, I hope I use your first names and not your family names when addressing both of you or let's say I hope I use the preferred way you'd like to be addressed. Please correct me if not.
I will send Indigo R3K patch in 2 months at the latest. I'll contact you later.
Again much obliged.I now also fetched my R4600 Indy from storage and will give it a try with the disk from the Indigo with R4000 version of irisboot and Indy kernel. I'll report back about the results.
Cheers, Frank