LWN Coverage of Linux Kernel and Hiring of Daroc Alden
-
LWN ☛ Kicinski: netdev in 2023
Networking maintainer Jakub Kicinski (along with several collaborators) has put up a summary of what happened in the kernel's network stack during 2023.
-
Linux ☛ netdev in 2023
The end of the Linux v6.2 merge coincided with the end of 2022, and the v6.8 window had just begun, meaning that during 2023 we developed for 6 kernel releases (v6.3 – v6.8). Throughout those releases netdev patch handlers (DaveM, Jakub, Paolo) applied 7243 patches, and the resulting pull requests to Linus described the changes in 6398 words. Given the volume of work we cannot go over every improvement, or even cover networking sub-trees in much detail (BPF enhancements… wireless work on WiFi 7…). We instead try to focus on major themes, and developments we subjectively find interesting.
-
LWN ☛ Please welcome Daroc Alden
When, at the beginning of November, we posted an open position at LWN, we were only so hopeful; experience has shown that finding writers who are both capable of and interested in writing our sort of material is a challenging task. This time, though, hope was justified: we got a surprising number of applications from highly qualified applicants. The hardest part of the task has, instead, been narrowing down the choice to a hiring decision.
-
LWN ☛ Kernel-text replication on NUMA systems
Kernel developers often go out of their way to reduce the memory used by the kernel itself; that memory is not available for the workloads that people are actually interested in running on their systems. Lower memory usage also tends to lead to better performance overall. But there are times when the expenditure of some extra memory can make the system faster. The replication of the kernel's text (executable code) and read-only data across a NUMA system may be a case in point; patch sets have been posted adding that capability to two architectures.
Non-uniform memory access (NUMA) systems are divided into nodes, each of which normally contains a number of CPUs and some memory. All of the memory in the system is addressable from any CPU, but access to memory attached to the local node will be significantly faster than access to memory on remote nodes. For this reason, the kernel goes out of its way to try to keep workloads and their memory on the same nodes, and there is an extensive set of system calls allowing applications to control where in the system their memory is placed.
-
LWN ☛ Some 6.7 development statistics
The 6.7 kernel was released on January 7 after a ten-week development cycle. This was, as it turns out, the busiest cycle ever with regard to the number of changesets merged. The time has come for our usual look at where all those changesets came from, with a side trip into how long kernel developers tend to stick around.
The 6.7 kernel saw the addition of 17,284 non-merge changesets to the mainline — the most for any development cycle ever. To a great extent, this volume is due to the merging of the bcachefs filesystem along with its full development history; bcachefs was responsible for nearly 3,000 commits in 6.7. Otherwise, it was a fairly normal cycle, incorporating the work of 1,973 developers (267 of whom made their first contribution in 6.7) and growing the kernel by 566,000 lines of code.