Linux Landing Change To Allow STIBP When Using Legacy IBRS

[mulenmar]

If I were running the world, at this point I'd simply say "[CENSORED] this dumpster fire!" make every OS drop support for this defective hardware entirely, and make Intel go spend their billions on developing a modern no-guessing no-CISC-decode no-cruft CPU which depends on compilers and JIT interpreters to optimize the code instead.

Like Itanium, except without the design choices which made it impossible to write said compilers for. And without trying to contort C onto it, use something which makes concurrency correctness easy.

Because clearly, this branch prediction tech is too hard to get correct.

Not like existing OS versions would stop existing, so whiners can just go put that argument right back in their pockets and let me have my fantasy, lol

[JEBjames]

Michael

Typo "now be ena bled while" should be "now be enabled while" (remove extra space)

[phoronix]

Phoronix: Linux Landing Change To Allow STIBP When Using Legacy IBRS

Ahead of the Linux 6.3-rc1 release later today, a set of "x86/urgent" patches were sent out Sunday morning that include the change to allow Single Threaded Indirect Branch Predictors (STIBP) to be used in the presence of legacy Indirect Branch Restricted Speculation (IBRS) for security reasons...

https://www.phoronix.com/news/Linux-6.3-STIBP-Legacy-IBRS