Loading…
LinuxCon + CloudOpen + ELC-E Europe 2014 has ended
To Learn More: LinuxCon Europe | CloudOpen Europe | Embedded Linux Conference Europe.

Attendees! Please provide us feedback on the sessions you attend! Click here to submit a brief survey for each session and win a $250 Amazon gift certificate. 

>> Tracing Summit: View the Full Schedule

Sign up or log in to bookmark your favorites and sync them to your phone or calendar.

LinuxCon [clear filter]
Monday, October 13
 

3:30pm

Slab Allocators in the Linux Kernel - Christoph Lameter
Slab allocators are providing memory allocation to kernel code. Often these allocations may occur in performance sensitive code (especially in the network and storage layer). Kernel performance is significantly affected by the slab allocators architecture and features.

The talk provides an overview of the slab allocator services available in the Linux kernel and covers the most frequent use cases. The approaches used in the three slab allocators (SLAB, SLUB, SLOB) are compared and ways to get the maximum performance out of each of them are presented.

Lastly a series of common gotchas when using slab allocators are reviewed followed by an open discussion on issues and the future of the slab subsystems.

Speakers
avatar for Christoph Lameter

Christoph Lameter

R&D Team Lead, Jump Trading LLC
Christoph Lameter is working as a lead in research and development for Jump Trading LLC (an algorithmic trading company) in Chicago and maintains the slab allocators and the per cpu subsystems in the Linux Kernel. He contributed to a number of Linux projects since the initial kernel... Read More →


Monday October 13, 2014 3:30pm - 4:20pm
Room 8
 
Tuesday, October 14
 

11:15am

Testing your Full Software Stack on a Single Host With cwrap - Andreas Schneider, Red Hat
Testing your full software stack on a single host with cwrap. Testing network applications correctly is hard. This talk will demonstrate how to create a fully isolated network environment for client and server testing on a single host, complete with synthetic account information, hostname resolution, and privilege separation.

The cwrap project aims to help client/server software development teams to gain full functional test coverage. It makes it possible to run several instances of the full software stack on the same machine and perform functional testing of complex network configurations. Daemons run with privilege separation and required user and group accounts, irrespective of the hosting system. The cwrap project does not require virtualization and can be used to build environments on different operating systems.

Speakers
AS

Andreas Schneider

Principal Software Engineer, Red Hat
Source Code Artist


Tuesday October 14, 2014 11:15am - 12:05pm
Room 14

2:30pm

x86 Instruction Encoding and Nasty Hacks We Do in the Linux Kernel - Borislav Petkov, SUSE
I have always wanted to understand x86 instruction encoding in detail but never gotten around to it. Of course not, who has time nowadays?! So, in order to force me to do it, I decided to write an x86 instruction decoder.

This talk attempts to show what I have learned in the process and how instruction encoding is done on x86.

As a practical aspect, the decoder I've scratched together tries to verbosely show some of the crazy low-level hacks^Wtechniques we do in the Linux kernel like alternatives patching, jump labels, exception tables, etc - they have a lot to do with deep knowledge of x86 instructions and how code is generally laid out in the binary kernel image. Maybe this talk can help shed some light on the whole fun that's happening under the hood in the kernel and so many are missing out on. And maybe it'll make it more interesting and palatable to people.

Speakers
BP

Borislav Petkov

SUSE
RAS/AMD kernel maintainer working currenly at SUSE Labs. Prior to that at AMDs Operating Systems Research Center doing Linux enablement and hardware debugging work.


Tuesday October 14, 2014 2:30pm - 3:20pm
Room 16

3:30pm

Where is My Crystal Ball? - Daniel Lezcano, Linaro
The increasing part of the embedded systems in the linux ecosystem forces the kernel developers to take into account an energy efficient approach when bringing a new platform. The different energy frameworks are standalone sub-systems acting independently and in a opportunistic way when there is nothing to do on the system. The energy efficient scheduler wants to integrate all these energy components in order to act proactively by having a better knowledge of the potential energy saving for each scheduling decision it will take. This presentation describes a new paradigm where the events occurring in a acceptable interval are considered predictable and can be tracked per task. It will describe the IO latency tracking fully integrated in the scheduler and, thanks to better predictions, allows to get ride of the cpuidle's governor by directly choosing an idle state from the scheduler.

Speakers

Tuesday October 14, 2014 3:30pm - 4:20pm
Room 14

3:30pm

Cut Power Consumption by 5x Without Losing Performance: a Big.LITTLE Software Strategy - Klaas van Gend, Vector Fabrics
Multicore silicon architectures are everywhere. Until recently, these architectures where homogeneous. Recent big.LITTLE silicon aims to trade-off power consumption against computational throughput.
Typically, the big cores are used for high performance tasks like web browsing, games and image processing. Unfortunately, the usage of big cores comes with high power consumption levels - resulting in a disappointing battery life.
The little cores have a much lower power consumption level, but typically used in ‘idle mode’.

In this presentation we show how we reworked Chromium to use all available ‘little’ CPUs in a parallel fashion to achieve a highly responsive browser without incurring the power penalty of using the big cores. Experiments with real websites show that this results in a 5x power consumption reduction - at the same performance.

Speakers
avatar for Klaas van Gend, Vector Fabrics

Klaas van Gend, Vector Fabrics

Vector Fabrics
Klaas van Gend is one of the experts for Vector Fabrics, a company specializing in multi-core and many-core programming. Vector Fabrics’ expertise and tools are used by companies around the globe to improve existing code for parallel operation. Visionaries in processor design and... Read More →


Tuesday October 14, 2014 3:30pm - 4:20pm
Room 8
 
Wednesday, October 15
 

11:15am

Ftrace Kernel Hooks: More Than Just Tracing - Steven Rostedt, Red Hat
The function hook facility of ftrace is what makes ftrace stand out from other kernel tracers. The ability of live modification of the kernel to convert a nop into a ftrace callback has revolutionized tracing inside the kernel. Because it is dynamic, ftrace gives you the ability to chose what functions are to be traced, as well as tracing all functions within the kernel. This talk is not about tracing though, it is about what is coming in the future and the hurdles that needs to overcome and how it will be done. The ftrace function hooks allows for "hijacking" of a function. That is, when the function is called, the hook can intercept the call and divert it to call another function. Live patching such as kgraft and kpatch were built on this facility. To use these new features of ftrace, more must be done. This talk will go over how the hooks work and what more needs to be done.

Speakers
avatar for Steven Rostedt

Steven Rostedt

Principal Software Engineer, Red Hat Inc
Steven Rostedt works for Red Hat and is the main developer for their Real Time kernel. Steven is the maintainer of the Real-Time stable releases. He works upstream mainly developing and maintaining ftrace (the official tracer of the Linux kernel). He also maintains trace-cmd and kernelshark... Read More →


Wednesday October 15, 2014 11:15am - 12:05pm
Room 2

12:15pm

Stateless Systems, Factory Reset, Golden Master Systems and systemd - Lennart Poettering, Red Hat
Many of systemd's most recent low-level changes and additions focus on stateless systems, factory reset logic and golden master systems, which are particularly interesting for container and embedded environments, as well as systems where the OS needs to be fully verified. In this talk I intend to discuss the various details. More specifically, I'll shed some light on what's necessary to make systems boot without /etc or /var, with only /usr populated. I'd like to discuss the details and in particular the challenges this means for the distributions, and what the benefits are.

Speakers
LP

Lennart Poettering

Principal Software Engineer, Red Hat
Lennart works on systemd, for Red Hat.


Wednesday October 15, 2014 12:15pm - 1:05pm
Room 2

2:30pm

Scaling Userspace @ Facebook - Ben Maurer, Facebook
Ben Maurer will discuss the server workloads that Facebook runs across its fleet. Ben will talk about work Facebook has done in various systems areas such as:

- Memory management: improvements we've made to memory allocation performance with jemalloc, reducing fragmentation in programs with large numbers of threads, and increasing performance with the use of NUMA.
- Synchronization: LifoSem -- a last, in first out semaphore built on top of futex
- Scheduling: using CPU queuing delay to measure capacity
- Networking: measuring network performance with TCP tracepoints.

Speakers
BM

Ben Maurer, Facebook

Facebook
Ben Maurer is the tech-lead of the Web Foundation team at Facebook. This team is responsible for managing the performance and reliability of Facebook's user facing infrastructure. Ben works at all layer's of Facebook's stack -- from javascript to the kernel. Ben joined Facebook in... Read More →


Wednesday October 15, 2014 2:30pm - 3:20pm
Room 8

4:30pm

High Performance Storage with blk-mq and scsi-mq - Christoph Hellwig
This presentation gives an overview over the problems of the existing Linux storage stack to deal with low-latency and high IOPS devices, and explains how these are addressed for future Linux releases. Blk-mq provides a replacement for parts of the Linux block layer and allows drivers to support low-latency I/O, and a high number of I/O operations as well as scale better to large number of CPUs. SCSI is the most important enterprise block storage protocol, and thus enabling the SCSI layer to use blk-mq allows to take advantage of it for a wide range of commercially available storage hardware, especially high performance storage arrays. This presentation will explain the high-level details of the architecture of blk-mq and scsi-mq, show performance comparisons to the previous architecture and will show developers how to take advantage of the new capabilities.

Speakers
CH

Christoph Hellwig

Christoph Hellwig has been working on Linux Storage and File system projects for 15 years. He works all the way up and down the Storage and File system stack, and runs a business focused on Linux Storage architecture and training.


Wednesday October 15, 2014 4:30pm - 5:20pm
Room 14