Security updates have been issued by Debian (libcaca), Fedora (beep and libgxps), Mageia (krb5, live, ffmpeg, mplayer, and vlc, and mbedtls), SUSE (helm-mirror, java-1_7_0-openjdk, and systemd), and Ubuntu (nss and python-django).
<p>Python has always touted itself as a "batteries included" language; itsstandard library contains lots of useful modules, often more than enough tosolve many types of problems quickly. From time to time, though, some havestarted to rethink that philosophy, to reduce or restructure the standardlibrary, for a variety of reasons. A discussion at the end of November on the python-dev mailing list revived that debateto some extent.
Security updates have been issued by Arch Linux (elfutils, polkit, and tar), Debian (python-django and ruby-loofah), and Mageia (ansible, avidemux, coreutils, discount, nettle, openafs, opensc, and qtbase5).
Many projects have adopted the "GitHub style" of development over the lastfew years, though, of course, there are some high-profile exceptions that still use patches and mailing lists. Many projects are leery of puttingall of their project metadata into a proprietary service, with limitedmeans of usefully retrieving it should that be necessary, which is whyGitLab (which is at least "open core") has been gaining some traction. A recently announcedeffort looks to kind of bridge the gap; Drew DeVault's sr.ht ("the hacker's forge")combines elements of both styles of development in a "100% free and open source softwareforge". It looks to be an ambitious project, but itmay also suffer from a lack of "social network" effects, which is part ofwhat sustains GitHub as the forge of choice today, it seems.
Version 5.0 of the Bash shell has been released."The most notable new features are several new shell variables: BASH_ARGV0,EPOCHSECONDS, and EPOCHREALTIME. The `history' builtin can remove ranges ofhistory entries and understands negative arguments as offsets from the endof the history list. There is an option to allow local variables to inheritthe value of a variable with the same name at a preceding scope. There isa new shell option that, when enabled, causes the shell to attempt toexpand associative array subscripts only once (this is an issue when theyare used in arithmetic expressions). The `globasciiranges' shell optionis now enabled by default; it can be set to off by default at configurationtime."
On the Red Hat community blog, Dave Neary writes about community governance and, in particular, how to choose who gets a vote, who can run, and how to decide a winner when electing a leader or council. He summarizes a number of different options that he has encountered with an eye toward avoiding the deep rat-hole conversations that picking a way to run elections can engender."Defining the activity metric and minimum bar for what qualifies as participation can become contentious, mainly because where you draw the line will be arbitrary, and will omit people who you want to include, or include people who you want to omit. For example, if you set the bar at the minimum contribution level of one commit to the project, you omit all whose contributions are significant but not code related. The typical fear is ballot stuffing or cohort effects — where large companies will dominate the representative bodies by having a large voting bloc, or where friends of candidates (or people with a certain agenda) will pass the low bar to become voters just to vote for their candidate."
Linus Torvalds released 5.0-rc1 onJanuary 6, closing the merge window for this development cycle andconfirming that the next release will indeed be called "5.0". At thatpoint, 10,843 non-merge change sets had been pulled into the mainline, about2,100 since last week's summary waswritten. Those 2,100 patches included a number of significant changes, though, includingsome new system-call semantics that may yet prove to create problems for existinguser-space code.
Linus has released the 5.0-rc1 kernelprepatch and closed the merge window for this development cycle."The numbering change is not indicative of anything special. If youwant to have an official reason, it's that I ran out of fingers and toes tocount on, so 4.21 became 5.0."
One of the useful features added during the 4.20 development cycle was theavailability of pressure-stall information,which provides visibility into how resource-constrained the system is.Interest in using this information has spread beyond the data-centerenvironment where it was first implemented, but it turns out that there someshortcomings in the current interface that affect other use cases. SurenBaghdasaryan has posted a patchset aimed at making pressure-stall information more useful for theAndroid use case — and, most likely, for many other use cases as well.
Security updates have been issued by Fedora (wget), Oracle (kernel), Red Hat (keepalived), Scientific Linux (keepalived), and SUSE (GraphicsMagick and mailman).
The fs-verity mechanism, created to protectfiles on Android devices from hostile modification by attackers, seemed tobe on track for inclusion into the mainline kernel during the current mergewindow when thepatch set was posted at the beginning of November. Indeed, it wasn'tuntil mid-December that some other developers started to raise objections.The resulting conversation has revealed a deep difference of opinion regardingwhat makes a good filesystem-related API and may have implications for howsimilar features are implemented in the future.
Security updates have been issued by Debian (jasper, libdatetime-timezone-perl, qtbase-opensource-src, thunderbird, and tzdata), Red Hat (rh-perl524-perl), and SUSE (libraw, polkit, and xen).
There are currently a number of sites bouncing emails from LWN.net due to analleged listing in the dnsbl.njabl.org blacklist. The only problem is thatthis blacklist hasbeen offline since 2013. That domain has been taken over by somebodyelse; checking web content at that address is not recommended as it is, toput it lightly, non-technical. If you are not getting expected email fromLWN, you might want to look into whether your local setup is still tryingto use that old, discontinued blacklist.
Antonio Cuni writesabout recent work to support running Python code with low latencyrequirements under PyPy. "As we said, the total cost of memorymanagement is less on PyPy than on CPython, and it's one of the reasons whyPyPy is so fast. However, one big disadvantage is that while on CPython thecost of memory management is spread all over the execution of the program,on PyPy it is concentrated into GC runs, causing observable pauses whichinterrupt the execution of the user program. To avoid excessively longpauses, the PyPy GC has been using an incremental strategy since 2013. TheGC runs as a series of 'steps', letting the user program to progressbetween each step."
The January 3 LWN.net Weekly Edition will be our first for 2019, markingour return after an all-too-short holiday period. Years ago, we made theill-considered decision to post some predictions at the beginning of theyear and, like many mistakes, that decision has persisted and become anannual tradition. We fully expect 2019 to be an event-filled year, withboth ups and downs; read on for some wild guesses as to what some of thoseevents may look like.
The Internet Archive (IA) has been around forover 20 years now; many will know it for its Wayback Machine, which is an archive ofold versions of web pages, but IA is much more than just that. Tracey Jaquith said that sheand her IA colleague David Van Duzer would relate a "love/hate, longadventure story—mostly love" about the migration of parts of IA toKubernetes. It is an ongoing process, but they learned a lot along theway, so they wanted to share some of that with attendees of KubeCon +CloudNativeCon North America 2018.
The linux-kernel mailing list carries the sad news that Shaohua Li, atalented contributor to much of the core kernel and the maintainer of theMD RAID subsystem, passed away over the holidays. Thank you for your work,Shaohua, you will certainly be missed.
Dylan O'Mahony, the cloud architecture manager for Bose,opened a presentation atKubeCon +CloudNativeCon North America 2018 by noting that many attendees may bewondering why a "50-year-old audio company" would be part of a presentationon Kubernetes. It turns out that Bose was looking for ways to support itssmart-speaker products and found the existing solutions to be lacking.Bose partnered with Connected, "a product development company fromToronto", to use Kubernetes as part ofthat solution, so O'Mahony and David Doyle from Connected were at theconference to describe the prototype that they built.
Cloud computing services that run customer code in short-lived processesare often called "serverless". Butunder the hood, virtual machines (VMs) are usually launched to run that isolatedcode on demand. The boot times for these VMs can be slow. This is the causeof noticeable start-up latency in a serverless platform like Amazon WebServices (AWS) Lambda. Toaddress the start-up latency, AWS developed Firecracker, a lightweightvirtual machine monitor (VMM), which it recentlyreleasedas open-source software. Firecracker emulates a minimal device modelto launch Linux guest VMs more quickly. It's an interesting exploration ofimproving security and hardware utilization by using a minimal VMM built withalmost no legacy emulation.
When the 4.20 kernel was released onDecember 23, Linus Torvalds indicated that he would try to keep to thenormal merge window schedule despite the presence of the holidays in themiddle of it. Thus far, he seems to be trying to live up to that;just over 8,700 changesets have been merged for the next release, whichseems likely to be called 5.0. A number of long-awaited features arefinally landing in the kernel with this release.
The New York Times reportsthe death of Dr. Lawrence G. Roberts, who was heavily involved inArpanet. "Dr. Roberts was considered the decisive force behind packetswitching, the technology that breaks data into discrete bundles that arethen sent along various paths around a network and reassembled at theirdestination. He decided to use packet switching as the underlyingtechnology of the Arpanet; it remains central to the function of theinternet." (Thanks to Paul Wise.)
Security updates have been issued by Arch Linux (go, go-pie, and webkit2gtk), Debian (c3p0, debian-security-support, libextractor, and tar), Fedora (electron-cash, leptonica, LibRaw, mingw-leptonica, mingw-openjpeg2, mingw-poppler, nettle, openjpeg2, php-pear, sqlite, and vcftools), Gentoo (GKSu and rust), Mageia (keepalived and libtiff), openSUSE (containerd, docker, go, go, GraphicsMagick, libraw, mozilla-nspr and mozilla-nss, netatalk, polkit, wireshark, and xen), and SUSE (containerd, docker, go, libqt5-qtbase, mailman, wireshark, and xen).
Most processors spend a great deal of their time doing nothing, waitingfor devices and timer interrupts. In these cases, they can switch to idlemodes that shut down parts of their internal circuitry, especially stoppingcertain clocks. This lowers power consumption significantlyand avoids draining device batteries. There are usuallya number of idle modes available; the deeper the mode is, the less power theprocessor needs. The tradeoff is that the cost of switching to and fromdeeper modes ishigher; it takes more time and the content of some caches is also lost. In theLinux kernel, the cpuidle subsystem has the task of predicting which choicewill be the most appropriate. Recently, Rafael Wysocki proposeda new governor for systems with tickless operation enabled that isexpected to be more accurate than the existing menu governor.
Kees Cook summarizesthe security-related improvements in the 4.20 kernel."Enabling CONFIG_GCC_PLUGIN_STACKLEAK=y means almost alluninitialized variable flaws go away, with only a very minor performancehit (it appears to be under 1% for most workloads). It’s still possiblethat, within a single syscall, a later buggy function call could use'uninitialized' bytes from the stack from an earlier function. Fixing thiswill need compiler support for pre-initialization (this is underdevelopment already for Clang, for example), but that may have largerperformance implications."
Security updates have been issued by Debian (ghostscript, graphicsmagick, libarchive, libsndfile, libvncserver, ruby-sanitize, and wireshark), Fedora (mosquitto and tinc), Mageia (monit, sqlite3, and thunderbird), and SUSE (openssl).
Linus has released 4.20 as expected."Let's face it, last week wasn't quite as quiet as I would have hopedfor, but there really doesn't seem to be any point to delay 4.20because everybody is already taking a break." Some of the headlinefeatures in 4.20 includenetwork flow dissectors in BPF,the tapriotraffic scheduler,peer-to-peer DMA support in the PCI layer,C-SKY architecture support,the pressure-stall instrumentation mechanism,the XArray data structure,and much more. The KernelNewbies 4.20 page iscoming together with more information.
This year's holiday gifts will include the 4.20 kernel; that can only meanthat it is time for another look at where the code going into this releasehas come from. This development cycle was typically busy and brought a lotof new code into the kernel. There are some new faces showing up in thestatistics this time around, but not a lot of surprises otherwise.
Greg Kroah-Hartman has announced the release of five new stable kernels: 4.19.12, 4.14.90, 4.9.147, 4.4.169, and 3.18.131. As usual, these contain importantfixes throughout the tree; users of those series should upgrade.
The kernel's live-patching (KLP) mechanism can apply a wide variety offixes to a running kernel but, at a first glance, the sort of highlyintrusive changes needed to address vulnerabilities like Meltdown or L1TFwould not seem like likely candidates for live patches.The most notable obstacles are the requiredmodifications of global semantics on a running system, as well as theneed for live patching the kernel's entry code. However, we at the SUSE livepatching team started working on proof-of-concept live patches for thesevulnerabilities as afun project and have been able to overcome these hurdles. The techniques we developed are generic and might become handy again whenfixing future vulnerabilities.
Security updates have been issued by CentOS (ntp), Debian (openssl1.0), openSUSE (salt), Oracle (firefox, ghostscript, and ntp), Red Hat (ntp), and SUSE (bluez, git, libnettle, ovmf, and tiff).
The December 20 LWN.net Weekly Edition is the final one for the year; asusual, we will be taking the last week of the year off for a brief rest.LWN, which is about to conclude its 21st year of publication, has had thetime to build up some traditions, one of which is a year-end retrospectivethat evaluates the predictions we made backin January. As usual, some of those predictions aged rather better thanothers; read on for our report card.
A year-old bug in Kubernetes was thetopic of a talk given by Michelle Au and Jan Šafránek at KubeCon+ CloudNativeCon North America, which was held mid-December inSeattle. In the talk, they looked at the details of the bug and theresponse from the Kubernetes product security team (PST). While the bug was fairly straightforward, it wassurprisingly hard to fix. The whole process also provided experience thatwill help improve vulnerability handling in the future.
Security updates have been issued by CentOS (ghostscript), Fedora (ansible and wireshark), openSUSE (go1.11, pdns, and pdns-recursor), Oracle (firefox), Red Hat (java-1.8.0-ibm), Scientific Linux (firefox), and SUSE (crash, libqt5-qtbase, perl, and qemu).
Back in late October, when we looked in onthe Python governance question, which came about due to the resignation of Guido van Rossum, things seemedto be mostly set for a vote in late November. There were six PythonEnhancement Proposals (PEPs) under consideration that would be ranked byvoters in a two-week period ending December 1; instant-runoffvoting would be used to determine the winner. In the interim, though,much of that changed; the voting period, winner-determination mechanism,and number of PEPs under consideration are all different. But the votingconcluded on December 16 and a winnerhas been declared; PEP 8016 ("TheSteering Council Model"), which was added to the mix in early November, cameout on top.
HardenedBSD has releasedversion 12 of its security-enhanced fork of FreeBSD. Improvements inthis release include Non-Cross-DSO Control-Flow Integrity (CFI) forapplications on amd64 and arm64; jailed bhyve; per-jail toggles forunprivileged process debugging; Spectre v2 mitigation with retpolineapplied to the entirety of base and ports; Symmetric Multi-Threading (SMT)disabled by default; and more.
Security updates have been issued by Debian (libapache-mod-jk and sleuthkit), Fedora (kernel, kernel-headers, mbedtls, php, php-symfony, php-symfony3, php-symfony4, and wireshark), openSUSE (pdns, pdns-recursor, and salt), Oracle (firefox and ghostscript), Red Hat (ansible, firefox, ghostscript, and kernel), Scientific Linux (firefox and ghostscript), and SUSE (ovmf).
LWN.net