Torvalds Blasts "Beyond Stupid" Flushing L1d On Context Switches - Reverts Code For Now

As part of the initial set of changes merged today for Linux 5.8 was the x86/mm material that included the controversial feature of opt-in flushing of the L1 data cache on context switching. Linus Torvalds ended up deciding to revert this functionality as for now at least he views it as crazy. While this feature is opt-in via new prctl options and not enabled by default and done in the name of helping those concerned about snoop assisted data sampling vulnerabilities or cache leakage via side channels and yet to be uncovered CPU vulnerabilities, for the time being Linux creator Linus Torvalds is not convinced. Original:

• Re: [GIT PULL] x86/mm changes for v5.8

  >  - Provide an opt-in (prctl driven) mechanism to flush the L1D cache on context switch.
  >    The goal is to allow tasks that are paranoid due to the recent snoop assisted data
  >    sampling vulnerabilites, to flush their L1D on being switched out.
  
  Am I mis-reading this?
  
  Because it looks to me like this basically exports cache flushing
  instructions to user space, and gives processes a way to just say
  "slow down anybody else I schedule with too".
  
  I don't see a way for a system admin to say "this is stupid, don't do it".
  
  In other words, from what I can tell, this takes the crazy "Intel
  ships buggy CPU's and it causes problems for virtualization" code
  (which I didn't much care about), and turns it into "anybody can opt
  in to this disease, and now it affects even people and CPU's that
  don't need it and configurations where it's completely pointless".
  
  To make matters worse, it has that SW flushing fallback that isn't
  even architectural from what I remember of the last time it was
  discussed, but most certainly will waste a lot of time going through
  the motions that may or may not flush the L1D after all.
  
  I don't want some application to go "Oh, I'm _soo_ special and pretty
  and such a delicate flower, that I want to flush the L1D on every task
  switch, regardless of what CPU I am on, and regardless of whether
  there are errata or not".
  
  Because that app isn't just slowing down itself, it's slowing down others too.
  
  I have a hard time following whether this might all end up being
  predicated on the STIBP static branch conditionals and might thus at
  least be limited only to CPU's that have the problem in the first
  place.
  
  But I ended up unpulling it because I can't figure that out, and the
  explanations in the commits don't clarify (and do imply that it's
  regardless of any other errata, since it's for "undiscovered future
  errata").
  
  Because I don't want a random "I can make the kernel do stupid things"
  flag for people to opt into. I think it needs a double opt-in.
  
  At a _minimum_, SMT being enabled should disable this kind of crazy
  pseudo-security entirely, since it is completely pointless in that
  situation. Scheduling simply isn't a synchronization point with SMT
  on, so saying "sure, I'll flush the L1 at context switch" is beyond
  stupid.
  
  I do not want the kernel to do things that seem to be "beyond stupid".
  
  Because I really think this is just PR and pseudo-security, and I
  think there's a real cost in making people think "oh, I'm so special
  that I should enable this".
  
  I'm more than happy to be educated on why I'm wrong, but for now I'm
  unpulling it for lack of data.
  
  Maybe it never happens on SMT because of all those subtle static
  branch rules, but I'd really like to that to be explained.
  
                      Linus
  

Today in Techrights

• Linux Without Linus
• What It Would Take for Linus Torvalds to Leave Linux Foundation Without the Linux Trademark and Without Linux
• The Past Does Not Go Away, Except From Short-Term Memories
• It’s a Common Mistake and Common Misconception/Error to Treat Microsoft as Just Another ‘Large Company’ (or ‘Big Tech’)
• Lessons of Michael Arrington (About Microsoft)
• IRC Proceedings: Sunday, May 31, 2020
• Links 1/6/2020: Linux 5.7, FOSSlife Born, LibreOffice 7.0 Beta1, Linux Mint 20 Making Early Promises
• Links 1/6/2020: OpenMandriva Lx 4.1 2020.05, Linux Lite 5.0 Release, FreeBSD 11.4 RC2

today's leftovers

• Want A More Secure Computer At The Cost Of Performance? Linux 5.8 Landing L1d Flushing

  For those very concerned about CPU data sampling vulnerabilities, the Linux 5.8 kernel comes with the ability to flush the L1 data cache on each context switch. That's good for security, but will hurt the system performance with all the excess L1 cache flushing. This work stems from a proposal earlier this year to flush the L1d cache on context switches due to recent snoop assisted data sampling vulnerabilites or the cache data leaked via side channels. This work was carried out by an Amazon engineer so presumably there is some interest in offering this functionality in the AWS space.

• AMD Radeon Linux Driver Sees Patches For New "Sienna Cichlid" GPU

  Still digging through the 207 patches for the AMD Radeon Sienna Cichlid, but will update if seeing anything else of note. For the most part it's leveraging the existing Navi code paths but the usual churn surrounding firmware, clock-gating / power management differences, and other modifications in the usual spots for bringing up new hardware. The main code additions primarily pertain to the new DCN3 and VCN3 blocks. Given the timing of these patches, the AMD Sienna Cichlid won't be mainlined until the Linux 5.9 merge window opening in August and then releasing in stable around October. That timeframe at least does point to Sienna Cichlid likely being the "RDNA 2" graphics card launch coming later in the calendar year.

• 2020-06-01 | Linux Headlines

  The Linux kernel packs version 5.7 with exciting additions, version 2.2 of the Foliate eBook reader is out with support for many more formats, and members of the Association of American Publishers sue the Internet Archive over their library lending practices.

• Ubuntu Cinnamon Remix 20.04 LTS overview | Ubuntu, traditionally modern.

  In this video, I am going to show an overview of Ubuntu Cinnamon Remix 20.04 LTS and some of the applications pre-installed.

• SUSE Update Infrastructure Access Through the Data Center

  In Step 2 Toward Enhanced Update Infrastructure Access the time-line for enabling access to the SUSE update infrastructure in the Public Cloud via routing through the data center was announced. As of June 1, 2020 we have started the work necessary to make this possible for all regions in AWS, Azure, and GCE. This marks the beginning of the final phase of a process that started almost 1 year ago with A New Update Infrastructure For The Public Cloud. We expect to have everything completed by no later than the end of June 2020, but will most likely be much faster. The changes from a global IP based access control mechanism to an instance based access mechanism apply to both SUSE Linux Enterprise Server (SLES) and SUSE Linux Enterprise Server For SAP Applications (SLES For SAP) on-demand instances and any images released in the future that might access the update infrastructure.

• Learn how to save money, reduce complexity with SUSE Manager [Ed: Linux has been around since the 1970s, it says. OK, whatever...]

  “The first is cost,” he says. “Linux has been around since the 1970s and has come a long way in that time. In one month (April 2020), Linux installations grew from 1,3% of the total installed base to a 3%. This might not sound like a lot, but it represents massive growth. For some Linux distributions, the grow rate was better than 600%.” [...] Brink points out that switching to a Linux front-end and an effective back-end management tool could save organisations a massive chunk of their end user license costs. SUSE Manager monitors an organisation’s infrastructure and manages how they deploy services on to front-end devices from a central point.

• OSI Charting a Course for 2020 and Beyond [Ed: Why does the OSI take pride in becoming a home for a Microsoft front group like ClearlyDefined?]

  The key to understanding how we move forward is to first remember how we got here. OSI as we know it didn't exist until 2013. Founded in 1998, the organization was held together in its first decade through strong board leadership in Michael Tiemann (2001-2012) and Danese Cooper (2002-2011). Deb Bryant (2012-present), Karl Fogel (2011-2014), Mike Milinkovich (2012-2018), and Simon Phipps (2010-2020) helped OSI begin professionalizing, by hiring General Manager Patrick Masson (2013-present), and becoming more democratic, with the introduction of a community-elected board. Molly de Blanc (2016-2020), Allison Randal (2014-2019), and Stefano “Zack” Zacchiroli (2014-2017) fostered better ties with the free software community. Richard Fontana (2013-2019) elevated legal discussions, taking OSI’s licensing work from knowledgeable hackers to expert practitioners and defining a review process. And Pam Chestek (2019-present) has brought a new level of professionalism to the license review process. This is a reductionist and inevitably incomplete view of OSI’s history, but the point is this: OSI has come a long way, and I am forever grateful to the talented and generous individuals who collectively invested decades to get us here. Over the last seven years, OSI has: sustained its core mission, shaped policy around the globe, worked tirelessly to mitigate open washing, built an alliance of more than 125 organizations representing hundreds of thousands of people, provided a home for projects like ClearlyDefined, and rolled out programs like FLOSS Desktops for Kids and Open Source Technology Management courses with Brandeis University.

• Priyanka Sharma Joins CNCF as General Manager

IBM/Red Hat/Fedora Leftovers

• Fedora Community Blog monthly summary: May 2020

  This is the first in what I hope to make a monthly series summarizing the past month on the Community Blog. [...] In May, we published 31 posts. The site had 4,964 visits from 2,392 unique viewers. Readers wrote 13 comments. 202 visits came from Fedora Planet, while 716 came from search engines.

• Red Hat Success Stories: A foundation for network automation and betting on OpenShift

  You hear the expression "betting" on platforms all the time. But Bilyoner Interactive Services in Turkey is really betting on Red Hat OpenShift by deploying a live betting platform on OpenShift with Red Hat Ansible Automation. When live sports betting was legalized in Turkey, Bilyoner Interactive Services needed a supported, scalable, and highly available technology foundation to support this new service. By migrating from community open source to Red Hat OpenShift and Red Hat Ansible Automation Platform, Bilyoner used container and microservices technology to quickly create and launch its new live betting platform. As a result, the company reports a five-fold increase in traffic and close to 100% service uptime.

• Kafka Monthly Digest – May 2020

  In this 28th edition of the Kafka Monthly Digest, I’ll cover what happened in the Apache Kafka community in May 2020.

• Free cloud native security conference hosted by IBM Developer

  Security concerns remain one of the key factors in enterprises unlocking the true value of the cloud. From modernizing applications with containerized microservices, to securing data while training AI models, or building continuous, secure DevOps pipelines in a growing complex hybrid cloud, developers face myriad challenges when it comes to security in a cloud native hybrid cloud environment. IBM Developer wants security to be one less thing you have to worry about when you’re building high-performance solutions. That’s why we put together the Digital Developer Conference: Cloud Native Security on June 24, 25, and July 1. [...] Learn the skills to react with speed and confidence by using solutions on IBM Cloud and Red Hat OpenShift alongside leading open source contributions by IBM and Red Hat to Kubernetes, Istio, Open Container Initiative, Cloud Native Computing Foundation, and Apache Foundation.

• Enable Sysadmin celebrates one-year anniversary with Sudoers Program

  What started as an idea in early 2019 has now blossomed into a publishing platform with a growing community with more than 100 writers. As we celebrate the one-year anniversary of the Enable Sysadmin publication, we’re excited to announce a new program for our community of writers. On May 5, 2020, we officially launched the Sudoers program for the Enable Sysadmin community. The Sudoers program recognizes our most trusted and committed contributors and provides a framework for becoming an established writer on the site. The editorial team has been working closely with 10 of our writers to help establish the first group of members in the Sudoer program. To date, this group of amazing sysadmins has collectively published more than 100 articles on the Enable Sysadmin publication.

• Enable Sysadmin: A year by the numbers