On 4/11/22 17:50, Tomoaki AOKI wrote:
On Mon, 11 Apr 2022 20:18:48 +0200
Ronald Klop <ronald-li...@klop.ws> wrote:
On 4/11/22 17:17, Dennis Clarke wrote:
Did the usual git pull origin main and buildworld/buildkernel but after
installkernel the machine will not boot.
The rev seems to be main-n254654-d4e8207317c.
I can boot single user mode and get a command prompt but nothing past
that. Is there something borked in ZFS in CURRENT ?
Up until now you are the only one with this error on the mailinglist today. So
I doubt something is borked.
You could consider to share more details about your setup to help people to
think along with you.
Regards,
Ronald.
I have main at git c79331a42c308139828c1117f49224bb83617a53 booting
fine, and no commits relatd with ZFS exists within git d4e8207317c.
I was just looking at that and there is one :
root@phobos:/usr/src #
root@phobos:/usr/src # /usr/local/bin/git --no-pager log -n 32
--pretty=oneline --abbrev-commit --graph
* d4e8207317c (HEAD -> main, origin/main, origin/HEAD)
vmm_instruction_emul.c: fix bhyve build
* be0d16b0b05 bsdinstall: filter out disks that are unavailable from the
list of options in ZFS
* 5580e5bd716 nfscl: Clean up the code by removing unused arguments
* 5a17f489d58 vmm: fix set but not used warning
* 5241577a223 vmm: fix set but not used warning
* 3587bfa797c vmm: fix set but not used warning
* 5c272efaba2 vmm: fix set but not used warnings
* f877977a034 vmm: fix set but not used warnings
* 893a3dd697e vmm: fix set but not used warning
* f3ef799f564 Only return a mapped address from efi_phys_to_kva
* 57e47ae514b Include the EFI Runtime Code in the DMAP
* bde57090337 UPDATING: Fix a few typos
* c79331a42c3 bhyve: use linker set for ipc commands
* 38c3cf6aede nfscl: Clean up the code by removing unused arguments
* c45d934f6b7 nfscl: Ansify a function header
* bd8701dede1 Document procstat(1) advlock command
* a5229a255ea Implement procstat(1) advlocks command
* e79866ddf1c procstat(1): add ability to specify subcommands not
requiring pid lists
* 50d3c72558f libprocstat: document procstat_getadvlock(3)
* 039d1496b07 libprocstat: add procstat_getadvlock(3)
* eca39864f70 Add sysctl KERN_LOCKF
* 6ead1379fd4 sys/user.h: Add kinfo_lockf structure to report advisory locks
* 147e4fe3f1f kern_lockf.c: remove no longer neeeded UFS headers
* 59e85819be6 lockf: remove lf_inode from struct lockf_entry
* 5c075d64049 ufs/acl.h: forward-declare struct inode
* 8cc19b1e47d Style.
* a3214fbe7ff mount: use pidfile_signal
* 287451fd019 pidfile: add pidfile_signal
* ecbdfbfd18d netgraph(3): Remove a double word in a source code comment
* d048e8c6196 ofed: Fix a typo in a source code comment
* 299fcf402dc fsck_ffs(8): Fix a typo in a source code comment
* 009727ed577 routed(8): Remove a double word in a source code comment
root@phobos:/usr/src #
I see be0d16b0b05 bsdinstall: filter out disks that are unavailable from
the list of options in ZFS
Not sure what that does however.
I am looking at :
root@phobos:/usr/src # git pull origin main
remote: Enumerating objects: 100, done.
remote: Counting objects: 100% (16/16), done.
remote: Compressing objects: 100% (3/3), done.
remote: Total 100 (delta 13), reused 13 (delta 13), pack-reused 84
Receiving objects: 100% (100/100), 189.37 KiB | 1.38 MiB/s, done.
Resolving deltas: 100% (57/57), completed with 12 local objects.
From git.freebsd.org:src
* branch main -> FETCH_HEAD
d4e8207317c..673bce11ced main -> origin/main
Updating d4e8207317c..673bce11ced
Fast-forward
lib/libc/sys/getdirentries.2 | 2 ++
sbin/ifconfig/ifconfig.8 | 5 +++--
stand/man/loader_lua.8 | 47
+++++++++++++++++++++++------------------------
sys/compat/linprocfs/linprocfs.c | 20 ++++++++++++++++++++
sys/compat/linux/linux_socket.c | 38
+++++++++++++++++++++++++++++---------
sys/compat/linux/linux_socket.h | 1 +
sys/dev/axgbe/if_axgbe_pci.c | 65
++++++++++++++++++++++++++++++++++++++++++++++++++++-------------
sys/kern/vfs_subr.c | 1 +
sys/kern/vfs_syscalls.c | 4 ++++
sys/netinet/ip_mroute.c | 26 ++++++++++++++++++++------
sys/netinet/ip_output.c | 2 +-
sys/netinet6/ip6_output.c | 2 ++
sys/netpfil/ipfw/ip_fw2.c | 27 ++++++++++++---------------
sys/netpfil/ipfw/ip_fw_log.c | 7 +++++--
sys/netpfil/pf/pf_ioctl.c | 2 ++
sys/sys/vnode.h | 2 +-
sys/vm/swap_pager.c | 4 ++--
17 files changed, 180 insertions(+), 75 deletions(-)
root@phobos:/usr/src # /usr/local/bin/git --no-pager log -n 32
--pretty=oneline --abbrev-commit --graph
* 673bce11ced (HEAD -> main, origin/main, origin/HEAD) linux(4): Copyout
actual size of addr to the user space in accept().
* bb0f644cd68 linux(4): Limit user-supplied sockaddr length in recvfrom().
* 68bfaefb3d9 linux(4): Remove unnecessary PTRIN().
* cf312f799a8 linux(4): Handle SO_DOMAIN in getsockopt syscall.
* c6487446d7e getdirentries: return ENOENT for unlinked but still open
directory.
* bb46e9b5107 linux(4): Prevent an attempt to copy an uninitialized
source address.
* 6ca0ca7b4cb IPv4 multicast: fix LOR in shutdown path
* 8e458a431eb Clean up some grammos I left behind.
* 67f5810e07c Correct typos and more precise wording.
* 632ea8ea984 ifconfig.8: Note that -l accepts -g in addition to -d and -u
.
.
.
.
So again nothing from yesterday until today that says a word about ZFS.
So the info at which commit was running before update would be needed.
I've encountered the situation like that, but it was far before OpenZFS
was introduced into base. If it's the same issue, /sbin/zfs SHALL be
updated in conjunction with kernel.
Right. I have seen it also but that was a long long time ago. Relatively
speaking and OpenZFS has never been an issue. I wish I had a serial
interface on the little laptop to boot to single usermode with and then
can look around and capture the output.
If you can boot with any emergency environment (memstick or another
installation) and import now-non-bootable pool with e.g. `zpool import
-R /mnt -f ZPOOL`,
I have been through that before and usually it was just to import a
strange zpool from some other storage array onto some other machine.
I don't think that is needed here.
Beware! If other updated components are mandatory for /sbin/zfs, the
above procedure is not at all enough. But would crash before actually
the pool is imported.
Yep. Let's avoid that.
I am going to see if I can build to commit 3587bfa797c on some other
hardware and have a look. Just because I am curious.
--
Dennis Clarke
RISC-V/SPARC/PPC/ARM/CISC
UNIX and Linux spoken
GreyBeard and suspenders optional
