Linus Torvalds 不满 Linux Kernel 4.20 出现性能下降


Linus Torvalds 不满 Linux Kernel 4.20 出现性能下降

之前我们曾报道过,Phoronix 近日测试发现,在几乎所有低中高端英特尔处理器上,最新测试版内核 Linux 4.20 的表现比最新稳定版 4.19 有显著的下降,而 AMD 的处理器不受影响。出现性能下降的原因与英特尔处理器漏洞修补有关,4.20 的主要改变是对英特尔支持超线程的处理器默认启用了 STIBP,防止间接分支预测器的跨线程控制。

针对此事,Linus Torvalds 感到惊讶,他在 邮件列表
中表示自己在讨论列表中居然没有看到任何关于会影响性能的字眼。当在某些负载下性能下降 50% 时,应该先反问是否值得,而不是不管不顾。

This was marked for stable, and honestly, nowhere in the discussion did I see any mention of just *how* bad the performance impact of this was.  When performance goes down by 50% on some loads, people need to start asking themselves whether it was worth it. It's apparently better to just disable SMT entirely, which is what security-conscious people do anyway.  So why do that STIBP slow-down by default when the people who *really* care already disabled SMT?  I think we should use the same logic as for L1TF: we default to something that doesn't kill performance. Warn once about it, and let the crazy people say "I'd rather take a 50% performance hit than worry about a theoretical issue".
—— Linus

他补充道,“我不认为代码需要还原,但是无条件启用 STIBP 的‘行为’需要更正。因为这付出的代价显然比你们之前告知的更大。 ”

来自 Intel 的 Linux 资深人士 Arjan van de Ven 也表示,“在文档中,AMD 官方已有建议不要默认这样做,我可以代表 Intel 说我们的立场也是如此。”

CLP(FD) Constraint Logic Programming over Finite Domains





Linus Torvalds 不满 Linux Kernel 4.20 出现性能下降