Intel + Microsoft Continue Work On Replacing More SMM "Black Boxes" With PRM

[phoronix]

Phoronix: Intel + Microsoft Continue Work On Replacing More SMM "Black Boxes" With PRM

Given all the headaches and concerns from the early days of UEFI SecureBoot, for longtime Linux users hearing Microsoft is working on another firmware-level standard in the name of security may raise concerns... Microsoft in conjunction with Intel has been spearheading the Platform Runtime Mechanism (PRM) that is about moving more code out of the System Management Mode (SMM) and executing it within the OS/VMM context. PRM remains a work-in-progress but the Windows support is already ready within Windows Insiders builds while the Linux support will come after the ACPI specification around it has been finalized...

http://www.phoronix.com/scan.php?page=news_item&px=Microsoft-PRM-OSFC-2020

[oiaohm]

Please fix the slide deck link it points to a local file instead of the website.

[uid313]

But doesn't this break backward compatibility with x86?
Isn't SMM a integral part of x86?

Since Apple's new M1 CPU seems to execute emulated x86 code faster than Intel and AMD executes native x86 code, maybe it is time for Intel and AMD just to ditch x86 and build a ARM CPU or a RISC-V CPU with emulation for x86?

[Michael]

Please fix the slide deck link it points to a local file instead of the website.

Fixed now, thanks.

[Michael]

But doesn't this break backward compatibility with x86?
Isn't SMM a integral part of x86?

As said in the article, SMM will still be around for things that need it.

[Charlie68]

Why does Linux support always come later? Why Intel, which always seems so interested in GNU / Linux, keeps considering GNU / Linux as B-series, because Intel® RST doesn't have any GNU / Linux support yet, why Intel?

[gabber]

But doesn't this break backward compatibility with x86?
Isn't SMM a integral part of x86?

Since Apple's new M1 CPU seems to execute emulated x86 code faster than Intel and AMD executes native x86 code, maybe it is time for Intel and AMD just to ditch x86 and build a ARM CPU or a RISC-V CPU with emulation for x86?

Geekbench. Last time I checked this is not used for any productive work. Also M1 has a hard max RAM of 16 GB.

Not saying the M1 is pointless, but it's not an "x86-Killer" as the church of apple wants you to believe.

[Danny3]

WTF is this?
Another vendor-locked crap from Microsoft?

[milkylainen]

But doesn't this break backward compatibility with x86?
Isn't SMM a integral part of x86?

Since Apple's new M1 CPU seems to execute emulated x86 code faster than Intel and AMD executes native x86 code, maybe it is time for Intel and AMD just to ditch x86 and build a ARM CPU or a RISC-V CPU with emulation for x86?

It doesn't. There is a difference between rewritten gadgets in native code and hot optimizing that code to native than "emulating".
Technically. I would not even call Rosetta 2 an emulator or an instruction simulator. More like a translator than anything else.
Not that the average user would care. But I think it's more a testament to the technology used for Rosetta 2 than the actual M1 CPU.
There are also a lot of stuff Rosetta 2 cannot do. Maybe in the future.

A more correct statement would be that Rosetta 2 is better at generating translated native code for the host and sometimes it does indeed execute faster than native x86 (of course, depending on what you compare it to).

[oiaohm]

I do hope these changes are able to reduce the performance jitter that comes from SMI stuff firing off in background.