PRs worth fixing

(carried over unclosed PRs from netbsd8)

• PR #53053: non-MULTIPROCESSOR hangs building Go (still reproducable with newer go?)
  PR #53173: "go test net/http" locks up the machine (both PRs tied )
• PR #53016: Clock not stable
• PR #53017: Kernel panics every now and then with "fpusave_lwp: did not" message
• PR #53155: Wedge after <12h uptime when >2 bnx network interfaces in use
• PR #53161: ATF test runs leave a rump_server process around
• PR #53286: hdaudio(4), iwm(4) and rtsx(4) detections fail in UEFI case
• PR #53291: GPT prevents installation

Regression since netbsd8:

• PR #53128: Boot says "default boot twice, skipping..."
• PR #54074: Test cases ending in 32 fail on sparc64 since jemalloc update
• PR #54089: vax installer does not boot (init dies)
• PR #54093: Recent test regressions on sparc
• PR #54300: lib/libutil/t_snprintb test regression on sparc
• PR #54345: kernel diagnostic assertion "xfer != NULL" in wdc.c
• PR #54503: Panic during attaching nvme(4) when # of logical CPUs >= 32 ?
• PR #54727: writing a large file causes unreasonable system behaviour (may not be a regression, but became more noticable with -9)
• PR #54761: nvme corruption on GENERIC without DIAGNOSTIC
• PR #54771: nvme(4) fails to attach

Bug Bountys

PRs in the above list marked with are likely hard and critical, and have a bug bounty of $100 attached. If you find a solution for one of the marked issues that leads to closing of the PRs, we will get in touch with you.

Features to finish/implement

• IPF/PF deprecation in favour of NPF
  • TODO: docs, what features to add to NPF ...
• ATA/NCQ
  • variants of PR #48214 "clearing WDCTL_RST failed" during boot - problem still present on some configurations
  • PR #52783 parallel fsck hangs during boot of 8.99.[5678] (done)
  • PR #52614 qemu virtual CD-ROM reports read errors since recent wdc changes (done)
  • PR #53183 System stops servicing I/O requests and eventually deadlocks (done)
• UBC using direct map - removal of pmap_enter()/pmap_remove()/pmap_update() for read()/write()-based I/O
  • PR #54759 (vm.ubc_direct deadlock when read()/write() into mapping of itself)
  • PR #53124 (FFS is slow because pmap_update doesn't scale)
  • confirmed working on amd64, alpha, aarch64
  • need sysctl to more easily turn on, will be left off by default for 9.0 (done)
• XEN AVX working
  • PR #50332 AVX instructions don't work but OSXSAVE flag is set (done)
• ZFS
• kASan for amd64 (done)
• kASan for aarch64 (done)
• compiler-rt in the base with enabled features for Clang/LLVM (done)
• LVM mirror target
• add NVMM (done)
• plug kernel pointer leaks (done)
• fix Nouveau, some GPUs stopped working after the update (done)
• umb(4) confirmed working with network traffic

Nice to have

• arm: GENERIC kernel config for armv7 boards. (done)
• arm64: Support for servers following SBSA/SBBR standards (done)
• updated graphics drivers (done)
• PVHVM and PVH
• ARM64 ACPI (done)
• WireGuard
• expand x86 intr masks to allow up to 56 interrupts per CPU and hence 40 MSI/MSI-X on cpu0 http://mail-index.netbsd.org/current-users/2018/12/01/msg034728.html

Nice to have, but likely won't be finished on time

• merge pgoyette-compat (done)
• address failing ptrace(2) tests, cover the remaining scenarios (there are planned ~2k ATF tests total for the elementary functionality) (done in the scope for 9.0)
• rework semantics of threading operations in ptrace(2) (done)

Abandoned for 9.0

• gcc 8+ from a sanitizers point of view (abandoned)

Current status and timeline

• release candidate 1 is available https://www.netbsd.org/releases/formal-9/
• several important fixes queued for pullup after RC1, releng decided to do a second RC
• expecting RC2 around mid of december 2019
• final release hopefully (very) early january 2020