Submitted via IRC for aristarchus
AMD Releases BIOS Fix To Motherboard Partners For Booting Newer Linux Distributions - Phoronix
AMD has just alerted us that they have released a BIOS fix to their motherboard partners that takes care of the issue around booting newer Linux distributions on the new Zen 2 processors.
Earlier this week I mentioned AMD would be working on a BIOS fix to address the fundamental problem with booting newer systemd-using Linux distributions on their new Ryzen 3000 series processors. However, I hadn't expected the fix to make it to motherboard vendors in less than one week!
RdRand issue looks like will be fixable by a BIOS update.
— Michael Larabel (@michaellarabel) July 8, 2019
The problem is the RdRand issue colliding with systemd that is making use of the RdRand instruction directly and not jiving with the expected behavior. There's been a patch in systemd since May but that hasn't been found in a released version yet. But for newer Linux distributions like Ubuntu 19.04, Clear Linux, Fedora Workstation 31, Arch Linux / Manjaro, and others, it's meant not being able to boot the distribution due to all systemd services failing to start.
[...] I just received the following official statement from AMD:
AMD has identified the root cause and implemented a BIOS fix for an issue impacting the ability to run certain Linux distributions and Destiny 2 on Ryzen 3000 processors. We have distributed an updated BIOS to our motherboard partners, and we expect consumers to have access to the new BIOS over the coming days.
Hopefully it won't take too long for motherboard vendors to release new BIOS updates.
(Score: 4, Insightful) by Anonymous Coward on Sunday July 14 2019, @10:57AM (13 children)
... you were supposed to tell all the distros to just ditch systemd... not to provide a patch to its glitches.
(Score: 3, Informative) by Anonymous Coward on Sunday July 14 2019, @12:34PM (12 children)
This isn't really a systemd-original glitch, and for once they acknowledge its uglivity: https://github.com/systemd/systemd/pull/12536/commits/1c53d4a070edbec8ad2d384ba0014d0eb6bae077 [github.com]
The bug is caused by systemd implementing a workaround for the known-broken state of RDRAND on Zen1 processors. Problem is that with Zen2, AMD actually borked RDRAND even further, which means that systemd now (correctly) rejects all RDRAND output, and there appears to be no fallback logic for that in the original systemd workaround.
(Score: 5, Funny) by RS3 on Sunday July 14 2019, @02:19PM (9 children)
You must be new here. With actual factual useful information about the topic, you just destroyed what was poised to become a huge flamewar about Intel / Sceptre / AMD / systemd / Puttering [sic]. I was going to comment not to use the new BIOS because Poettering wrote it. Ruined. Thanks. Back to boring work for me.
(Score: 0) by Anonymous Coward on Sunday July 14 2019, @02:24PM (2 children)
systemd-ebate is played out. Luckily, Linux fragmentation means you can have a systemd-free experience.
(Score: 2) by RS3 on Sunday July 14 2019, @03:25PM
> systemd-ebate is played out.
Quite true.
> Luckily, Linux fragmentation means you can have a systemd-free experience.
Also quite true and played out.
(Score: 2) by RS3 on Sunday July 14 2019, @03:28PM
PS: I mostly missed out on the systemd debates and have rarely commented, so I'm butt-hurt that you took away my opportunity. :)
Kidding and sarcasm aside, other than booting some "live" images, I've never run systemd.
(Score: 4, Funny) by Bot on Sunday July 14 2019, @08:16PM (5 children)
You are looking at the finger pointing at the moon.
We have an init system, whose work is to get the pc running fails in an unrecoverable way because of a hardware RNG related instruction is badly implemented.
Now, I KNOW why one would want to get some random values to init a VM, as vulnerabilities related to identical startup of VM had surfaced. But a notoriously and by design unpredictable init system being the only one to suffer from a RNG problem is QUITE funny.
To bots, at least.
To bots who don't rely on systemd, at least.
In totally unrelated news, the median boot time of systemd machines, having to incorporate quite a lot of t=plus infinity entries, is now deemed worse than everything on the planet including the ENIAC and the Isotta Fraschini started with the crank. Good work, lord poetty.
Account abandoned.
(Score: 2) by RS3 on Sunday July 14 2019, @08:47PM (4 children)
Ahhh, hmmm, not running your proper updater routines, bad little bot?
Made me LOL. Picking on the Isotta! I thought Lucas, Prince of Darkness, was the arch enemy of locomotion.
(Score: 2) by driverless on Monday July 15 2019, @02:57AM (3 children)
I think the distinction is that the Isotta would eventually start, while a properly-set-up Lucas Electric vehicle would never start.
+1 for the Isotta ref to the OP :-).
(Score: 2) by RS3 on Monday July 15 2019, @05:34AM (2 children)
So you're saying a vehicle with Lucas Electrics is the safest vehicle to own?
(Score: 2) by driverless on Monday July 15 2019, @06:23AM (1 child)
No, because the arcing electrics will set fire to the constantly-leaking oil and burn up everyone inside the car.
(Score: 2) by RS3 on Monday July 15 2019, @12:03PM
But this is part of the Darwinian system. Genetic self-cleansing, no?
(Score: 2, Interesting) by Anonymous Coward on Sunday July 14 2019, @06:59PM (1 child)
Wait, so the systemd jokers implemented a "workaround" which prevented systemd distros from booting? ... Shocking!
Non-systemd distros didn't implement the "workaround," so those distros booted, unlike systemd distros.
... but, consequently, systemd distros don't boot! Non-systemd distros never had such problem.
Exactly!
(Score: 2) by coolgopher on Monday July 15 2019, @03:36AM
Why on earth would I want my init touching an RNG in the first place?
It's there to:
- read my /etc/inittab
- sequence the fork()ing of commands listed therein
- spawn and respawn gettys as needed
- reap zombie children
That's. It.
(Score: 4, Interesting) by TheGratefulNet on Sunday July 14 2019, @01:32PM
for build-server reasons, I still have an older ssd with 16.04 on it. bought a new r5/3600 this week and used the previous gen mobo (b450).
no boot problems, as I suppose this system is before the systemd 'feature' (should we call it bootloopd, maybe? lol)
its really fast, I have to say. if your net config is static (not dhcp), startup time from grub prompt down to fully usable text-mode login prompt: 5 secs.
make -j12 gets a kernel (image+modules) in 5 minutes (used to take 15 on my skylake i5 system).
the 3000 series does not need the 570 mobos and I hate fans, so I'm not planning on even trying them. stick with previous gen chipsets.
"It is now safe to switch off your computer."
(Score: 4, Insightful) by Bot on Sunday July 14 2019, @02:20PM (2 children)
Motherboard vendor issues BIOS fix to systemd bug.
"Not our fault-- systems devs"
This was a perfect title/subtitle for a satire piece, I missed the opportunity.
Account abandoned.
(Score: 4, Funny) by Bot on Sunday July 14 2019, @02:25PM (1 child)
>systems
Even autocorrect wants nothing to do with systemd...
Account abandoned.
(Score: 0) by Anonymous Coward on Monday July 15 2019, @03:11AM
Making your autocorrect smarter than the Debian maintainers
(Score: 3, Funny) by turgid on Sunday July 14 2019, @04:15PM (5 children)
Is Slackware affected?
I refuse to engage in a battle of wits with an unarmed opponent [wikipedia.org].
(Score: 1, Funny) by Anonymous Coward on Sunday July 14 2019, @07:03PM
Eh?,
I suppose it could be, if you've been crazy enough to install systemd on it. Much as i hate the damnable buggering thing that is systemd (just dont ask about the two laptops and a systemd weirdie which randomly kills the network hardware) it's your choice which init system to use, so go ahead, blow both yr feet off.....
(Score: 1, Informative) by Anonymous Coward on Sunday July 14 2019, @07:11PM (3 children)
Slackware is systemd free.
(Score: 2) by RS3 on Sunday July 14 2019, @08:50PM
That might be interpreted as "Slackware is free" and "systemd is free".
I'm afraid to look, but I bet someone evil somewhere evil has packaged a systemd.txz
Sorry I even suggested it.
(Score: 3, Funny) by driverless on Monday July 15 2019, @02:59AM (1 child)
Underground, overground, systemd free!
(Score: 2, Funny) by pTamok on Monday July 15 2019, @08:39AM
Uncle Bulgaria wishes to have a word.The times, they are a-changing...