[SOLVED] About slackware and NVME SSD.

bassmadrigal · 10-04-2017, 07:56 AM

From the many posts on this forum that I've seen about UEFI, I will say that based on some of the results of those posts, there are just some buggy UEFI implementations out there. The first thing I would check is to see is if there is a "BIOS" update (not sure if they still call them that or moved to UEFI update) for the motherboard. That could fix some issues. Others ended up using rEFInd to work around some of those bugs.

On my NVMe system, I was able to use the eliloconfig from -current to set up everything and then I modified elilo.conf to my particular taste. Unfortunately, I didn't run into any of the issues you had, so I have not played with efibootmgr at all.

brobr · 10-04-2017, 09:34 AM

Thanks for the clear description of the problem, and yes it looks weird. Refind could be a try although it appears to use the efibootmgr that is on the system to edit the bootloader-menu, just as you would do manually. It might do something with 'signed' keys, ubuntu does as well I think, that could be needed to get things changed; but best to look at the refind install script and its documentation.

Cheers

Rob

sungjin · 10-06-2017, 04:57 AM

Yes. I think the problem might arise from:

some buggy UEFI implemantation, (and/or)
'signed keys' which Slackware lacks.

However since I guess we have a degree of agreement about this matter, I think it's sufficient to close this subject at this time. (And I have a working system now).

Again, thank your for the help you gave. I learned a lot. I really appreciate them.

Best regards,
sungjin

ps. BTW, for the completeness, unfortunately, this mainboard's firmware is up-to-date (released 2017-07-03).

Didier Spaier · 10-06-2017, 05:20 AM

Let me just add that as a fallback you can include in the main directory of the ESP a small EFI shell script named startup.nsh with just this line: (assuming that EFI is the ESP itself and it contains a partition named Slacwkare including the elilo.efi boot loader):

Code:

EFI\Slackware\elilo

Then if your firmware includes an EFI shell (I assume that most recent ones do) as last resort it will execute startup.nsh if found in an ESP, hence in this case start elilo. That's what we do in Slint and that works. The advantage is that even if there is no boot entry for your system in the firmware's boot menu it will be picked up.

Cf: the UEFI Shell specification (version 2.2 at time of writing):
http://www.uefi.org/sites/default/fi..._Shell_2_2.pdf

All UEFI specifications are available from this page:
http://www.uefi.org/specifications

sungjin · 10-06-2017, 07:18 AM

Dear Didier,

Your last post makes me be in agony.

On one hand, I have now a working system, but on the other hand, I know I have the responsibility to report anything about this subject.

OK. I'll try it, since I have a few free hours, finished my daily work.

Let's go step-by-step.

I ssh to the machine at my lab. This is current configuration.

Code:

root@1080ti:~# efibootmgr 
BootCurrent: 0003
Timeout: 1 seconds
BootOrder: 0003,0001
Boot0001  Hard Drive
Boot0003* UEFI OS

And the ESP directory.

Code:

root@1080ti:~# ls /boot/efi/EFI/
BOOT/
root@1080ti:~# ls /boot/efi/EFI/BOOT/
bootx64.efi*  elilo.conf*  vmlinuz*

Then what we will do are;

Remove ./EFI/BOOT/
Create ./EFI/Slackware/
create a startup.nsh

I'll merge the first and second step into one.

Code:

root@1080ti:/boot/efi/EFI# ls
BOOT/
root@1080ti:/boot/efi/EFI# mv BOOT/ Slackware 
root@1080ti:/boot/efi/EFI# cd Slackware/
root@1080ti:/boot/efi/EFI/Slackware# ls
bootx64.efi*  elilo.conf*  vmlinuz*
root@1080ti:/boot/efi/EFI/Slackware# mv bootx64.efi  elilo.efi
root@1080ti:/boot/efi/EFI/Slackware# ls
elilo.conf*  elilo.efi*  vmlinuz*

And finally make a startup.nsh to the root of ESP.

Code:

root@1080ti:/boot/efi# ls
EFI/
root@1080ti:/boot/efi# cat << EOF > startup.nsh
> EFI\Slackware\elilo
> EOF
root@1080ti:/boot/efi# cat startup.nsh 
EFI\Slackware\elilo

so that now I have

Code:

root@1080ti:/boot/efi# ls
EFI/  startup.nsh*

and

Code:

root@1080ti:/boot/efi# efibootmgr 
BootCurrent: 0003
Timeout: 1 seconds
BootOrder: 0003,0001
Boot0001  Hard Drive
Boot0003* UEFI OS
root@1080ti:/boot/efi#

Now, in theory it should be able to boot normally. Right? I reboot the machine and cross my fingers.

Code:

root@1080ti:/boot/efi# shutdown -r now

Oops. It does not boot. I'll go to my lab and fix it. What a moment.

OK, I came back from my lab.

So I insist that the startup.nsh method does not work in my case.

And beside what I have done to recover this system are:

boot from Slackware64-14.2 udb
mount ESP
mv ESP/EFI/Slackware to ESP/EFI/BOOT
rename elilo.efi to bootx64.efi

So we finally have

Code:

root@1080ti:~# efibootmgr 
BootCurrent: 0004
Timeout: 1 seconds
BootOrder: 0004,0001
Boot0001  Hard Drive
Boot0004* UEFI OS

And I think this is another evidence that this BIOS does not properly handle the UEFI entry. What do you think about?

Thank you for your kind help.

Regards,
sungjin

Didier Spaier · 10-06-2017, 07:35 AM

I don't know. Maybr you don't have an EFI shell, or your firmware doesn't implement the specification properly

sungjin · 10-06-2017, 07:53 AM

Quote:

Originally Posted by Didier Spaier

I don't know. Maybr you don't have an EFI shell, or your firmware doesn't implement the specification properly

Thank you for you concern. Sorry to bother you. I think next time I'll buy a mainboard, I would rather skip the MSI mainboard, so that to get rid of this weird BIOS phenomenon.

However, I really want to tell you helped me a lot. Thank you for your kind help.

Best regards,
sungjin

brobr · 10-06-2017, 07:54 AM

Sungjin: at least you got it booting with linux; you might have seen these, but it was not trivial (in February) on windows either: http://www.tomshardware.co.uk/answer...-nvme-960.html.

What is the make of your NVME SSD? Could be a relevant factor too: https://forum-en.msi.com/index.php?topic=281664.0 and specific instructions could be needed for setting up the bios, see for example, intel..nvme_boot_guide

sungjin · 10-06-2017, 08:16 AM

Fist, please pardon me that I can answer at now very few of the questions you asked. I'll append it later.

Quote:

Originally Posted by brobr

Sungjin:

What is the make of your NVME SSD?

answer:

http://www.xpg.com/en/feature/423

Code:

root@1080ti:~# sdparm  -i  /dev/nvme0n1
    /dev/nvme0n1: NVMe      ADATA SX8000NP    .2.1
Device identification VPD page:
  Addressed logical unit:
    designator type: SCSI name string,  code set: UTF-8
      SCSI name string:
      126fADATA SX8000NP                          10002H1720050704

I'll read the link you suggested but it may take some time.

kjhambrick · 10-06-2017, 09:49 AM

Thank you sungjin for all the extra work you've done to debug this NVME boot issue.

I imagine it may be very helpful for someone else in the future.

One Q I have, in addition to the Make / Model of your NVME Card that you already posted above is ...

What is the DMI: entry for your MoBo in your `dmesg` output.

Mine is below.

Thanks again !

-- kjh

The DMI: entry for my Sager NP9778-S Laptop is this ( I edited out a LOT of extra spaces ):

Code:

# dmesg |grep DMI:

[    0.000000] DMI: Notebook   P7xxDM(-G)   /P7xxDM(-G), BIOS 1.05.03LS1 09/23/2015

brobr · 10-06-2017, 11:07 AM

Hi kjhambrick, what would that tell, apart from model no and bios? Mine, a rebranded clevo N750HU, gives:

Code:

bash-4.4# dmesg | grep DMI
[    0.000000] DMI: Notebook   N750HU   /N750HU, BIOS 1.05.01 05/16/2017

lspci -v tells more about the nvme card:

Code:

3c:00.0 Non-Volatile memory controller: Samsung Electronics Co Ltd NVMe SSD Controller SM961/PM961 (prog-if 02 [NVM Express])
	Subsystem: Samsung Electronics Co Ltd NVMe SSD Controller SM961/PM961
	Flags: bus master, fast devsel, latency 0, IRQ 16
	Memory at dc000000 (64-bit, non-prefetchable) [size=16K]
	Capabilities: [40] Power Management version 3
	Capabilities: [50] MSI: Enable- Count=1/32 Maskable- 64bit+
	Capabilities: [70] Express Endpoint, MSI 00
	Capabilities: [b0] MSI-X: Enable+ Count=8 Masked-
	Capabilities: [100] Advanced Error Reporting
	Capabilities: [148] Device Serial Number 00-00-00-00-00-00-00-00
	Capabilities: [158] Power Budgeting <?>
	Capabilities: [168] #19
	Capabilities: [188] Latency Tolerance Reporting
	Capabilities: [190] L1 PM Substates
	Kernel driver in use: nvme
	Kernel modules: nvme

@sungjin: duckduckgo spits this out when searching for "ADATA SX8000NP", could be your card: adata-sx8000np-m-2-512-gb and adata.com..423

kjhambrick · 10-06-2017, 12:01 PM

brobr --

Sometimes on a with a Machine built from components, The DMI String will give some info about the MoBo in addition to the BIOS Version.

-- kjh

brobr · 10-06-2017, 12:33 PM

Ah, thanks; (I assume that in that case one of the entries that looks in our case duplicated (apart from the /) will be different).

kjhambrick · 10-06-2017, 03:27 PM

brobr --

Yes, exactly. My Sager Laptop is also a rebranded Clevo so you and I see much the same DMI string.

-- kjh

OTOH, this is what I see on a CentOS 6 NUC Box on the bench:

Code:

# dmesg |grep DMI:

DMI:   /NUC6CAYB, BIOS AYAPLCEL.86A.0040.2017.0619.1722 06/19/2017

This is a ZOTAC ZBOX BI323 on CentOS 6:

Code:

# dmesg |grep DMI:

DMI: ZOTAC ZBOX-ID88/ID89/ID90/ZBOX-ID88/ID89/ID90, BIOS 4.6.5 07/17/2013

And this is an HP Proliant ML10 v2 running Slackware 14.2:

Code:

# dmesg |grep DMI:

[    0.000000] DMI: HP ProLiant ML10 v2, BIOS J10 02/02/2015

sungjin · 10-06-2017, 10:54 PM

DMI info:

Code:

root@1080ti:~# dmesg | grep DMI
[    0.000000] DMI: MSI MS-7A71/Z270-A PRO (MS-7A71), BIOS 1.30 06/28/2017

and the result of lspci:

Code:

04:00.0 Non-Volatile memory controller: Silicon Motion, Inc. Device 2260 (rev 03) (prog-if 02 [NVM Express])
	Subsystem: Silicon Motion, Inc. Device 2260
	Flags: bus master, fast devsel, latency 0, IRQ 16, NUMA node 0
	Memory at df100000 (64-bit, non-prefetchable) [size=16K]
	Capabilities: [40] Power Management version 3
	Capabilities: [70] Express Endpoint, MSI 00
	Capabilities: [b0] MSI-X: Enable+ Count=16 Masked-
	Capabilities: [100] Advanced Error Reporting
	Capabilities: [158] #19
	Capabilities: [178] Latency Tolerance Reporting
	Capabilities: [180] L1 PM Substates
	Kernel driver in use: nvme