sockstat -np 1234 Is there a reason this is more than items with that port? Using -4 or -6 limits it to what I would expect.

nothing about -p implies not showing non-internet sockets

and from man sockstat: "If neither -4, -6 or -u is specified, sockstat will list sockets in all three domains."

I'm still unhappy about!

i run top and RES is the physical memory actually being used by a process. but in top's summary at the top, there is no Res total in the Mem line. wtf?

I think that's because adding RES for all processes would count memory used by shared libraries multiple times. But I've been known to be wrong and my understanding of memory stats is weak.

^ is the correct answer, it's basically impossible to calculate memory usage of a process on a shared/virtual memory system. 'RES' is just a fudge that's sometimes useful

i do with top had more useful system memory statistics though

spent way to long trying to work out why poudriere was flagging a bunch of ports as ignore even though I have since ditched mysql-57, then finally realised I had set 5.7 as the default version in the make.conf for the jail....

hi all

Fresh freebsd 14.0 install at vmware pro , no display manager nor window manager. simply at console. nothing is installed additionally after the fresh installation. file system is UFS. 'top' shows 199M 'wired' memory. I edited the kernel config file and removed 'not needed' parts (devices such as NICs and WIFIs) and recompiled the kernel, booted up but still 'wired' memory is 199M and if i do a installation for example it goes up. but never

goes down. why a fresh install uses 200M by default ? and how can i lower or minimize the 'wired' memory ?

CountryBall0, I actually don't know how big the base 14 kernel is. I did a check on some 13.3 systems, x86 w/ zfs shows about 40M worth of kernel in memory, arm64 about 20M

usking kldstat as an estimate for the lower bounds of what you would expect in wired memory

userland can still be wired too though, so perhaps world has some things wired down too?

there is a huge amount of memory tuning options available though through various sysctl options. It could be that the defaults are less conservative than you'd have expected?

something like 'vmstat -h -m' might also help tell you where the memory is being used

it could be something like the network buffers using up some of this too?

edenist, thanks for the reply. yeah I was expecting those values (40-20M) for my setup too but unfortunately it 200M. there are no network activity tho so no network buffers. but memory tuning and kldstat I will be looking into (thanks) , about userland, hmm how to pratically identify/diagnose the userland in this aspect to see which userland programs are wired and how much wired ?

edenist, kldstat shows 7 other kernel modules as loaded. almost all of them related to vmware, virtualization. it seems I may discard the wired memory when in virtualized enviroment.. they cant be kldunloaded either. vmstat -h -m provided me interesting results, thanks again

is there a partition resizing tool for FreeBSD?

something like gparted Iguess?

dch: afaik gparted supports ufs

arino: FreeBSD has the strongest performance and has a lot of features (openzfs support, jails, strong virtualisation etc), its security is very Linux-like with support for MAC and containierisation/virtualisation. OpenBSD focuses on code correctness and security, they have rigid code standards and proactively audit their own codebase to ensure any rotting code is cut out... anything not required

is additional surface area for attack in their minds, and thus should be deleted. OpenBSD maintains a small, minimalistic, clean codebase whcih aims to be secure by design, however they do not agree with the mainstream Linux security ideas such as MAC (argued it simply is too complex and is switched off) and for containerisation its either chroot or unveil(). Whether you agree with their extreme

stance against the norm of "security", OpenSSH for example is shared between almost every major system in the world... and pf came from OpenBSD too... along with OpenSMTPD which seems to be a common choice of MTA

NetBSD focuses on hardware support, for example I believe they just announced RISCV support.....

If you want good support for hardware, NetBSD "just works" afaik

Yet to try NetBSD, I have used FreeBSD and OpenBSD only

drobban: I would need to install linux for that tho

you also have DragonflyBSD which diverged from FreeBSD due to arguments on optimising performance

iirdc

dch: gpart(8) and docs.freebsd.org/en/books/handbook/disks

Title: Chapter 20. Storage | FreeBSD Documentation Portal

polarian: I'm familiar with gpart, I'm looking for one that can handle resizing, not just extension thanks

ohhh

apologies I misunderstood

however gpart does support resize no?

gpart resize ...?

polarian: thanks

lw, ping

polarian: only into space at the end. If you want to resize a middle partition, it can't.

I tried to ask last night, but people were arguing the merits of the BSDs at the time

Anyone around using pkgBase?

what's your question?

anybody familiar with mtools and gpart/glabel etc?

jbo: hi

I am trying to set a partition label EFISYS such that Linux `blkid -L EFISYS` works

lw, hey

glabel & gpart don't do what I need, they only set `PARTLABEL=EFISYS` and not `LABEL`

so I'm wondering if its possible with base tools. FWIW emulators/mtools was able to change it

rafe: plenty of people by now

(>3? at least)

Anyone know why st_dev in struct stat has a full 64 bit value but kinfo_vmentry returned by KERN_PROC_VMMAP is a 64bit field but only seems to contain the lower 32bits of the value obtained by stat?

(for ZFS)

Hello, I cannot get FreeBSD 14.0 to boot on Banana Pi M1+ (ARM A20). The last messages I see before it hangs are: Using DTB provided by EFI at 0x78f0f000. Kernel entry at 0x72200200.Kernel args: (null). Does anyone have any suggestions on what I can try? Thanks!

Can I run a different version of FreeBSD that will give me more debug output during boot? Would UART and serial console output help with dianosing?

BananaPiHalp, do you see a menu similar to the one in docs.freebsd.org/en/books/handbook/bsdinstall/#bsdinstall-view-probe (perhaps with ASCII art instead)? From "Boot options" in that menu, you should be able to request verbose boot.

Title: Chapter 2. Installing FreeBSD | FreeBSD Documentation Portal

V_PauAmma_V: Unfortunately, no. I am using SD CARD so there is no installation and boot does not have a menu.

OK. Then perhaps someone familiar with that installation method can help. Or you could try the 14.1 beta release to see if it helps.

This is as far as I my boot gets. I don't see any debug statements in bi_load so it looks like I need to do a custom build to add them myself? github.com/freebsd/freebsd-src/blob…tand/efi/loader/arch/arm/exec.c#L77

Title: freebsd-src/stand/efi/loader/arch/arm/exec.c at 2ae0f5a4d0931067c672be9a791909f0e32d5a0e · freebsd/freebsd-src · GitHub

are you sure it's not actually booting but with a serial console? you might try mounting the sd card elsewhere and change boot_serial to NO in /boot/loader.conf

lw: Did not know about that! I will try that, thank you!

if you can't mount the sd card an alternative is to connect it to a network and see if it gets a dhcp lease, if it does you can log in as freebsd/freebsd via ssh

will try that too!

it might not be this, i haven't booted the sd card images for ages (and never on that hardware) but i know it does enable the serial console by default... i thought it was meant to be in dual console mode though

that would be my assumption as well, it would be strange to give up on hdmi output on these boards. nevertheless, it may offer some clues

there any real benefit to keeping logs 640 instead of 644? like obviously assuming there aren't any secrets in the log. is it actually less secure to run 644 for the convenience?

that's arguably something that can depend on your threat model, but principle of least privilege overall seems like an easy way be secure without having to overthink it. taking away perms from those not the user and outside the group seems to be something wise that you should have good reason to go against