Exploring Parallel Linux Kernel Execution Using kexec
328 | Thu 31 Jul 3 p.m.–3:45 p.m.
Presented by
-
Cong Wang
https://wangcong.org
Cong Wang is a professional Linux kernel developer mainly focuses on networking and eBPF, he is also a Linux kernel maintainer for the networking traffic control subsystem. He has contributed over 1000 patches to Linux kernel.
Cong Wang
https://wangcong.org
Abstract
This presentation introduces an experimental approach that repurposes the Linux kexec mechanism to enable parallel execution of multiple Linux kernels on a single physical system. While kexec has traditionally been used for fast kernel switching and crash recovery, our prototype demonstrates the feasibility of using this technology to create multiple lightweight Linux kernels. The implementation establishes isolated execution environments through careful memory partitioning and hardware resource allocation, allowing concurrent kernel instances to operate independently while maintaining communication channels between kernels. Our proof-of-concept prototype successfully validates the core functionality of parallel kernel execution, including the challenges of memory management, interrupt handling, and device sharing strategies. This talk will explore the technical implementation details, current limitations, and potential future directions for this approach.
This presentation introduces an experimental approach that repurposes the Linux kexec mechanism to enable parallel execution of multiple Linux kernels on a single physical system. While kexec has traditionally been used for fast kernel switching and crash recovery, our prototype demonstrates the feasibility of using this technology to create multiple lightweight Linux kernels. The implementation establishes isolated execution environments through careful memory partitioning and hardware resource allocation, allowing concurrent kernel instances to operate independently while maintaining communication channels between kernels. Our proof-of-concept prototype successfully validates the core functionality of parallel kernel execution, including the challenges of memory management, interrupt handling, and device sharing strategies. This talk will explore the technical implementation details, current limitations, and potential future directions for this approach.