Re-Install w/ New Mobo AMD CPU EFI

Hi:

This is a “re-install” done via SLED12 SP3 DVD (actually UPDATE). Re-install is actually having to move my two HDDs (1TB each) and 1 SSD (14GB) with a DVD R/W unit to a new MOBO, w/ AMD Ryzen 4 Core CPU and 8GB of RAM (similar to prior system that had a powersupply failure that took out the MOBO).

Went into the BIOS to figure out what it was set for, defaults, etc. Had to make a config change so that it would boot from the DVD/CD any time it needed to restart and bootable media was in the tray.

In order to get everything connected where it needed to be (across the SATA connectors) I booted Knoppix 7.6. This failed. Got a copy of 7.6.1 and that was able to boot and run with this setup:

ASUS Prime B450-Plus motherboard, AMD Ryzen w/O graphics, external Graphics (RADEON and I’ve forgotten the exact model, but it works), two monitors, and bluetooth keyboard/mouse (USB).

So now the two HDDs are connected in SATA 1 & 2 as they had been in the original machine.

Booted using a Knoppix 7.6.1 live DVD, I mounted the various partitions and verified they are there and good (so power supply failure didn’t wreck the HDDs).

Here is where things go down hill.

Attempted to boot off the harddrive(s) and the system could not find the O/S.

[Also, I do not put in the registration code during the running of the SLED12 SP3 installer. The SUSE system counts those – even though you are working on the same system. But it doesn’t have a problem when the system “registers” or “re-registers” once booted since it will have the name it had from the last time.]

Ok, so the prior machine did have EFI, so I figured an UPGRADE using the SLED12 SP3 DVD should find and solve the problems with “Grub” [I am not sure that these drives were done using GPT if this even matters at this point.].

No joy. The upgrade logic couldn’t find the correct “/” partition. I don’t know why, but it was looking at a test partition (sdb2) that was not a SLED partition (hadn’t needed to deal with this for a long time and had forgotten it was there). Yes, you can tell it which one to use, and I did that and that didn’t work either.

So I restarted all of this and did an 'INSTALL" where I told it to format the SWAPs (1 per each drive), and to format the system partition(s) (/home was to be untouched as it is RAID 0 mirrored between the two HDDs).

This worked as I expected, and I caught the forced rename to “linux” and corrected it to the correct name for this workstation. [Been down this road before so I know that this type of install works.]

The install code then went off to reboot (and I removed the DVD from the tray), and the BIOS did not find any boot (should have found Grub). All I got was a flashing underscore about 3 lines down the screen.

Ok, so I booted the SLED DVD again and told it to boot the system on disk. Same problem – it couldn’t find the O/S and so immediately rebooted the DVD.

I have made a change so that the EFI partition would be mounted (entries in fstab done by the install DVD). That didn’t help.

I would like to NOT set this system to Legacy mode – but that may be what is required.

I am at a loss. From all I understand, from having done this kind of thing with other distros, It should have set this so it would boot. Even if the HDDs were swapped, GRUB should have picked up the correct partition on the correct drive, right (system is using UUID)?

Any idea what option I should be specifying that I’m not to get this to setup?

If you need screen captures, I’m not sure that pictures of the monitor(s) would be readable when posted here. <<

Thanks in advance.

Hi
So when you boot from the knoppix system, can you check the system BIOS nvram contents with;

efibootmgr -v

You should have a small partition /boot/efi of type ef00 ~260MB with the efi files in it?

You may need to use the efibootmgr command to re-implement the booting…

Anyway, you may have fun with the old kernel and the ryzen cpu, they are still tweaking the 4.18+ kernels for performance…

[QUOTE=malcolmlewis;55179]Hi
So when you boot from the knoppix system, can you check the system BIOS nvram contents with;

efibootmgr -v

You should have a small partition /boot/efi of type ef00 ~260MB with the efi files in it?

You may need to use the efibootmgr command to re-implement the booting…

Anyway, you may have fun with the old kernel and the ryzen cpu, they are still tweaking the 4.18+ kernels for performance…[/QUOTE]

The problem here is, I don’t recall ordering this level of AMD processor. I thought I ordered a lower one with graphics. I like to stay away from the bleeding edge unless I’m part of a group creating it. :wink:

Now I find that Knoppix 7.6.1 doesn’t have the efibootmgr command. So I am now trying to get 8.1 downloaded and get a USB thumbdrive bootable (16GB). Hopefully it will have it.

Meanwhile, In running Knoppix 7.6.1 doing a backup of the /home (raid 0 mirroring) it got lost somewhere so after letting it run all night, it had backed up 1GB of 140GB. I was wanting this backup so I could just blow the whole system and start over. Restoring the /home partition would get my wife’s system back to a usable state for her.

According to the partitioners that have looked at my HDDs, I do have the efi partition (about 500MB) and the drives were made all using a GUID type “partitioner”.

I guess the bottom line at this point is, would/does SLED 15 have all the right stuff with a kernel that will work with this ryzen CPU? I really have no problem going to SLED 15 as long as I can get V-Box working.

Hi
You could always try an openSUSE Live Tumbleweed USB to check the efi
side?
https://software.opensuse.org/distributions/tumbleweed

What model ryzen cpu? I think you will have better support in the 4.12
kernel range.

Don’t envisage issues with V-Box except the usual module rebuilding
gotchas at times. Perhaps consider KVM/qemu as an alternative?


Cheers Malcolm °¿° SUSE Knowledge Partner (Linux Counter #276890)
SLES 15 | GNOME Shell 3.26.2 | 4.12.14-25.25-default
If you find this post helpful and are logged into the web interface,
please show your appreciation and click on the star below… Thanks!

[QUOTE=malcolmlewis;55205]Hi
You could always try an openSUSE Live Tumbleweed USB to check the efi
side?
https://software.opensuse.org/distributions/tumbleweed

What model ryzen cpu? I think you will have better support in the 4.12
kernel range.

Don’t envisage issues with V-Box except the usual module rebuilding
gotchas at times. Perhaps consider KVM/qemu as an alternative?


Cheers Malcolm °¿° SUSE Knowledge Partner (Linux Counter #276890)
SLES 15 | GNOME Shell 3.26.2 | 4.12.14-25.25-default
If you find this post helpful and are logged into the web interface,
please show your appreciation and click on the star below… Thanks![/QUOTE]

Sorry for the delay. Been battling getting this to work.

Ryzen 5 1400 SMT 3.4 GHz

The problem is not the CPU, it is UEFI.

I have done strange things to get this to work on two different systems now. I have to use the install DVD and select “more” and then start linux and then have it find the installed linux and then run it from there on BOTH these systems (one is now running LEAP 15 and the other is running SLED12 SP3).

Wicked is the most aptly named thing in that it is not well behaved. Even on an update it changes the host to “linux”.

Bottom line is, the installer needs some work.

Hi
You need to configure the hostname in YaST → System → Network
Settings, under the Hostname/DNS tab, then against the interface under
the Overview tab, select the interface and edit…

UEFI can be troublesome as it’s a hardware/vendor issue with their
implementation. Is the BIOS firmware the latest?

So what UEFI options to you have in the BIOS; Full UEFI, CSM? Did you
manage to boot from a Tumbleweed Live CD to check the efibootmgr
command?


Cheers Malcolm °¿° SUSE Knowledge Partner (Linux Counter #276890)
SLES 15 | GNOME Shell 3.26.2 | 4.12.14-25.25-default
If you find this post helpful and are logged into the web interface,
please show your appreciation and click on the star below… Thanks!

[QUOTE=malcolmlewis;55398]Hi
You need to configure the hostname in YaST → System → Network
Settings, under the Hostname/DNS tab, then against the interface under
the Overview tab, select the interface and edit…

UEFI can be troublesome as it’s a hardware/vendor issue with their
implementation. Is the BIOS firmware the latest?

So what UEFI options to you have in the BIOS; Full UEFI, CSM? Did you
manage to boot from a Tumbleweed Live CD to check the efibootmgr
command?


Cheers Malcolm °¿° SUSE Knowledge Partner (Linux Counter #276890)
SLES 15 | GNOME Shell 3.26.2 | 4.12.14-25.25-default
If you find this post helpful and are logged into the web interface,
please show your appreciation and click on the star below… Thanks![/QUOTE]

Malcolm:

Please don’t assume that I don’t know how to fix the hostname during install or update. This is why I am complaining about Wicked – It is an aptly named misbehaved product. Better would be that the installer gets fixed so that during UPDATE it goes and gets the information from the system being updated, rather than arbitrarily changing the hostname. I have had to go delete several instances in the SCC.SUSE.COM pages relative to my subscriptions because of this behavior.

I have gotten the system that is running SLED12 SP3 to respond to “efibootmgr -v” from the running SLED12 SP3!
This is what it returned:

efibootmgr -v

[HTML]BootCurrent: 0003
Timeout: 1 seconds
BootOrder: 0000,0003,0001,0002
Boot0000* sled-secureboot HD(1,GPT,6b27d312-1aff-4e73-bab3-4e4c0ea7ce8e,0x800,0xfa000)/File(\EFI\sled\shim.efi)
Boot0001* Hard Drive BBS(HD,0x0)…GO…NO…u.W.D.C. .W.D.1.0.E.Z.E.X.-.0.0.B.N.5.A.0…A…>…Gd-.;.A…MQ…L. . . . .W. .-.D.C.W.3.C.5.F.H.C.9.7.J.D…BO…NO…u.W.D.C. .W.D.1.0.E.Z.E.X.-.0.0.B.N.5.A.0…A…>…Gd-.;.A…MQ…L. . . . .W. .-.D.C.W.3.C.1.F.Y.Y.H.L.2.S…BO…NO…u.K.i.n.g.F.a.s.t…A…>…Gd-.;.A…MQ…L.3.0.1.2.0.2.5.1.3.1.6.8. .8. . . . . . …BO
Boot0002* CD/DVD Drive BBS(CDROM,0x0)…GO…NO…u.M.e.m.o.r.e.x. .M.R.X.-.5.5.0.L. .v.4…A…>…Gd-.;.A…MQ…L.S.C.J.A.4.5.1.6.0.1.2.2.9.7. . . . . . …BO
Boot0003* UEFI: Memorex MRX-550L v4 PciRoot(0x0)/Pci(0x1,0x3)/Pci(0x0,0x1)/Sata(5,65535,0)/CDROM(1,0x29e,0x219c96)…BO
[/HTML]

And the above system has CSM.

If I boot this system using the SLED 12 SP3 install CD and send it to “more” and Load Linux, then it can get the installed SLED12 SP3 system running.

My temp desktop system is now running SLED 15. And last boot it appeared (I wasn’t there to see it all) that it booted directly into SLED15, but it may have gone through the SLED15 installer CD.

Bottom line at this point, I have usable systems (SLED13 SP3 is my wife’s workstation, SLED15 is my temporary workstation).

[QUOTE=wylbur;55430]Malcolm:

Please don’t assume that I don’t know how to fix the hostname during install or update. This is why I am complaining about Wicked – It is an aptly named misbehaved product. Better would be that the installer gets fixed so that during UPDATE it goes and gets the information from the system being updated, rather than arbitrarily changing the hostname. I have had to go delete several instances in the SCC.SUSE.COM pages relative to my subscriptions because of this behavior.

I have gotten the system that is running SLED12 SP3 to respond to “efibootmgr -v” from the running SLED12 SP3!
This is what it returned:

efibootmgr -v

[HTML]BootCurrent: 0003
Timeout: 1 seconds
BootOrder: 0000,0003,0001,0002
Boot0000* sled-secureboot HD(1,GPT,6b27d312-1aff-4e73-bab3-4e4c0ea7ce8e,0x800,0xfa000)/File(\EFI\sled\shim.efi)
Boot0001* Hard Drive BBS(HD,0x0)…GO…NO…u.W.D.C. .W.D.1.0.E.Z.E.X.-.0.0.B.N.5.A.0…A…>…Gd-.;.A…MQ…L. . . . .W. .-.D.C.W.3.C.5.F.H.C.9.7.J.D…BO…NO…u.W.D.C. .W.D.1.0.E.Z.E.X.-.0.0.B.N.5.A.0…A…>…Gd-.;.A…MQ…L. . . . .W. .-.D.C.W.3.C.1.F.Y.Y.H.L.2.S…BO…NO…u.K.i.n.g.F.a.s.t…A…>…Gd-.;.A…MQ…L.3.0.1.2.0.2.5.1.3.1.6.8. .8. . . . . . …BO
Boot0002* CD/DVD Drive BBS(CDROM,0x0)…GO…NO…u.M.e.m.o.r.e.x. .M.R.X.-.5.5.0.L. .v.4…A…>…Gd-.;.A…MQ…L.S.C.J.A.4.5.1.6.0.1.2.2.9.7. . . . . . …BO
Boot0003* UEFI: Memorex MRX-550L v4 PciRoot(0x0)/Pci(0x1,0x3)/Pci(0x0,0x1)/Sata(5,65535,0)/CDROM(1,0x29e,0x219c96)…BO
[/HTML]

And the above system has CSM.

If I boot this system using the SLED 12 SP3 install CD and send it to “more” and Load Linux, then it can get the installed SLED12 SP3 system running.

My temp desktop system is now running SLED 15. And last boot it appeared (I wasn’t there to see it all) that it booted directly into SLED15, but it may have gone through the SLED15 installer CD.

Bottom line at this point, I have usable systems (SLED13 SP3 is my wife’s workstation, SLED15 is my temporary workstation).[/QUOTE]
Hi
Might be worth a Support Request for the installer…? I’ve only done fresh installs of SLE 15 on SCC it sets hostname as installer, then after I change the host name on next check-in it updates. It can also be checked/changed at the SCC registration page (Top Right - Network button).

CSM can be a bit funky on different systems (I’ve seen this on HP ProBook 44xx series). I’m guessing the BIOS is actually using the HDD entry (expects Windows), 0001, even though the default says 0000.

You can verify, if you look down in /boot/efi/EFI is there a ‘Boot’ folder? If not create one and then copy over shim.efi as bootx64.efi and see if it comes up. The other option is to check the BIOS and can you change the order here?

Another option is to add a new efi entry and make it thinks your booting Windows;

efibootmgr -c -L "Windows Boot Manager" -l "\\\\EFI\\\\Boot\\\\bootx64.efi"

Or try a /boot/efi/EFI/Microsoft/Boot/bootmgfw.efi (which is a copy of shim.efi) and modify the above.

I have on ASUS K55A laptop and it played fine with SLED 15 and dual booting.