9
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 08 Sep 2026
9 points (58.8% liked)
retrocomputing
5459 readers
1 users here now
Discussions on vintage and retrocomputing
founded 3 years ago
MODERATORS
By spychip are you referring to IME and PSP? Chips that old are going to have security holes larger than a CEO's ego. Those chips will be vulnerable to Spectre and Meltdown. Iirc from the fallout of those CVEs, only AMD bothered to patch chips that went back to 2011
“Meltdown affects Intel x86 microprocessors, IBM Power microprocessors,[1] and some ARM-based microprocessors”
I don’t see AMD on that list. Do you?
“At the time of disclosure (2018), this included all devices running any but the most recent and patched versions of iOS,[5] Linux,[6][7] macOS,[5] or Windows.”
So this vuln can be fixed by patching OS kernels, and your reaction is to switch to a CPU that runs embedded closed-source software controlled by a corporate third party who decides what is authorized to execute on your own system? You can fix the problem with or without being nannied. I’ll take the latter.
Spectre affected “All pre-2019 microprocessors”. So no, the spy chip did not protect you. Intel injected the spychip from 2008 forward and AMD did it from 2013 forward.
More generally, it’s not smart infosec to introduce extra complexity. It’s profoundly naive to stick a big attack surface in the core of your CPU. I think it’s quite well established that vulns exploit defects, and defects are proportional to complexity. Signing up for a closed source blob in the core of your processor is adding complexity that’s far from wise.
The /chase the shiny/ mentality (of “latest is greatest!”) neglects the fact that you sign up for the worst kind of vulns -- the unknown variety. With old gear the vulns are more of the known variety, which you have a fighting chance of controlling for.
Lashing out at me was unnecessary. I was pointing out that there are 2 catastrophic vulnerabilities that are guaranteed to be present in every chip that meets your criteria.
I'm not defending the IME or the PSP. The most generous thing I can say about them is that they are unnecessary. I'm pointing out that the chips that you want to use have been unsupported for so long that they have catastrophic vulnerabilities that have never even been attempted to be patched. There are more catastrophic vulnerabilities than just Spectre and Meltdown.
It was an attack on your bullshit. Not on you personally.
And I was pointing out that you are wrong.
You are advocating for chips that are /more/ vulnerable, not less. You are advocating for chips with a much larger attack surface which brings copious unknown vulns. Overall, you are giving poor advice from an infosec standpoint. It is easier to secure a simple known thing than a complex unknown thing.
I’m listening.
No, you weren't. You were being pedantic. AMD will have Spectre. Intel will have Spectre and Meltdown. 2 vulnerabilities.
No, I wasn't. Stop putting words in my mouth. Unless you know how to write microcode, I have serious doubts that you are capable of successfully patching the vulnerabilities on those chips.
Do whatever you want. I don't care.
If you don’t like the facts, what more is there to say? The facts failed to support your claims. If you will not let the facts shape your world view, then it’s on you to go off and find different facts.
Advocating for only running “supported” chips is inherently contempt for old hardware. (But note you said that apparently without knowing about the 15h.org project).
How are you still failing grasp this? The fix was made. And it was done in the kernel without writing microcode. You’ve been told this already. You did not counter it yet to still failed to absorb it. And now you try recycling defeated arguments. You don’t even have to patch Meltdown on chips unaffected by Meltdown (AMD). The spychip failed to protect from both Meltdown and Spectre.
Patching is not the only way to control for a vuln. I am not going to give you the whole infosec discipline here in this thread. There are many ways to controlling for a vuln apart from patching. Depending on your threat model and use cases, there may be no need to do any control. The 15h.org project is another kind of control. Air gapping is another. Detecting the malicious code is another.
You act like you're the only one that understands information security. I took a security class for my computer science degree too. Productive conversation cannot be had when you strut around arrogantly dismissing everyone that disagrees with you.
Spectre was a microcode patch. Meltdown was a kernel patch, so in firmware. Only AMD bothered to fix chips that fit within your timeframe. The fix has not been made for the chips you want to use.
You aren't going to stop branch prediction with clever tricks. These are hardware level flaws. Patching is the only way to completely mitigate these flaws.
What’s being dismissed is irrelevant facts. You continue (for a 3rd time) to still fail to grasp the fact that Meltdown was not found to affect AMD chips. It’s wholly irrelevant.
Those are two different things. There was a firmware patch. And separately there was a kernel mitigation. You don’t need both.
You mean AMD’s f/w patch was not made. Yet you’ve been told about the 15h.org project. If you absorbed that, then citation needed that Coreboot fails to mitigate. In the absence of Coreboot, the os mitigation was implemented in linux. So you’re pushing a bullshit problem.
I'm aware of coreboot. I was not aware that the project managed to increase support to include configurations for all chips that predate IME and PSP. Last I recall when I looked at coreboot, there were chips that they didn't think could ever be supported, even without nannyware. I wasn't pushing a bullshit problem, I was raising legitimate concerns for problems that I didn't know had solutions. If you don't want people to raise honest, good faith concerns to you then you better be extremely vigilant that you don't step in shit.