NetBSD-10: write to rw null mount on top of LFS fs hangs up

Hi. I configured a pkgsrc bulk build on NetBSD-10 virtual machine and
selected LFS file system for temporary directory where object files are
created.

I know the status of LFS is "unknown" but this is not a production
machine. So, I just wanted to play with LFS a little bit.

Virtual machine was run on Linux/KVM as the following:

qemu-system-x86_64 -accel kvm \
    -smp cpus=12 \
    \
    -drive file=netbsd-10.qcow2,if=none,id=hd0 \
    -device virtio-blk-pci,drive=hd0 \
    \
    -drive file=netbsd-10_srv.qcow2,if=none,id=hd1 \
    -device virtio-blk-pci,drive=hd1 \
    \
    -drive file=netbsd-10_newsrv.qcow2,if=none,id=hd2 \
    -device virtio-blk-pci,drive=hd2 \
    \
    -object rng-random,filename=/dev/urandom,id=viornd0 \
    -device virtio-rng-pci,rng=viornd0 \
    \
    -netdev tap,id=vioif0 \
    -device virtio-net-pci,netdev=vioif0,mac=00:06:70:A3:85:FC \
    \
    -m 12288

Packages are built in 18 parallel subprocesses in chroots where /tmp/
directory is null mounted on top of LFS. So, IO was relatively high.
After a few hours of work writing to file system is stuck.

File system was created by newfs_lfs(8) without options.

Currently ps looks like the following

- /usr/bin/make -f Makefile -f /home/cheusov/pkg_distbb/etc/distbb.local.mk -f /home/cheusov/pkg_distbb/share/distbb/distbb.mk DEPENDS_TARGET=nonexistant BATCH=yes PKG_VERBOSE=1 build
`-- /bin/sh -c set -e;\t\t\t\t\t if test -n "" &&  /usr/sbin/pkg_info -K /usr/pkg/pkgdb -qe mariadb-server-10.11.7; then  echo ===\\> "Skipping installation of already handled package";  else
  `-- /usr/bin/make _MAKE=/usr/bin/make NATIVE_OPSYS=NetBSD NATIVE_OS_VERSION=10.0 NATIVE_OPSYS_VERSION=100000 HOST_MACHINE_ARCH=amd64 NATIVE_LOWER_OPSYS=netbsd _PKGSRCDIR=/srv/pkgsrc/current P
    `-- /usr/bin/make -f Makefile all
      `-- /usr/bin/make -s -f CMakeFiles/Makefile2 all
        `-- /usr/bin/make -s -f storage/maria/CMakeFiles/aria.dir/build.make storage/maria/CMakeFiles/aria.dir/build
          `-- /tmp/obj_pkgsrc/databases/mariadb1011-server/work/.gcc/bin/gcc -fcommon -fstack-protector-strong -D_FORTIFY_SOURCE=2 -Wl,-zrelro -fPIC -DHAVE_CONFIG_H -DWITH_S3_STORAGE_ENGINE -I/
            `-- /usr/libexec/cc1 -quiet -I /tmp/obj_pkgsrc/databases/mariadb1011-server/work/mariadb-10.11.7/wsrep-lib/include -I /tmp/obj_pkgsrc/databases/mariadb1011-server/work/mariadb-10.11


That's it. It seems that writing to LFS file system hangs up. System itself
is fully functional. Kernel didn't panic.

Is there anyone interested in debugging? I can provide all required
information.

mount:
/dev/ld2a on /srv2 type lfs (local)
/srv2/tmps/tmp1 on /srv/sandboxes/sandbox1-nb10/tmp type null (local)

df:
Filesystem      1K-blocks         Used        Avail %Cap Mounted on
/dev/ld2a       300911577     45415971    225401031  16% /srv2

uname:
NetBSD builder_nb10 10.0 NetBSD 10.0 (GENERIC) #0: Thu Mar 28 08:33:33
UTC 2024
[email protected]:/usr/src/sys/arch/amd64/compile/GENERIC amd64