F
Florine Kamdem
Guest
Original article here.
Consider this when replying.
I’m very grateful that the FreeBSD Foundation sponsored my trip to AsiaBSDCon 2026 in Taipei, Taiwan. The conference ran from March 19 through 22, with the first two days dedicated to the FreeBSD Developer Summit and the final two days to the main conference. It was an incredibly productive experience, and I’m glad I had the opportunity to attend.
The Developer Summit ran for two days, 10 AM to 5 PM each day. Most sessions were short presentations covering topics like FreeBSD on WSL 2 and Ansible on FreeBSD. Between talks, developers worked on patches, followed up on mailing-list threads, and reviewed one another’s work. I found this format extremely valuable as I was able to get useful feedback on my LLVM pull requests during these gaps, which would have taken much longer to sort out over email. I had a productive conversation with Shingyi Hung (aokblast@) about my LLDB kernel debugging pull requests, and his review comments helped me refine my approach going forward.
The main conference began on Saturday and drew roughly three to four times as many attendees as the DevSummit. Kenichi Yasukata’s talk on a portable userland TCP/IP stack was fascinating — he outlined how separating OS-specific glue from the core stack allows it to remain portable
while still achieving strong performance. He also discussed techniques for intercepting system calls without relying on LD_PRELOAD, using binary overwriting with architecture-specific hooks. Pierre Pronchery’s demonstration of smolBSD was equally impressive, showing container-oriented NetBSD images that boot in under one second by stripping the kernel and bypassing the traditional UEFI boot path.
On Sunday, George Neville-Neil gave a thought-provoking talk on runtime reoptimization for modern heterogeneous hardware. His core argument — that Unix-like systems still carry assumptions rooted in the PDP-11 era — resonated with me, and his approach of feeding live profiling data back into LLVM to produce architecture-specific binaries is something I plan to revisit once the recording is available. Brooks Davis presented the CheriBSD upstreaming effort, with a target of FreeBSD 16 in December 2027. The security demonstrations were compelling. CHERI caught all 13 memory bugs intentionally injected into NGINX, plus two additional ones introduced unintentionally. At lunch, I had the chance to speak with Olivier (olce@) about designing a hybrid scheduler and IntelHFI patch review, which is directly relevant to the hybrid scheduling work I’ve been doing for FreeBSD.
As helpful as mailing lists and code review platforms can be, they don’t compare to the quality of communication you can have with someone in person. Being able to sit down with Olivier and Shingyi to discuss my ongoing work face-to-face was far more productive than weeks of asynchronous back-and-forth would have been.
During the closing session, Li-wen Hsu announced that AsiaBSDCon 2027 will be held in Singapore, and the organizers are actively looking for volunteers and sponsors. I also want to note that I will be presenting a talk on hybrid scheduling at BSDCan 2026.
Conference schedule: https://2026.asiabsdcon.org/entry/schedule/
My blog recap: https://minsoo.io/asiabsdcon-2026-recap/
Thank you again to the Foundation for making this trip possible.
Respectfully,
Minsoo Choo
The post AsiaBSDCon 2026 Trip Report – Minsoo Choo first appeared on FreeBSD Foundation.
Continue reading...
Consider this when replying.
I’m very grateful that the FreeBSD Foundation sponsored my trip to AsiaBSDCon 2026 in Taipei, Taiwan. The conference ran from March 19 through 22, with the first two days dedicated to the FreeBSD Developer Summit and the final two days to the main conference. It was an incredibly productive experience, and I’m glad I had the opportunity to attend.
The Developer Summit ran for two days, 10 AM to 5 PM each day. Most sessions were short presentations covering topics like FreeBSD on WSL 2 and Ansible on FreeBSD. Between talks, developers worked on patches, followed up on mailing-list threads, and reviewed one another’s work. I found this format extremely valuable as I was able to get useful feedback on my LLVM pull requests during these gaps, which would have taken much longer to sort out over email. I had a productive conversation with Shingyi Hung (aokblast@) about my LLDB kernel debugging pull requests, and his review comments helped me refine my approach going forward.
The main conference began on Saturday and drew roughly three to four times as many attendees as the DevSummit. Kenichi Yasukata’s talk on a portable userland TCP/IP stack was fascinating — he outlined how separating OS-specific glue from the core stack allows it to remain portable
while still achieving strong performance. He also discussed techniques for intercepting system calls without relying on LD_PRELOAD, using binary overwriting with architecture-specific hooks. Pierre Pronchery’s demonstration of smolBSD was equally impressive, showing container-oriented NetBSD images that boot in under one second by stripping the kernel and bypassing the traditional UEFI boot path.
On Sunday, George Neville-Neil gave a thought-provoking talk on runtime reoptimization for modern heterogeneous hardware. His core argument — that Unix-like systems still carry assumptions rooted in the PDP-11 era — resonated with me, and his approach of feeding live profiling data back into LLVM to produce architecture-specific binaries is something I plan to revisit once the recording is available. Brooks Davis presented the CheriBSD upstreaming effort, with a target of FreeBSD 16 in December 2027. The security demonstrations were compelling. CHERI caught all 13 memory bugs intentionally injected into NGINX, plus two additional ones introduced unintentionally. At lunch, I had the chance to speak with Olivier (olce@) about designing a hybrid scheduler and IntelHFI patch review, which is directly relevant to the hybrid scheduling work I’ve been doing for FreeBSD.
As helpful as mailing lists and code review platforms can be, they don’t compare to the quality of communication you can have with someone in person. Being able to sit down with Olivier and Shingyi to discuss my ongoing work face-to-face was far more productive than weeks of asynchronous back-and-forth would have been.
During the closing session, Li-wen Hsu announced that AsiaBSDCon 2027 will be held in Singapore, and the organizers are actively looking for volunteers and sponsors. I also want to note that I will be presenting a talk on hybrid scheduling at BSDCan 2026.
Conference schedule: https://2026.asiabsdcon.org/entry/schedule/
My blog recap: https://minsoo.io/asiabsdcon-2026-recap/
Thank you again to the Foundation for making this trip possible.
Respectfully,
Minsoo Choo
The post AsiaBSDCon 2026 Trip Report – Minsoo Choo first appeared on FreeBSD Foundation.
Continue reading...