@Not_Oles said:
* retiring OpenBSD on this server, for the moment, to free up more space
that's something I am planning to do soon.
That's done now - resizing a NetBSD filesystem is always a it fiddly, as you first have to make sure to remove to filesystem journal, then resize the filesystem, and finally re-create the journal again (if you are using wapbl).
I have installed qemu (with nvmm) to run a test NetBSD VM to see if I can reproduce the virtio issue I am seeing on the linveo VPS.
I looks like it could be a memory ordering issue in the virtio communication between the guest and the host, as adding an additional mfence in the driver code seems to fix it.
@cmeerw said:
I have installed qemu (with nvmm) to run a test NetBSD VM to see if I can reproduce the virtio issue I am seeing on the linveo VPS.
I looks like it could be a memory ordering issue in the virtio communication between the guest and the host, as adding an additional mfence in the driver code seems to fix it.
@cmeerw said:
I have installed qemu (with nvmm) to run a test NetBSD VM to see if I can reproduce the virtio issue I am seeing on the linveo VPS.
I looks like it could be a memory ordering issue in the virtio communication between the guest and the host, as adding an additional mfence in the driver code seems to fix it.
Been pretty busy the past week or so with other stuff - mainly replacing my 10 year old low-end laptop with a shiny new low-end laptop. With a 6.17.0 rc5 Linux kernel, even hibernate seems to work fine now.
Other than that, the virtio driver has been fixed on NetBSD HEAD now, waiting to make it's way into 9, 10, and 11 branches.
Went for the Chuwi Corebook X 7430U, really seems like amazing value for money, and works absolutely fine under Linux (Ubuntu 25.04 with an updated kernel).
Went for the Chuwi Corebook X 7430U, really seems like amazing value for money, and works absolutely fine under Linux (Ubuntu 25.04 with an updated kernel).
As a past Chuwi user, I support your choice! I had their dual booting windows/android tablet. Lasted longer then my old lenovo yoga tablet at just over 1.5 years. It died cause I let the battery drain too low and forgot to charge it for a few weeks... When I got back to it, it was dead and would not take a charge. Hopefully they have fixed that by now.
Just thought I'd give LVM a try on NetBSD. Note that approximately half of the disk space is used by Debian (most of it on logical volumes - and there is a bit of free space on the volume).
NetBSD can actually work with the logical volumes created by Linux LVM. So I thought I just use some of the remaining free space to create a test volume for NetBSD and create a new filesystem in it:
newfs -O2ea /dev/M247-X10E-9N-vg/rnetbsd-test
/dev/M247-X10E-9N-vg/rnetbsd-test: 2048.0MB (4194304 sectors) block size 16384, fragment size 2048
using 12 cylinder groups of 170.67MB, 10923 blks, 21184 inodes.
super-block backups (for fsck_ffs -b #) at:
160, 349696, 699232, 1048768, 1398304, 1747840, 2097376, 2446912, 2796448,
...............................................................................
Ok, now we have a new filesystem for NetBSD to use.
Actually, I probably should have enabled lvm=YES (in /etc/rc.conf) and/or started /etc/rc.d/lvm first to make sure all the existing volumes get discovered. Anyway, seemed to work fine without it (but I did do that later).
Mounting that file system also worked fine:
mount /dev/mapper/M247--X10E--9N--vg-netbsd--test /mnt/test
So I then put it into /etc/fstab and rebooted. But the server didn't come back up. So what went wrong?
In /etc/fstab I added rw,log as the options (so we get wapbl meta-data journaling), and that appears to be an issue on a logical volume:
mount -o rw,log /dev/mapper/M247--X10E--9N--vg-netbsd--test /mnt/test
mount_ffs: /dev/mapper/M247--X10E--9N--vg-netbsd--test on /mnt/test: Inappropriate ioctl for device
So somehow that seems to be incompatible for logical volumes. Need to do some investigation...
@cmeerw said:
Just thought I'd give LVM a try on NetBSD. Note that approximately half of the disk space is used by Debian (most of it on logical volumes - and there is a bit of free space on the volume).
NetBSD can actually work with the logical volumes created by Linux LVM. So I thought I just use some of the remaining free space to create a test volume for NetBSD and create a new filesystem in it:
newfs -O2ea /dev/M247-X10E-9N-vg/rnetbsd-test
/dev/M247-X10E-9N-vg/rnetbsd-test: 2048.0MB (4194304 sectors) block size 16384, fragment size 2048
using 12 cylinder groups of 170.67MB, 10923 blks, 21184 inodes.
super-block backups (for fsck_ffs -b #) at:
160, 349696, 699232, 1048768, 1398304, 1747840, 2097376, 2446912, 2796448,
...............................................................................
Ok, now we have a new filesystem for NetBSD to use.
Actually, I probably should have enabled lvm=YES (in /etc/rc.conf) and/or started /etc/rc.d/lvm first to make sure all the existing volumes get discovered. Anyway, seemed to work fine without it (but I did do that later).
Mounting that file system also worked fine:
mount /dev/mapper/M247--X10E--9N--vg-netbsd--test /mnt/test
So I then put it into /etc/fstab and rebooted. But the server didn't come back up. So what went wrong?
In /etc/fstab I added rw,log as the options (so we get wapbl meta-data journaling), and that appears to be an issue on a logical volume:
mount -o rw,log /dev/mapper/M247--X10E--9N--vg-netbsd--test /mnt/test
mount_ffs: /dev/mapper/M247--X10E--9N--vg-netbsd--test on /mnt/test: Inappropriate ioctl for device
So somehow that seems to be incompatible for logical volumes. Need to do some investigation...
That appears to be something that got broken in the NetBSD 11 branch. I have done some analysis and filed PR #59674
Thanks @Hosteroid for the nice server, allowing me to test NetBSD 11.0_BETA.
Comments
That's done now - resizing a NetBSD filesystem is always a it fiddly, as you first have to make sure to remove to filesystem journal, then resize the filesystem, and finally re-create the journal again (if you are using wapbl).
(there is still Debian installed using the other half of the disk)
Thanks @cmeerw!
Much appreciated! 
Thanks to Hosteroid for our nice community server! Which is now dual booting Debian and NetBSD. . . .
Anybody have a project and want to jump on? Please read "How To Apply" in the OP, and then come and join us!
I hope everyone gets the servers they want!
I have installed qemu (with nvmm) to run a test NetBSD VM to see if I can reproduce the virtio issue I am seeing on the linveo VPS.
I looks like it could be a memory ordering issue in the virtio communication between the guest and the host, as adding an additional mfence in the driver code seems to fix it.
@cmeerw How's it going?
Anybody else want to share this fine server?
Thanks @Hosteroid!
I hope everyone gets the servers they want!
Been pretty busy the past week or so with other stuff - mainly replacing my 10 year old low-end laptop with a shiny new low-end laptop. With a 6.17.0 rc5 Linux kernel, even hibernate seems to work fine now.
Other than that, the virtio driver has been fixed on NetBSD HEAD now, waiting to make it's way into 9, 10, and 11 branches.
Sorry, can't resist asking which one, please?
I hope everyone gets the servers they want!
Went for the Chuwi Corebook X 7430U, really seems like amazing value for money, and works absolutely fine under Linux (Ubuntu 25.04 with an updated kernel).
As a past Chuwi user, I support your choice! I had their dual booting windows/android tablet. Lasted longer then my old lenovo yoga tablet at just over 1.5 years. It died cause I let the battery drain too low and forgot to charge it for a few weeks... When I got back to it, it was dead and would not take a charge. Hopefully they have fixed that by now.
Never make the same mistake twice. There are so many new ones to make.
It’s OK if you disagree with me. I can’t force you to be right.
Greetings from our lovely Community Server, kindly donated by Hosteroid!
If you want to jump on the server, please read the OP section about applying, and please get in touch! Thank you!
Best wishes and kindest regards,
Tom
I hope everyone gets the servers they want!
Just thought I'd give LVM a try on NetBSD. Note that approximately half of the disk space is used by Debian (most of it on logical volumes - and there is a bit of free space on the volume).
NetBSD can actually work with the logical volumes created by Linux LVM. So I thought I just use some of the remaining free space to create a test volume for NetBSD and create a new filesystem in it:
Great, that worked.
Ok, now we have a new filesystem for NetBSD to use.
Actually, I probably should have enabled
lvm=YES(in/etc/rc.conf) and/or started/etc/rc.d/lvmfirst to make sure all the existing volumes get discovered. Anyway, seemed to work fine without it (but I did do that later).Mounting that file system also worked fine:
So I then put it into
/etc/fstaband rebooted. But the server didn't come back up. So what went wrong?In
/etc/fstabI addedrw,logas the options (so we get wapbl meta-data journaling), and that appears to be an issue on a logical volume:So somehow that seems to be incompatible for logical volumes. Need to do some investigation...
Awesome!
I hope everyone gets the servers they want!
That appears to be something that got broken in the NetBSD 11 branch. I have done some analysis and filed PR #59674
Thanks @Hosteroid for the nice server, allowing me to test NetBSD 11.0_BETA.