Linux Plumbers Conference 2023
2023-11-13..2023-11-15
-- online --
[Linux Plumbers Conference 2022, Linux Plumbers Conference 2023, Linux Plumbers Conference 2025]
| title | author | length |
|---|---|---|
| Introduction | Palmer Dabbelt | 02m |
Short logistical briefing for the conference: organizers skip slides so sessions are discussion oriented and attendees are expected to converse rather than… | ||
| Deprecating Stuff | Palmer Dabbelt | 23m |
Proposes removing niche kernel bits that nobody boots or builds regularly: mark seldom-tested configs as deprecated in Kconfig, emit an on-boot warning, and… | ||
| Run ILP32 on RV64 ISA (RV64ILP32) | Ren Guo | 19m |
The rv64ilp32 patch series combines u64ilp32 and s64ilp32 to run ILP32 userland on RV64 hardware and adds the first 64ilp32 kernel; user and kernel changes are… | ||
| RISCV patchwork CI | Björn Töpel | 26m |
Pulls patches from Patchwork and reports results back, using the kernel patches daemon (KPD) to sync repos and GitHub Actions with ephemeral GitHub Runners… | ||
| Proposal of porting Trusted Execution Environment Provisioning (TEEP) Protocol... | Akira Tsukamoto | 15m |
Proposal to port the IETF TEEP reference implementation (released April 20, 2023 after a four‑year approval delay) from its current PMP + Keystone build to… | ||
| SBI Supervisor Software Events | Clément Léger | 26m |
Adds an SBI extension that injects high‑priority supervisor events from M mode that preempts all traps and interrupts by diverting execution to an SSE entry… | ||
| Perf feature improvements in RISCV | Atish Patra | 16m |
RISC-V perf currently relies on esop_pmf and SBI PMU with firmware-provided event-to-counter maps but new ISA extensions add counter delegation allowing S-mode… | ||
| RISC-V Vector: Current Status and Next? | Tao Chiu | 12m |
RISC-V currently runs Vector in user mode with eager save/restore and deferred restore before return to user space, and kernel-mode Vector support is in… | ||
| Control Flow Integrity on RISCV | Deepak Gupta | 20m |
RISC-V CFI splits into ZICFI-LP landing pad checks that force all indirect branches to target an embedded label (with exceptions for AUIPC+JALR idioms) and… | ||
| RISC-V irqbypass with KVM | Andrew Jones | 20m |
Proposal implements kvm_arch_irq_bypass_add_producer for RISC-V using AIA VS files and MRIFs so devices can inject MSIs without hypervisor mediation; RISC-V… | ||
| Android MC Intro | Kareem ymore, John Sumit | 04m |
Room and agenda briefing that lists concrete logistics and agenda items and flags the few technical hooks worth watching: 11th Edition microconference, focus… | ||
| Driver Development Kit DDK and Vendor Workflow | John Moon | 15m |
Qualcomm migrates its external kernel modules for about 50 out-of-tree modules from build.sh to Google’s Kleaf framework and the Driver Development Kit (DDK)… | ||
| Simplified Android Kernel Driver Development with DDK v2 | Matthias Männich, Mr Yifan Hong | 15m |
DDK v2 claims to let a developer drop into an empty directory and, using hermetic toolchain and Kbuild under the hood, produce a flashable kernel module built… | ||
| BPF Access Control and CO-RE in Android | Neill Kapron | 15m |
Claim: enabling CO-RE via libbpf on Android lets BPF programs compiled once run across kernel versions but currently vendors only have socket-filter access and… | ||
| Binder: fixing contention in buffer allocations | Carlos Llamas | 15m |
Android binder suffers from alloc-mutex contention when many clients fault pages into a single mmaped service VMA, causing background tasks to hold the mutex… | ||
| Android Kernel testing with platform integration | Betty Zhou | 13m |
Android kernel testing integrates AOSP presubmit and postsubmit pipelines to catch regressions early using arm64 on physical phones watches tablets plus… | ||
| Improving suspend/resume time and runtime PM on Android | Saravana Kannan | 15m |
Proposes adding a device-tree property called post_init_supplier to mark resources not required at probe time so fw_devlink can break three-way cycles… | ||
| RISC-V support in Android | Curtis Galloway | 15m |
Android kernels build and boot in an emulator and basic platform plumbing is functional; qemu 8.1 and cuttlefish run RISC‑V images while full test CI is not… | ||
| Adding Third-Party Hypervisor to Android Virtualization Framework | Elliot Berman, Prakruthi Heragu | 13m |
Adds support for the Gunyah EL2 hypervisor into Android Virtualization Framework by landing a new CrosVM hypervisor crate and a Linux kernel driver (upstream… | ||
| Porting Android Automotive on Xen | Leo Yan | 13m |
Naro ports AOSP Android Automotive to a Xen guest VM and demonstrates a working build with fewer than ten guest patches while host work requires maintaining… | ||
| Pixel 6 support on androidmainline | Peter Griffin, William McVicker | 16m |
Pixel 6 runs Android Mainline kernels since 5.10 and the project charters GKI presubmit and postsubmit validation on a real phone, tests every GKI change and… | ||
| Can mainline Linux run on Android without vendor hooks | Qais Yousef | 15m |
Argues that many Android-only patches are now implemented as vendor hooks (tracepoints added to the GKI) to override kernel behavior because real devices… | ||
| 16KB Page Size Support | Juan Yescas, Kalesh Singh | 15m |
Testing 16KB base pages on a Pixel 6 shows concrete wins and concrete costs: binary load experiments report 4x fewer page faults and reduced ad-load latencies… | ||
| AOSP Devboards | Sumit Semwal | 14m |
Supports specific boards: HiKey, DB410c, 820c, DB45C and RB5 with an upcoming 85096 port; validators build and boot mainline kernels and AOSP userlands… | ||
| Welcome message and DL Server | Daniel Bristot de Oliveira | 22m |
A patch set implements a DL server that uses SCHED_DEADLINE to schedule an entire fair scheduler so FIFO/RT busy loops cannot starve CFS tasks, replacing… | ||
| Do nothing fast: How to scale idle cpus? | Mathieu Desnoyers | 19m |
A global raw spinlock in the idle path and a small delay loop during do_idle unexpectedly speeds a scheduler-heavy hackbench on a 192-core AMD EPYC, taking the… | ||
| system pressure on CPUs capacity and feedback to scheduler | Vincent Guittot | 19m |
Proposes adding a new kernel interface rscale_cpu_cap that reports sustainable CPU capacity for the next ~10–100 ms, distinct from existing rscale_cpu_capacity… | ||
| Optimizing Chromium Low-Power Workloads on Intel Notebooks | Len Brown, Ricardo Neri, Vaibhav Shanka | 21m |
Prototype exposes a per-task EQOS class in task_struct, Chromium is updated to tag threads, and the kernel applies that tag to Intel HWP’s EPP value on context… | ||
| How to reduce complexity in Proxy Execution | John Stultz | 21m |
Proxy execution aims to eliminate priority inversion by keeping mutex-blocked tasks on run queues and running their runnable owners instead; a userland… | ||
| 60 Adaptive userspace spinlocks with rseq | André Almeida, Mathieu Desnoyers | 17m |
Proposes exposing thread run state to userspace via rseq so userspace can do cheap, syscall-free checks and only spin when the lock owner is actually running… | ||
| CPU Isolation state of the art | Frédéric Weisbecker | 23m |
Yearly status report shows concrete fixes and remaining work: VM stat folding is disabled on nohz_full CPUs which largely removes the per-second disturbance… | ||
| Improving CPU Isolation with per-cpu spinlocks: performance cost and analysis | Leonardo Bras Soares | 18m |
Work shows replacing queue_work_on(isolated_cpu) with per‑cpu spinlocks avoids IPIs and preserves CPU isolation while costing surprisingly little: a normal… | ||
| QA about PREEMP_RT | Thomas Gleixner | 18m |
Threaded printk replaces fully synchronous console output with background console threads and a lockless handover state machine that lets a CPU either request… | ||
| Introducing PAGEMAP_SCAN IOCTL for Windows syscalls translation... | Andrei Vagin, Muhammad Usama Anjum | 22m |
Adds a kernel-side scanner that returns compact ranges of pages with selected PTE bits and can atomically get and clear write-tracking state for a mapping… | ||
| User namespaces with host-isolated UIDs/GIDs | Aleksandr Mikhalitsyn, Stephane Graber | 30m |
Proposal converts kuid_t/kgid_t to 64-bit so high UIDs/GIDs can be host-isolated while preserving 32-bit userland view, and implements VFS ID-mapped mounts to… | ||
| In Containers We Trust? Building Trust in Containerized Environments | Avery Blanchard | 08m |
Extends a hardware-to-container chain of trust by measuring container images at creation and mapping executable files during runtime, using IMA logs and TPM… | ||
| Fuse mounts recovery and Checkpoint/Restore | Aleksandr Mikhalitsyn, Stephane Graber | 24m |
Proposes adding a FUSE API to enable mount healing and CRIU-based checkpoint/restore by cleaning kernel-side FUSE state, dropping pending requests, and… | ||
| Cgroups and Enterprise Users | Kamalesh Babulal, Tom Hromatka | 34m |
Enterprise users report concrete migrations: libcgroup now supports v1, v2 and hybrid modes and provides a practical libcgroup abstraction/mapping that… | ||
| Protecting Sensitive Data in Container Checkpoints | Adrian Reber, Radostin Stoyanov, Wesley Armour | 24m |
Extends CRIU to encrypt checkpoint images end-to-end by generating a symmetric token, saving it as a CRIU-image crypto file with the cluster public key, and… | ||
| Closing session | Stephane Graber | 01m |
Session ends exactly at 6 pm with an announcement that the microcon closes now and attendees are thanked for coming this year and invited back next year, a… | ||
| Opening session | Daniel Borkmann, Jakub Kicinski | 04m |
Opening logistics for the networking and BPF Summit lay out three full days with 40 talks organized into 30-minute slots that include discussion and strict… | ||
| Evolving the BPF Type Format | Alan Maguire | 34m |
Proposes moving the ~8 MB vmlinux BTF into a separate vmlinux BTF module so embedded kernels avoid image bloat, and adding a standalone BTF renumbering step at… | ||
| Exceptions in BPF | Kumar Kartikeya Dwivedi | 28m |
Adds a new runtime primitive where BPF throw aborts execution and invokes a global exception callback with a cookie to choose the return value; the loader runs… | ||
| When BPF programs need to die: exploring the design space for early BPF... | Dan William, Raj Sahu | 30m |
Proposal offers a fast-path termination that dynamically patches running BPF bytecode to force helper calls to fail except those that free resources, relying… | ||
| Verifying the Verifier eBPF Range Analysis Verification | Harishankar Vishwanathan | 26m |
Tool automatically extracts the verifier’s C semantics into SMT and checks a formal soundness spec for the eBPF range and bitwise analyses using Z3 and LLVM… | ||
| BPF Memory Model Two Years On | Paul E. McKenney | 36m |
A concrete BPF memory model is proposed that honors dependencies and maps BPF instructions to specific ordering guarantees: exchange/compare-exchange = full… | ||
| Overflowing the kernel stack with BPF | Dan Williams, Sai Roop Somaraju, Siddharth Chintamaneni | 23m |
Demonstrates concrete exploits where eBPF code overflows the kernel stack by abusing verifier assumptions: the verifier assumes 8KB assumption of available… | ||
| BPF for Security and LSM updates | KP Singh | 30m |
Implemented a BPF-based safe_setid LSM that successfully blocks an unauthorized UID transition in a live VM, using libbpf bootstrap and clang 18, but hit… | ||
| BPF_LSM + fsverity for Binary Authorization | Song Liu, Boris Burkov | 22m |
Proposal implements binary authorization using fs-verity hashes signed by a secure signing service, signatures stored in xattrs, and enforcement by BPF_LSM… | ||
| Sysarmor Metas eBPF Security Detection and Enforcement Tool | Liam Wisehart, Shankaran Gnanashanmugam | 38m |
Sysarmor is an eBPF security daemon deployed across collocated hosts, Meta Network Appliances, development servers, cloud gaming and AWS/GCP that evaluates… | ||
| Extending Non-Repudiable Logs with eBPF | Avery Blanchard, George Almasi | 20m |
Adds a kernel module API that lets eBPF programs call bpf_process_measurement to push formatted measurements into the IMA log and extend them into a TPM… | ||
| Advancing Kernel Control Flow Integrity with eBPF | Jinghao Jia | 32m |
Proposes an eBPF-based kCFI implementation that rewrites indirect-call no-ops into a trampoline which saves registers, packs caller/callee into the BPF… | ||
| Modernizing Android BPF and the Android Security Model | Neill Kapron | 21m |
Proposal to enable modern BPF features like libbpf CO-RE and iterators on Android while keeping release devices secure; current stack uses a single early BPF… | ||
| Buzzing Across Space: The Illustrated Children’s Guide to eBPF | Quentin Monnet | 12m |
An illustrated children’s book explains eBPF by staging bees inside a starship: the kernel engine room requires verifier goggles to prevent broken fixes, a JIT… | ||
| Resolve and standardize early access to hardware for automotive industry with Linux | Khasim Syed | 41m |
Linux cannot meet automotive early-KPI needs so OEMs offload boot-time tasks to MCU firmware; the result is cost, fragmentation and non‑Linux friendly designs… | ||
| Rust for Linux | Miguel Ojeda | 46m |
Rust support is now merged (initial in v6.1) and the project reports active growth — ~500 mailing-list subscribers, ~70 regular chat users and ~8,000 messages… | ||
| Beginner Linux kernel maintainers toolbox | Krzysztof Kozlowski | 43m |
Practical checklist: add your maintainer tree to linux-next so bots and community CI (LKP, ukci, lkft) actually build and test patches before merge, use b4 to… | ||
| Speeding up Kernel Testing and Debugging with virt-meng | Andrea Righi | 44m |
A forked and maintained virtme-ng boots a recompiled kernel inside a copy‑on‑write live snapshot using verfs instead of 9p, and mounts the whole root with… | ||
| Emulating NT synchronization primitives in Wine | Zeb Figura | 32m |
Wine currently offloads NT synchronization to a single wine server process which becomes a bottleneck for multithreaded games. A proof of concept kernel driver… | ||
| Optimizing sysfs and procfs | Ajay Kaher, Vamsi Krishna Brahmajosyula | 35m |
Observed that ~40% of sysfs/procfs entries get accessed at boot, leaving many inodes/dentries resident; applying DCACHE_DONTCACHE to all entries reduces… | ||
| Powering up "discoverable bus-attached" devices on DT-based platforms | Abel Vesa | 45m |
Linux DT platforms fail to power devices that must be live for bus enumeration, forcing marking regulators "always-on" and draining battery; example is the… | ||
| Improving kexec boot time | Usama Arif | 44m |
Kernel boot time drops from about 4s to ~1s in tests by enabling deferred struct page init and applying parallel SMP boot so secondary CPUs bring themselves… | ||
| Security Features status update | Kees Cook, Qing Zhao, Bill Wendling | 54m |
Toolchains now match on fstrict-flex-arrays=3, -fsanitize=bounds and __builtin_dynamic_object_size(); GCC landed changes in 13–15 and Clang in 16 for trailing… | ||
| Synthesized CFI for hand-written assembly in GNU assembler | Indu Bhagat | 48m |
GNU assembler gains a new SCFI mode that tries to synthesize full DWARF call‑frame info for hand‑written assembly via --scfi=all, currently tested on x86_64… | ||
| Graph-based ABI analysis for fun and profit | Matthias Männich | 51m |
Graph-based ABI toolchain STG reads BTF and DWARF, builds a unified symbol/type graph and uses an SCC-based diff to avoid repeated work; it consumes 18 million… | ||
| Compiling for verified targets BPF | Jose E. Marchesi | 24m |
GCC BPF support reaches parity with clang for most kernel BPF selftests, with one or two failing due to LLVM-specific optimizations; goal is to make compiled… | ||
| Towards data type profiling | Namhyung Kim | 47m |
A working prototype links perf memory sampling with DWARF location lists to map PMU load/store samples to C data types and struct fields, and lands as perf… | ||
| VSCode for kernel development | Florent Revest | 48m |
A complete VSCode workflow that actually sends and receives kernel patches: a vscode-git-send-email extension that remembers series metadata, runs… | ||
| Callsite Trampolines | Mr. Aleksei Vetrov, Steven Rostedt, Suren Baghdasaryan | 01h 10m |
Compiler extension generates a per-callsite wrapper that creates a small static record (placed in a custom ELF section) and replaces calls with trampoline… | ||
| Generating BPF Verifier Friendly Code with Clang | Yonghong Song | 24m |
Describes concrete fixes and techniques to make Clang/LLVM emit code the kernel verifier accepts: backend passes that insert intrinsics, targeted LLVM patches… | ||
| Welcome | Sasha Levin, Shuah Khan | 01m |
Welcome opens the microconference and sets a concrete goal: produce a one‑year roadmap for kernel dependability and testing with measurable targets and shared… | ||
| the path to achieve a bug-free build on the mainline | Philip Li | 19m |
Summary of concrete outcomes and methods: build escapes are tracked across rc1–rc6 commits and attributed to specific tools so fixes map to smatch and… | ||
| Storing and Outputting Test Information: KUnit Attributes and KTAPv2 | Rae Moar | 24m |
Creates KUnit test attributes upstream to store supplemental data (speed and module name today) using flexible void-pointer attributes, exposes a struct and… | ||
| Quality in embargoed patches | Sasha Levin | 28m |
Embargoed fixes routinely land with minimal testing, produce regressions and performance cliffs, and sometimes break unrelated platforms; the concrete proposal… | ||
| Detecting failed device probes | Laura Nao | 35m |
Proposal: a generic regression detector that verifies devices actually instantiate and bind to drivers by parsing kernel driver match tables and checking the… | ||
| Unifying and improving test regression reporting and tracking | Gustavo Padovan, Ricardo Cañuelo | 33m |
A proof-of-concept ingests CarCI and other kernel CI outputs into the KCB aggregator, post-processes test histories and logs to filter infrastructure noise and… | ||
| Welcome/intro | Mr Davidlohr Bueso | 01m |
Welcome announces a daylong CXL microconference running until six PM with sponsor thanks and an anti-harassment policy reminder. Schedule gives a… | ||
| CXL Emulation in QEMU - Progress, status and most importantly what next? | Fan Ni, Jonathan Cameron | 28m |
QEMU now implements CXL emulation beyond basics: Dynamic Capacity support with one-to-8 regions, DPA extent tracking and basic read/write backed by mailbox… | ||
| CXL Type-2 core support | Ira Weiny | 15m |
Describes which CXL Type‑2 features exist upstream and which still need work: P2P, Back‑Invalidate, Cache‑ID configuration are required for accelerators, and… | ||
| Plumbing challenges in Dynamic capacity device | Ira Weiny, Jonathan Cameron, Navneet Singh | 29m |
CXL 3.0 allows asynchronous release of extents but kernel code in review keeps the response lightweight and defers real reclamation, with an RFC v3 patch set… | ||
| Adding RAS Support for CXL Port Devices - Terry Bowman | Terry Bowman | 24m |
Adds RAS handling by changing where internal-error callbacks live: a proof-of-concept CXL port service is tried then rejected because it duplicates CXL… | ||
| Shared CXL 3 memory what will be required | John Groves | 25m |
Argues that CXL 3 sharable tagged capacity must never be onlined as system RAM and instead needs a more file-like layer so existing apps can mmap datasets… | ||
| CXL Memory Tiering for heterogenous computing | Ravi Kiran Gummaluri | 26m |
CXL adds capacity but hardware interleaving is rigid, hides NUMA and forces iso‑DIMM sizing; the proposed Linux changes push allocator‑level, software weighted… | ||
| A move_pages() equivalent for physical memory | Mr Gregory Price, Svetly Todorov | 30m |
Proposal adds a syscall that migrates physical pages between NUMA nodes by reusing move_pages infrastructure, accepting device or host PFNs and requiring… | ||
| Hypervisor-Enforced Kernel Integrity (Heki) for KVM | Mr Madhavan Venkataraman | 41m |
Heki is a KVM-based prototype that moves kernel self-protection out of the kernel by enforcing control-register pinning and page permissions via EPT/GFN… | ||
| Multi-KVM Abstract | Anish Ghulati, Sean Christopherson | 26m |
Proposal builds a way to run multiple independent KVM modules on one host so you can upgrade or A/B test KVM without unloading the old module. It privatizes… | ||
| Unifying KVM API for protected VM and utilities | Isaku Yamahata | 16m |
Proposes adding VFIO-side region fragments so a portion of a device BAR gets a separate mapping with different attributes, driven by customer demand for… | ||
| pkernfs: Persisting guest memory and kernel/device state safely... | Alexander Graf, James Gowans | 43m |
Proposal to preserve guest RAM and kernel/device state across kexec by storing page tables and driver blobs in a new in‑memory persistent filesystem: pkernfs… | ||
| Hyper-V's Virtual Secure Mode in KVM project update | Nicolas Saenz Julienne | 23m |
A VSM proof-of-concept for KVM exists and a first patch series is imminent; two concrete designs compete: keep VTLs as separate vCPUs with kernel VTL awareness… | ||
| Supporting guest private memory in Protected KVM on Android | Fuad Tabba | 21m |
Port of the proposed guest memory API to pKVM on Arm64 implements guest private memory as a FD-like interface and a kernel-only no-map attribute updated on… | ||
| libside: Giving the preprocessor a break with a tracer-agnostic instrumentation API | Mathieu Desnoy | 18m |
libside provides a new ABI/API that lets kernel and user-space tracers attach to static and dynamic user-space instrumentation, exposing a type system and… | ||
| Graphing tools for scheduler tracing | Julia Lawall | 21m |
Trace-cmd traces scheduler events and the new graphing tools summarize them: dat-to-graph draws per-core run intervals across 128 cores so gaps show unused… | ||
| Function return hook integration with Function graph tracer | Mr Masami Hiramatsu | 16m |
Current kernel has two independent return-hook systems—kretprobe and function graph tracer—that both overwrite return addresses and keep per-task shadow… | ||
| pt_regs - the good the bad and the ugly | Florent Revest | 20m |
A history of struct pt_regs shows ptrace → kprobe → kretprobe → ftrace → fprobe → uprobe → redhook → eBPF → perf and a recurring problem: many users propagate… | ||
| RTLA Requests and TODOs | Daniel Bristot de Oliveira | 21m |
RTLA is a set of user-space binaries that measure scheduling latency and root‑cause it: a timerlot sampler flags latencies over ~30 µs and breaks them into… | ||
| Function parameters with BTF | Steven Rostedt, Mr Masami Hiramatsu | 21m |
Proposal shows using BTF to print return values and every argument for function graph and ftrace by building a quick lookup table that maps IPs to BTF… | ||
| HiSilicon Performance Monitor Control Unit | Jie Zhan | 19m |
Performance Monitor Control Unit (PMCU) offloads core PMU work to an external DMA-backed controller that configures events, switches them, and writes samples… | ||
| DTrace and eBPF: new challenges | Kris Van Hee | 19m |
DTrace reaches 98% feature parity and now runs on upstream kernels after removing non-upstream patches, but generating BPF from scripts hits the verifier hard… | ||
| Implementing sframes | Steven Rostedt, Indu Bhagat | 24m |
sframes are implemented in binutils to let the kernel unwind user stacks without frame pointers by reading an ELF sframe section for static binaries, but… | ||
| The taming of the kernel dump | Petr Tesarik | 29m |
A portable library called libkdumpfile parses many kernel dump formats, implements page-table translation for several ISAs, and can replace parts of… | ||
| drgn Writing to Memory and Breakpoints, Safely in Production? | Omar Sandoval | 29m |
drgn is read-only today and the concrete proposal is to add write arbitrary addresses and file:line breakpoints so scripts can patch running kernels to… | ||
| Beyond DWARF: Debugging the Kernel with Drgn BTF/CTF & kallsyms Stephen Brennan | Stephen Brennan | 29m |
Drgn adds a pluggable symbol finder and a DWARFless workflow that uses kallsyms lookup plus either CTF via libctf or parsed BTF and relies on frame-pointer/ORC… | ||
| When kdump is way too much | Guilherme G. Piccoli | 27m |
Kdump needs ~200 MB prereserved RAM and can fail during panic boot, so pstore is used on Steam Deck as a lightweight alternative that avoids rebooting a panic… | ||
| Minidump to debug end user device crashes | Mukesh Ojha, Elliot Berman | 28m |
Qualcomm ships a firmware-backed minidump that records a table of physical regions the kernel or subsystems register and, on a crash, firmware copies only… | ||
| Kernel Livedump | Lukáš Hruška | 19m |
Implements live kernel memory dump without reboot by using copy-on-write and dumping into a reserved target that currently writes a vmcore format image on x86… | ||
| Sensors aggregation | Alexandre Bailon, Daniel Lezcano | 16m |
Proposal builds an in‑kernel mechanism that combines multiple per‑core or per‑component sensors into a single thermal zone so one governor can use an… | ||
| New thermal trip types | Daniel Lezcano | 22m |
Proposes a new user trip point writable by userspace that only triggers a notification (no kernel mitigation) and requires sensors with programmable interrupt… | ||
| Use of Netlink for thermal kernel-user notification is problematic | Srinivas Pandruvada | 17m |
Kernel thermal notifications use NETLINK_GENERIC and NETLINK_KOBJECT_UEVENT and are shown to fail under event storms: experiments reproduce a full system… | ||
| CPUfreq/sched and VM guest workload problems | David Dai, Saravana Kannan | 15m |
Workloads in VMs show large CPUfreq/sched regressions versus host; a virtual MMIO CPU frequency control device plus DT property changes are implemented so… | ||
| VM-CPUFreq for x86: Scaling the guest frequency for performance and power savings | Mr Wyes Karny | 15m |
VM-CPUFreq V2 ported to x86 EPYC exposes guest vCPU utilization to the hypervisor via hypercall and ACPI P‑state hints (min, mid, max). Experiments use tbench… | ||
| Virtualized Frequency Control for Telco Workloads | Mr Chris Macnamara, Srinivas Pandruvada | 11m |
Telco packet-processing apps poll and appear as 100% polling workloads, so kernel governors see no idle time and give zero savings. A user-space power governor… | ||
| uclamp in CFS: Fairness, latency, and energy efficiency | Dietmar Eggemann, Morten Rasmussen | 17m |
Current upstream uclamp uses max aggregation, which lets many tasks share a low frequency and hides true compute demand, causes PELT saturation and frequency… | ||
| Make sync_state()/handoff work for the common clk framework | Stephen Boyd | 16m |
Problem: clock_disable_unused runs at lateinit but actually races with deferred probe, so hardware-enabled clocks get turned off unpredictably. Attempts… | ||
| Intel Low Power Mode Daemon on Hybrid CPUs | Rui Zhang, Srinivas Pandruvada | 14m |
intel_lpmd monitors procfs, kernel hints and user prefs and moves tasks onto the low‑power A‑cores to force the compute die into D6 and approach CPU‑hotplug… | ||
| Enabling DDR segments on demand during memory pressure for DDR power reduction | Sudarshan Rajagop | 13m |
A userspace C++ daemon registers to PSI and signals a kernel driver that uses memory hotplug to dynamically offline/online 1 GB DDR blocks, turning OFF… | ||
| Improving monitoring of power saving states | Mr Stanislaw Kardach (Google), Sven van Ashbrook | 15m |
ChromeOS measures CPU idle, module/package idle and system deep‑sleep residencies across x86 and ARM and finds current kernel interfaces fragmented: cpuidle… | ||
| bpftime: Fast uprobes with user space BPF runtime | Yusheng Zheng | 31m |
bpftime is a userspace eBPF runtime that uses binary rewriting to inject into running processes without restart and claims bpftime prototype uprobes that are… | ||
| Make ftrace_regs a common trace interface for function entry/exit tracing | Masami Hiramatsu | 29m |
Proposal replaces pt_regs with ftrace_regs for function entry/exit tracing, claiming a smaller, parameter-focused register set that avoids saving control… | ||
| Where have all the kprobes gone | Jiri Olsa | 27m |
Many BPF-attached tracing probes silently drop execution because different attach layers enforce different re-entry checks and some layers never exported miss… | ||
| xprobes: Hybrid User/Kernel eBPF Probes for Cross-Layer Observability | Lucas Castanheira | 30m |
Combines uprobes and kprobes into xprobes that keep a BCC shadow stack in eBPF maps and merge kernel events with userspace entry/exit to attribute on‑CPU… | ||
| BPF programmable netdevice | Daniel Borkmann | 31m |
A BPF-based driver called netkit programmable netdevice embeds BPF execution into the driver's xmit path so Cilium can move endpoint programs into the pod peer… | ||
| Application network security and observability in an encrypted future | John Fastabend | 30m |
Tetragon runs L7 parsing and enforcement entirely in-kernel, using BPF parsers attached via sockmap hooks to capture DNS, TLS handshakes and then transition to… | ||
| Safe sharing of the network with eBPF | Balasubramanian Madhavan, Prankur Gupta | 28m |
Meta uses eBPF to push concrete transport fixes: a receiver-driven rwnd rewrite at TC that cuts packet discards, lowers buffers and packets-in-flight; a… | ||
| BPF struct_ops - current status and the last developments | Kui-Feng Lee | 24m |
BPF struct_ops lets user BPF programs implement kernel-visible method tables so subsystems can call user-defined functions; TCP congestion control is already… | ||
| BPF Static Keys | Anton Protopopov | 27m |
Adds a BPF Static Keys API that compiles helpers into asm goto-generated NOP/JUMP patches and registers a list of patch points so user space can toggle… | ||
| Troubles and Tidbits from Datadogs eBPF journey | Guillaume Fournier, Hemanth Malla | 34m |
Datadog describes concrete failures using eBPF in production: building Cloud Security Management with kprobes/tracepoints and Cilium hooks exposes blind spots… | ||
| eBPF Shenanigans with Flux | Barret Rhoden | 24m |
Flux is a framework that runs schedulers written entirely in eBPF on Google’s Ghost kernel scheduling class and it successfully composes hierarchical… | ||
| Developing Continuous eBPF Profiler to look Beneath the Kernel to Beyond the Clouds | Sumera | 36m |
Built a continuous profiler that uses parca-agent and eBPF to sample kernel and user stacks at 19 Hz sampling, shipping compact profiles every 10s to a server… | ||
| Towards a standardized eBPF ISA - Conformance testing | Alan Jowett | 15m |
The BPF Conformance Suite runs a test runner plus runtime-specific plugins to validate a draft BPF ISA against real implementations: Linux kernel BPF (via a… | ||
| Standardizing Linux DRM drivers implementations by interfacing DRM Bridge as a single... | Jagan Teki | 27m |
Argues that encoder logic migrates into DRM bridges to create a single bridge API so host drivers can remove encoder ops and register one dummy encoder for… | ||
| Enabling Large Block Size devices in Linux | Luis Chamberlain, Pankaj Raghav | 47m |
Adds a page-cache API to encode a folio minimum order so filesystems can request folios at least as large as their block size, implements XFS changes and… | ||
| Linux Kernel Autotuning | Cong Wang | 46m |
Automated tuning of Linux kernel parameters uses Bayesian optimization and simple adapters to search configuration space and it produces concrete wins: a… | ||
| Standardizing CPUID data for the open-source x86 ecosystem | Ahmed S. Darwish | 47m |
Proposal standardizes CPUID by collecting and encoding the publicly known 830+ bitfields into an extensible XML schema and choosing XML over JSON/YAML for… | ||
| Hunting Heisenbugs | Paul E. McKenney | 47m |
Defines practical anti-heisenbug techniques: detect versus instrument, amplify rare events, and run targeted stress tests that make timing bugs happen… | ||
| nouveau and kernel GPU VMA management | David Airlie | 39m |
Nouveau now uses NVIDIA GSP firmware so reclocking works and GPUs run at expected speeds; firmwares are large at 20–30 MB firmwares which forces… | ||
| Putting Linux into Context – Towards a reproducible example system with Linux... | Philipp Ahmann | 42m |
Project builds a reproducible example system combining Linux, Xen, and Zephyr on real hardware and a CI that daily builds images, runs OpenQA tests and… | ||
| Dynamic vCPU priority boosting in KVM for latency sensitive workloads | Joel Fernandes, Vineeth Pilla | 45m |
A prototype for x86_64 implements host–guest communication using shared memory and hypercalls so the guest signals latency intent and the host temporarily… | ||
| RCU Office Hour | Paul E. McKenney | 45m |
Q&A-style session focused on concrete RCU fixes and tooling: two external pull requests land changes to RCU bus numbering, an Anamaria-made reproducer that… | ||
| Improve Linux Perf tool to account for task sleep | Ajay Kaher, Alexey Makhalov | 41m |
Current perf misses samples while a task is sleeping because the hrtimer handler stops on sleep; two kernel-side fixes are proposed and prototyped: calculate… | ||
| Powering up “discoverable bus-attached” devices on DT-based platforms | Abel Vesa | 45m |
Proposes creating a virtual PCI slot device bound as a platform child of the PCI host controller so DT node compatibles drive a slot/hotplug driver that powers… | ||
| Installing and Using the Linux-Kernel Memory Model (LKMM) | Paul E. McKenney | 37m |
Step-by-step setup of the Linux‑Kernel Memory Model shows how to install herd7 and OCaml, satisfy opam/dune dependencies and build the tools, then run litmus… | ||
| kdevops sync up | Luis Chamberlain | 35m |
vagrant is unmaintained, so kdevops moves toward alternatives: ktest/Kconfig, Terraform/OpenTF, or running guests inside Kubernetes using virt (KVM-in-pods). A… | ||
| Empowering Engagement: Introducing a Dynamic Dashboard for Proactive Retention Strategies | Prof. Anita Sarma et al. | 52m |
A retention-focused web dashboard called Survival dashboard mines GitHub activity and uses statistical survival models to predict group- and individual-level… | ||
| KDLP: Kernel Development Learning Pipeline | Joel Savitz | 44m |
Red Hat–sponsored open source project publishes full course materials and runs a four-credit course at UMass Lowell that teaches Git, patch workflow and… | ||
| resctrl filesystem | Peter Newman | 43m |
Resctrl is being refactored so the filesystem and core APIs support non-x86 QoS and monitoring like MPAM on ARM and CBQRI on RISC‑V; patches split locking… | ||
| Securing build platforms | Joshua Lock | 35m |
Maps SLSA concepts to distribution build platforms and demonstrates a Yocto proof of concept: generates SLSA provenance metadata, signs it to bind artifacts to… | ||
| Improving UAPI Compatibility Review with Automated Tooling | John Moon | 40m |
Describes adding libabigail-based automation and a shell wrapper check_uapi.sh that compiles headers, runs ABI-diff on binary artifacts, and flags incompatible… | ||
| kernel: build system outputs and workflows (and how to balance them) | Bruce Ashfield | 38m |
Describes how the Yocto/OpenEmbedded kernel integration actually works: build outputs are kernels, modules, DTBs, firmware, Linux-lipsy uapi headers, SBOM and… | ||
| How big of a problem are Un-upstreamed patches | Jon Mason | 32m |
Yocto/OpenEmbedded shows concrete counts: 200 backports, 253 pending, 452 inappropriate, 232 submitted, 19 denied, 21 inactive and 69 CVE-related patches, with… | ||
| Building for Heterogeneous Systems | Alejandro Enedino Hernandez Samaniego | 26m |
Build system shows how to produce Linux plus an RTOS/bare‑metal image in one run using bitbake multiconfig, creating separate cross toolchains (arm‑none‑eabi… | ||
| Confidential Computing Microconference Introduction | Dhaval Giani, Joerg Roedel | 01m |
Microconference opens with logistics and ground rules: organizers ask for volunteer note-takers and point to a shared EOP patch session for collaborative… | ||
| COCONUT Secure VM Service Module Discussion | Joerg Roedel | 13m |
COCONUT‑SVSM currently boots Linux guest at VMPL2 on AMD SEV‑SNP hardware, implements core protocol pieces and is around 12,000 lines of Rust with enforced… | ||
| Remote Attestation in AMD SEV-SNP Confidential VMs | Claudio Carvalho | 15m |
Runs a vTPM inside the SVSM of AMD SEV‑SNP Confidential VMs so hypervisor cannot read TPM state, with a modified kernel TPM driver and reuse of userland TPM… | ||
| Shrinking The Elephant - A Confidential Computing Attestation Sequel | Samuel Ortiz | 24m |
Confidential Containers ships a working open source attestation framework: the Attestation Service plus the Key Broker Service (KBS), both HTTP/HTTPS based… | ||
| How to Build a Confidential Attestation Client | Tobin Feldman-Fitzthum | 17m |
Explains how an attestation client inside a confidential guest can be turned into an evidence factory attack that impersonates many tenants by producing valid… | ||
| Supporting Live Migration of Confidential VMs in KVM | Pankaj Gupta | 20m |
Design proposes SEV‑SNP live migration using a VMPL0 SVSM that generates an SVSM-generated attestation report containing the hash of a newly created public… | ||
| Secure IO | Dan Williams, Jeremy Powell, Samuel Ortiz, Steffen Eiden, Thomas Lendacky | 31m |
Combined proposals converge on concrete mechanics: full guest images encrypted with hardware-bound keys, device identity attestation via SPDM, and a TSM/PSP… | ||
| Taming the Incoherent Cache Issue in Confidential VMs | Mingwei Zhang | 16m |
AMD's SME_COHERENT hardware flag does not cover device DMA, so confidential guest pages can leave dirty CPU caches and later corrupt host memory… | ||
| Towards unified confidential computing ABIs | Dan Williams | 13m |
Proposal claims Linux should host a common confidential‑computing ABI and enforces it via kernel ABI docs; implemented a configfs TSM register that creates… | ||
| Update on RISC-V Confidential VM Extension (CoVE) | Atish Patra, Ravi Sahita | 10m |
Proposes a RISC‑V memory tracking table (MTT) ISA extension that checks the final physical address after page-table walks and enforces per‑supervisor‑domain… | ||
| Secure TSC for AMD SEVSNP guests | Nikunj Dadhani | 11m |
SEV‑SNP exposes secure TSC so guests can run RDTSC/RDTSCP without trusting the hypervisor: the TSC scale and offset live in the encrypted VMSA, and the AMD… | ||
| Secure AVIC: Securing Interrupt Injection from a 'malicious' Hypervisor | Kishon Vijay Abraham I | 14m |
Secure AVIC moves the APIC backing page into guest-owned memory and adds Guest Owned APIC Backing Page fields plus allow IR / requested IR registers so the… | ||
| Improve Xeon IRQ throughput with posted interrupt | Jacob Pan | 29m |
Enables posted MSI on bare metal and adds a kernel RFC that coalesces delivery so only the leading interrupt in a burst notifies the CPU, avoiding slow APIC… | ||
| PCI Endpoint Subsystem Open Items Discussion | Manivannan Sadhasivam | 29m |
Three concrete open items: adding VIRTIO-based EPF drivers, device integration so EPF binds to EPC via devicetree, and replacing the custom outbound-window… | ||
| Non-discoverable devices in PCI devices | Lizhi Hou, Rob Herring | 29m |
Modern PCI endpoints expose multiple MMIO peripherals behind BARs and those peripherals change with FPGA reprogramming, so the practical solution is to attach… | ||
| IOMMU overhead optimizations and observability | Pasha Tatashin, Yu Zhao | 46m |
Observed IOMMU page table overhead reaches many gigabytes on real workloads; a small reproducer iova_stress cycles VA space and shows system free memory… | ||
| iommufd discussion | Mr Jason Gunthorpe | 39m |
iommufd exposes CPU IOMMU features to VMMs via the new vfio cdev interface, provides virtual IO page tables and nested translation using user-space hardware… | ||
| Linux-wpan updates | Mr Stefan Schmidt | 40m |
Kernel work lands progressive 802.15.4 features across 6.0–6.8: small releases (8, 13 patches), a 40‑patch push for management, then 26 patches for… | ||
| TSCH@Zephyr: IEEE 802.15.4 SubG IIoT in the Making | Florian Grandel | 31m |
Zephyr’s native IEEE 802.15.4 L2 now handles SubG, SUN O‑QPSK, FSK, OFDM and initial HRP UWB support and is running TSCH on TI CC13/26xx hardware with a… | ||
| Zephyr Retro-&-Prospective: Project Growth, Long Term Support, & Linux Interoperability | Chris Fri | 19m |
Zephyr reports concrete growth: support expands from ~200 to 626 boards, base architectures rise to 13 (22 exported), and contributor count doubles to ~1,000… | ||
| Shared FPU Support in Zephyr for ARM64 and RISC-V | Nicolas Pitre | 39m |
Zephyr implements lazy FPU context switching for ARM64 and RISC‑V to avoid saving 512‑byte FPU state on every switch, using a global owner tag and per‑thread… | ||
| Challenges in Device Tree Sync - kernel, Zephyr, U-boot, System DT | Nishanth Menon | 29m |
Work on AM625/TDA4VM shows a real gap: Linux, U-Boot and Zephyr each keep their own device-tree view for the same hardware (4x A53, R5, M4), resulting in two… | ||
| Breaking Barriers: Arduino Core API advancements in Zephyr, Linux and IoT Systems | Dhruva Gole | 27m |
Upstreamed an Arduino Core API module that runs Arduino sketches on Zephyr and as a Linux executable via a native_posix simulator, demonstrates three… | ||
| Livepatch Visibility at Scale | Breno Leitao | 34m |
Meta collects console output from millions of servers using netconsole and adds a kernel version to each message, but livepatches break that mapping. A… | ||
| KLP for Clang LTO Kernel | Yonghong Song | 24m |
Clang LTO live patching runs on top of kpatch by comparing linker-produced pre-link objects instead of the full vmlinux, using clang’s lto object-pass dump to… | ||
| Kbuild support for klprelocation generation | Lukáš Hruška | 29m |
Kernel build adds a small utility that rewrites livepatch ELF relocations into the loader’s required format and currently implements only the annotation path… | ||
| Simplify Livepatch Callbacks Shadow Variables and States handling | Petr Mladek | 18m |
Proposal changes kernel livepatch internals to tie callback lifetime and shadow-variable garbage collection to versioned states, so transitions and downgrades… | ||
| Moving livepatching module building to kselftests | Marcos de Souza | 35m |
Proposal moves livepatch test modules from lib/livepatch into tools/testing/selftests/livepatch and adds a kselftest target that compiles modules on the test… | ||
| Arm64 live patching | Mark Rutland | 30m |
Status: Arm64 kernel live patching is blocked by lack of reliable stack unwinding across exception boundaries despite extensive prerequisite work: implemented… | ||
| Klint: Compile-time Detection of Atomic Context Violations for Kernel Rust Code | Dr Gary Guo | 27m |
Klint is a compile-time checker for Kernel Rust that enforces compile-time detection of sleeping-in-atomic-context errors by tracking a function’s relative… | ||
| pin-init: Solving Address Stability in Rust | Benno Lossin | 26m |
Explains a concrete API that lets kernel code initialize pinned Rust values without unsafe code, demonstrating a working pin-init API used in Rust-for-Linux… | ||
| Coccinelle for Rust | Julia Lawall | 24m |
Coccinelle for Rust implements a semantic-patch engine that uses Rust Analyzer for parsing and type inference and relies on rustfmt for output, matching… | ||
| Using Rust in the binder driver | Alice Ryhl | 27m |
A Rust reimplementation of the binder driver reaches feature parity with the C driver, boots a device, passes the test suite, and matches microbenchmark… | ||
| Block Layer Rust API | Andreas Hindborg | 25m |
Project delivers a Rust block-layer API plus a Rust NVMe driver (PCI transport only) and a null_blk driver (210 lines, memory-backed), with a long list of Rust… | ||
| Rust in V4L2: a status report | Daniel Almeida | 21m |
A patchset with initial Rust patchset yields Rust abstractions for V4L2: vb2 queue/allocations, V4L2 ioctls and format enumeration, device probe and a sample… | ||
| Converting a DRM driver to Rust | Maíra Canal | 38m |
A 400-line VGEM DRM driver gets rewritten in Rust as a drop-in equivalent that uses Asahi Lina DRM bindings and Rust-for-Linux helpers for timers, DMA… | ||
| Zero Copy Receive using io_uring | David Wei | 33m |
Proposes a kernel-side zero copy RX path that delivers payload DMA straight into registered user memory while headers stay in kernel via header splitting, uses… | ||
| Enhancing Homa Linux for Efficient RPC Transportation | Xiaochun Lu | 24m |
Homa gets a dynamic per-peer window: an RTT probe every 1 ms measures real-time RTT and classifies RTT into low/mid/high to shrink or grow unscheduled and… | ||
| An introduction to the DPLL subsystem | Vadim Fedorenko | 15m |
A new Linux 6.7 subsystem exposes DPLL hardware via a netlink API to configure inputs, priorities and pin routing and to monitor lock status, phase offset and… | ||
| Unblocking the softirq lock for PREEMPT_RT | Sebastian Siewior | 31m |
Claims that the global local-BH disable on PREEMPT_RT behaves like a per-CPU Big Kernel Lock and blocks higher-priority threads; offers a proof-of-concept that… | ||
| Offloading QUIC Encryption to Enabled NICs | Andy Gospodarek | 30m |
Proposal describes kernel module that exposes a netlink control plane to register QUIC flows, a driver ndo that returns an xid connection ID, and per-packet… | ||
| Extending AF_XDP with hardware metadata | Stanislav Fomichev | 22m |
AF_XDP now exposes NIC hardware metadata on receive via XDP helper calls so userland can read receive timestamp and receive hash (implementation merged around… | ||
| connect() - why you so slow?! | Frederick Lawler | 26m |
Opening ~50k connections to one origin makes connect() latency bimodal and wildly slow: experiments and flamegraphs show the inet_hash_connect loop plus parity… | ||
| Container Networking: The Play of BPF and Network NS with different... | Martin Lau, Takshak Chahande | 26m |
Shows that per-container IPv6 plus cgroup-bpf bind/connect/sendmsg actually lets multiple containers bind the same port by rewriting bind and source addresses… | ||
| Evolution of Direct Server Return (DSR) implementation for containerized applications | Lalit Gupta | 25m |
Meta replaces IPVS with BPF XDP and reports handling ~3x more traffic using ~7x less CPU, then builds XDP Chainer to let multiple teams attach ordered XDP… | ||
| SYN Proxy at Scale with BPF | Kuniyuki Iwashima | 23m |
A stateless SYN proxy is implemented to protect multi‑terabit services by validating custom SYN cookies at the edge and avoiding per‑connection state in the… | ||
| bpfilter: a BPF-based packet filtering framework | Quentin Deslandes | 33m |
A userspace daemon plus tiny lib on GitHub receives iptables-legacy ipt_replace structs or netlink NFtables messages and generates BPF bytecode into one… | ||
| Blinking Lights getting it wrong again again and again | Andrew Lunn | 17m |
RJ45 LEDs often end up configured by brittle device tree hacks (Marvel’s 2010 register-write example), and many patches get NACKed for encoding vendor register… | ||
| Syzbot: 7 years of continuous kernel fuzzing | Aleksandr Nogikh | 37m |
Syzbot runs syzkaller across 25 instances on arm64 and x86 VMs and reports roughly 1,000 new findings per year; since 2017 it records 17,000 findings, 6,000… | ||
| Linux Virtualization Based Security (LVBS) | Thara Gopinath | 50m |
Proposal and prototypes for Linux Virtualization Based Security (LVBS) implement hypervisor-backed kernel protections with a hypervisor-agnostic common layer… | ||
| Is Linux Suspend ready for the next decade | Len Brown | 47m |
A decade-old tool called sleepgraph runs local Python traces and browser reports to measure suspend and resume latencies, Wi‑Fi reconnect times, and per‑device… | ||
| Encryption for filesystems with advanced features new fscrypt functionality | Sweet Tea Dorminy | 33m |
fscrypt gains per-extent encryption with a key stored in each extent, enabling reflink and snapshot-safe encryption rather than inode-only keys. btrfs patches… | ||
| Trust, confidentiality, and hardening: the virtio lessons | Michael S. Tsirkin | 45m |
Virtio drivers get concrete hardening: core vring and device code are audited and now validate inputs (example: block driver rejects zero vring queues), and… | ||
| Using hardware hints for optimal page placement | Bharata Bhasker Rao | 45m |
Kernel consumes AMD EPYC IBS instruction-sampling to get precise physical addresses and data-source (cache/DRAM, local vs remote), driving NUMA balancing and… | ||
| Linux perf tool metrics | Ian Rogers | 45m |
Linux perf now ships hundreds of metrics (Valkyrie contributions) encoded as JSON and compiled into perf via create_perf_json.py, letting metric expressions… | ||
| Ship your Critical Section, Not Your Data: Enabling Transparent Delegation with... | Vishal Gupta | 44m |
Proposes transparent delegation that encodes a waiter’s critical section on its stack and runs it on the lock holder via a lightweight context switch… | ||
| Kernel handling of CPU and memory hot un/plug events for crash | Eric DeVolder | 31m |
Kernel now registers a generic crash notifier on CPU and memory notifiers and updates only the affected elfcorehdr segments in-kernel instead of triggering… | ||
| A kernel documentation update | Jonathan Corbet | 59m |
Architecture docs move into Documentation/arch/ and about 30–40 device docs are slated to move next, reducing ~100 top-level entries; Rust docs are integrated… | ||
| Tips and Strategies for Reducing Stress and Burnout by Creating Psychological... | Gloria Chance | 46m |
Psychological safety cuts stress by building trust and empathy and by using concrete practices: breath breaks, short cross-crawls, the lazy eight eye‑trace… | ||
| DAMON Current Status and Future Plans | SeongJae Park | 46m |
DAMON provides page-granularity sampling (defaults: one page per region, NR=10, sample=5ms, aggregation=100ms) and DAMON-based operation schemes (DAMOS) that… | ||
| Kernel Samepage Merging (KSM) at Meta and Future Improvements to KSM | Stefan Roesch | 43m |
KSM is used at Meta to deduplicate millions of anonymous pages and saved ~6 GB on a 64 GB node (up to ~5 million pages in stress tests). Installation uses a… | ||
| VSOCK: From Convenience to Performant VirtIO Communication | Amery Hung | 38m |
VSOCK gains datagram support and shows concrete wins: in virtio-over-vhost tests (8 vCPUs, one sender thread per vCPU, 1 RX/1 TX queue per CPU, equal buffer… | ||
| Improving resource ownership and life-time in linux device drivers | Bartosz Golaszewski | 53m |
Many subsystems mix reference‑counted resources with regular data and embed implementation structs inside driver private data, leading to use‑after‑free and… | ||
| How to make syzbot reports easier to debug? | Aleksandr Nogikh | 53m |
Syzbot produces hundreds of automated kernel bug reports a year and the hard facts that speed triage are: kdump memory dumps (useful for postmortem but large… | ||
| UEFI Setvariable at runtime -- Problems status and solutions | Ilias Apalodimas | 40m |
A patchset enables SetVariable at runtime by sandboxing standalone MM inside OP-TEE, using the eMMC RPMB as tamper‑resistant backing store, and running a… | ||
| Embedded Linux BOF | Tim Bird | 49m |
Group discussion catalogs concrete wins and pain points: prioritized boot time, system size, testing and build systems; practical tactics include U-Boot HTTP… | ||
| PCI device authentication encryption | Jonathan Cameron | 01h 00m |
A proof-of-concept in-kernel CMA‑SPDM RFC→non‑RFC patch set now aims to make the PCI core perform device authentication before driver probe, exposing an… | ||
| Secure IO BoF | Dhaval Giani | 55m |
Discussion argues for a single Linux guest‑to‑host ABI that maps generic verbs (notably info verb and a guest‑initiated rebind), with per‑vendor hypercall… | ||
| Android BOF | Kalesh Singh | 14m |
Reclaim for the multi‑generation (mgu) LRU stalls when pages remain in a movable Zone one index larger than the requested zone, preventing the oldest… | ||
| Android: memcg v1 - v2 | T.J. Mercier | 08m |
Upgrade effort moves Android memcg from v1 to v2 to add real dma-buf accounting using a per-CPU counter that flushes to global totals, plus kernel-initiated… | ||
| Reporting and tracking regressions across the ecosystem | Thorsten Leemhuis | 34m |
Mail-list reports become tracked by sending a rebot command in mail, then the system watches replies, patch submissions and link and closes tags to mark fixes… | ||
| 16k page size support | Juan Yescas, Kalesh Singh | 18m |
Android moves to 16 KB base pages and hits two concrete barriers: a USB host controller rejects transfers larger than 4 KB (workaround requires roughly 20… | ||
| Testing Drivers with KUnit (Does hardware have to be hard?) | David Gow | 28m |
KUnit is the in-kernel unit test framework (since Linux 5.5) and now hosts almost 1,000 upstream tests. The talk lays out concrete tooling: static stubbing /… | ||
| toolchain-agnostic build time improvements | Tanzir Hasan, Nick Desaulniers | 23m |
Automated header refactoring plus include-what-you-use (IWYU) and selective precompiled headers aim to cut front‑end work by removing indirect and dead… | ||
| nothing matches | ||