Linux Decides To Disable RNG On All AMD fTPMs

[phoronix]

Phoronix: Linux Decides To Disable RNG On All AMD fTPMs

As a follow-up to the first-on-Phoronix article last month that highlighted Linus Torvalds' frustrated views on the AMD fTPM random number generator continuing to cause problems for users even with updated firmware/BIOS, as of today the Linux kernel has gone ahead and blanket disabled RNG use for all current AMD fTPMs...

https://www.phoronix.com/news/Linux-Disables-RNG-AMD-fTPMs

[RejectModernity]

I love getting my PC's functions disabled!

[tildearrow]

I love getting my PC's functions disabled!

What? System freezes, even if short, are intolerable. They ruin audio, video and other tasks which require precision.

[Jumbotron]

I used my trusty Commodore 64 from 1982 until 1990 when I bought my first PC in the form of a 12 pound laptop with an AMD 286 CPU that I later hot rodded with a plug in the socket Cyrix Math Co-Processor. I've been Team Green ever since

When my desktop and laptop, both of which have Bristol Ridge APUs, which was the pinnacle of the Bulldozer/Steamroller/Piledriver/Carrizo HSA era, I'm done with AMD. It's not enough to have a great engine when your chassis and suspension suck. When it comes to AMD the engine (CPU) is great but their chassis and suspension ( dev software/firmware and OEM BIOS and OEM packinging and retail offerings ) sucks.

I know it stems from long ago with Intel and all their shenanigans that they were hauled into court over. And I know Lisa Su basically stripped the last 20 years before her off the table and rebuilt the company.

But CPUs are bricks without good dev software and firmware. Plus the enforcement of feature complete and non broken BIOS in their partners laptops and desktops.

[Paradigm Shifter]

What? System freezes, even if short, are untolerable. They ruin audio, video and other tasks which require precision.

Yup. They just ruin everything.

Windows 10 had a similar problem with some hardware - I think it was updating from 1607 to 1703 where Microsoft tweaked the SATA controller driver for the Intel laptop chipset at the time (I forget), and this resulted in it not playing nicely with some SSD controllers (e.g.: whatever Sandisk were using at the time). This resulted in my mothers laptop having regular soft locks as the SATA controller on the motherboard and the controller on the SSD would spend 20+ seconds renegotiating after every 3 to 5 seconds of actually being able to use it. Installing the Intel driver fixed this, but was an exercise in extreme frustration as the 20+ second soft locks even happened in safe mode. Getting that driver installed was an experience I do not wish to repeat in a hurry.

[RejectModernity]

What? System freezes, even if short, are untolerable. They ruin audio, video and other tasks which require precision.

Then fix your shit. Turning off functions ain't no "fix".

[gbcox]

My concern about this is apparently when something like this is done in the kernel, it is permanent. I just encountered an issue with Seagate USB drives and Smartmontools. Apparently, back in the day some USB Seagate drives were having issues (see attached link), however seagate drives haven't had this issue for years. Technology changes all the time and it seems like there should be a better approach than doing a permanent blacklist for all AMD chips, regardless if they have the issue or not.

[emansom]

Then fix your shit. Turning off functions ain't no "fix".

Sadly microcode is not open-source.

[NotMine999]

Then fix your shit. Turning off functions ain't no "fix".

Ever think that it has been turned off "as an interim fix" while AMD and other devs quietly work on a solution in the background?

Nah! That would be too easy. It's much more fun to go off on a public rant.

[Democrab]

I used my trusty Commodore 64 from 1982 until 1990 when I bought my first PC in the form of a 12 pound laptop with an AMD 286 CPU that I later hot rodded with a plug in the socket Cyrix Math Co-Processor. I've been Team Green ever since

When my desktop and laptop, both of which have Bristol Ridge APUs, which was the pinnacle of the Bulldozer/Steamroller/Piledriver/Carrizo HSA era, I'm done with AMD. It's not enough to have a great engine when your chassis and suspension suck. When it comes to AMD the engine (CPU) is great but their chassis and suspension ( dev software/firmware and OEM BIOS and OEM packinging and retail offerings ) sucks.

I know it stems from long ago with Intel and all their shenanigans that they were hauled into court over. And I know Lisa Su basically stripped the last 20 years before her off the table and rebuilt the company.

But CPUs are bricks without good dev software and firmware. Plus the enforcement of feature complete and non broken BIOS in their partners laptops and desktops.

I'm sure you'll be back on AMD before long if this kind of bug is all it's taken to get you to swear them off after ~30 years, especially given the main competition for AMD isn't really that much better in the "low level-polish" type of areas you're talking about here. Even the simple fact that CPU schedulers still aren't able to handle the P+E core configuration particularly well when Intel is on its second generation of parts using it, or the state of their GPUs and I'm not just meaning the currently well-known Arc driver issues there. Heck, there's a bunch of Intel chips which have lost potentially significant amounts of performance due to SPECTRE/Meltdown workarounds.

Take it from someone whose running all kinds of systems from more than just the two major companies dating from a modern zen4 vcache chip in my main desktop all the way back to an AMD 5x86-133 and Pentium MMX 166 when you consider the various retro PCs I own: Neither company is better than the other overall, sometimes one will have a clearly better platform or product but more often than not you're just picking which set of trade-offs you prefer.