Most of the commercial Unix systems such as HP-UX and AIX and such did not have alloca() in their libraries and at the time it was considered a BSD feature. I am surprised that it was in Unix v7. It was not in System V.

And since it was not in a major release such as System V it could not be in the standard. Caution: This is all from my vague memory.

IMNHO: The problem with Rust is the toolchain for it. It's a huge task to bootstrap it. I am not fond of the language. But it is the toolchain that makes it problematic.

i donno, i think you'd figure it out eventually, checkpoint

cyric: thanks, i’ll take a look at that.

hm. so i upgraded fbsd.. had to reboot.. and gmirrors still don't show on reboot

i really need to sort that out

i don't want to keep rebuilding the mirror every time i reboot

probably easier to just give up on the mirrored encrypted swap

do you load the module

Macer: or is the module presently loaded

yes. it loads on boot

but the mirror is not the boot drive?

i thought the metadata was just supposed to magic itself

no

the mirror is just a swap

or at least that's what it's supposed to be

i thought so too lol

so if it loads on boot then the gmirror should just 'show up'

*if the module loads on boot

i just gave up on the mirroring

good luck macer

checkpoint, so malloca() is pre '89, but not necessarily present in any C89-compliant compiler.

rwp, the alloca() man page says: LIBRARY: Standard C library , which confused me.

Hmm... I think it is a semantic thing. The Standard C Library is libc but that "standard" is the system standard library and not a standard like POSIX. It's system standard not official blessed standard. A different use of the word.

I look at various doc and see this in older linux documentation, "There is evidence that the alloca() function appeared in 32V, PWB, PWB.2, 3BSD, and 4BSD. There is a man page for it in 4.3BSD. Linux uses the GNU version."

I remember porting code from BSD which used alloca() onto HP-UX which was System V and did not have it available. I had to patch code to use malloc()/free() rather than alloca().

win 45

:(

is there a single command you can do to essentially fully reset the TCP stack? Like drop all connections, drop all states, and start over clean as if you had just rebooted . . . except without rebooting?

service netif restart && service routing restart

Apart from cgit.freebsd.org/src/tree/gnu is there any other GPL code within freebsd?

wiki.freebsd.org/GPLinBase I throught it was finished already, but seems like it wasn't

much to my confusion a manpage exists on my 15.0-RELEASE system for `zfs-rewrite` but the `zfs` command does not claim to support it. Any of you running a zfs based machine able to take a look to see if it shows up in `zfs --help` >_<

15.0-RELEASE-p3 and: there is a manpage, and it does show up in zfs --help

hm

kFreeBSD durian 15.0-RELEASE-p4 FreeBSD 15.0-RELEASE-p4 GENERIC amd64

how fishy

I think this might be on account of a goofed up freebsd-update on my part, I thought I did correct that by grabbing a new base tarball a while back

in a newer created jail I see the rewrite command off its zfs

confirmed, the `file` output looks like the zfs is from the wrong OS version, so another side effect of my halfway state of freebsd-update

Hi

I have some weird problem

for no reason the media directory from my root disappeared put it listed in 'ls' output

when I try accessing it, I get No such file or directory

*but listed

I can't remove it.. even rm -f failed

also it's same when i try create new on e

rmdir is for removing (empty) directories

rmdir: media: Not a directory

if ls shows it, then what does _not_ show it? have you ruled out permission problems?

my filler gui

ridcully_: yeah permissions is fine

I scared that can be input/output

what does `ls -lad /media` claim it to be?

I experienced this with ntfs back then, it was returning input/output error with then same behavior with my current media dir

ridcully_: I can't access it

No such file or directory

but it still listed also can't create new one

I am nearly 0 experience with ufs

if it's listed ls, then `ls -la | grep media`?

ls: media: No such file or directory

it listed in standard output

So is it hardware problem?

mosaid: what do you mean it listed in standard output?

$ ls

COPYRIGHT fsck_msdosfs.core rescue

Xorg.core home root

bin lib sbin

boot libexec sys

compat media tmp

dev mnt usr

entropy net var

etc proc xdm.core

OK no need to paste the whole thing here, you can use a pastebin service. but like ridcully_ mentioned earlier, what is the output of: ls -la | grep media

should be a single line you can just paste here.

zie: I mean plain output no options

sorry

ls: media: No such file or directory

OK run: ls -la

then just paste to us the line with media in it.

zie it tried and No such file or directory no need to paste it's only in the plain output

is ls -la checks for directory realtime or something and plain command not?

you can read the man page, ls -la just says give us a long listing of all files. -la are options to the ls command.

but it seems like your ls command isn't taking those options. ls is almost always a built-in, but I wonder if you somehow overwrote that.

maybe try from a fresh shell/terminal?

I don't think ls is ever built in

I also need to mention I run rm -rf media and it returns nothing..

mosaid: why were you running rm -rf media? I thought you wanted your media? rm -rf removes stuff.. forcefully.

zie: you thinking of something unreal.. the problem isn't from ls when I try to do any thing to this dir all commands returns errors

zie: media is an empty dir

mosaid: how do you know either of those statements are true?

it's now corrupted for no reason.. so I want to delete it and create new one

zie: mkdir: .: No such file or directory

again, how do you know that it's corrupted? how is it corrupted, etc. since your ls command seems to be misbehaving, I think you might have larger problems than a media directory.

so mkdir is also corrupted

mosaid: try running: /bin/ls -la

It sounds like hard disk problem.. same error

"/rescue/ls" is same also

probably because you deleted everything mistyping your rm

isley: very possible, especially if mosaid is the root user while doing it.

no my all system functioning well all files accessibly also I rebooted the system

it's one dir problem

accessible

mosaid: except you can't seem to run ls with any options, mkdir, rm, etc.

with this dir only

it tried other dirs

this dir seems to be /, so you sure about that?

no

it's not /

mkdir returns that bec dir is not exist for it.. it corrupting mkdir

not exist for mkdir

can you: mkdir media2

this bizarre behavior means as I know and as I experienced disk error.. but I don't know why this happened to my new ufs system

zie: yeah also rm it

and why it happens is this disk failure or fstab

was media a different disk/partition?

no

may fstab broke my partition?

at boot

is fstab run after I wake pc from sleep zzz command

you can look at your fstab: cat /etc/fstab

yeah it's fine /dev/ada0p2 / ufs rw 1 1

default options for /

I made my pc sleep many times recently

due to some bug and I don't know why it happens some linux apps like seamonkey never runs only after zzz and wake up

I tried to run it without my gpu drivers running and it fixed but i need drivers

amdgpu

as I see there is many bugs here in freebsd noone talk about

especially for old hp Compaq

sleep/wake shouldn't affect this at all, but you can try rebooting just for funsies. I don't see how GPU issues would affect a 'media' directory.

no I was just throwing it here.. maybe someone know a fix heh

also feeling this will brake my pc.. "shouldn't affect this at all" ah, good

zie: tried

mosaid: that no such file or directory sounds like you tried ls -la media instead of plain ls -la did you really try ls -la in /?

yeah

and it listed in top of output that

it existed but corrupted.. so it confuse the system

I would have tried stat media but ls -la should also show everything relevant (plus info about everything else in /)

existed and not existed in same time XD

everything else is fine

I need to deep search and learn more about ufs and fstab to fix this

but my fears is maybe it's start of fail

if ls in / shows media as suggested in the paste above, and ls -la does not -- what is 'not' shown? is the line missing? is ls segfaulting because it sees/reads an abomination?

if it's ufs i would reboot and fsck

there's also `echo *` if your system is hosed

does anyone use ipmitool with their freebsd

i'm tring to figure out why `ipmitool delloem powermonitor powerconsumption` works on one dell host and not another one

echo * what is this?

cyric: finally got a chance to look at that. It certainly looks useful, but doesn’t help with what I want. It has avahi as a hard dependency, which depends on dbus, which is what I’m trying to avoid.

mosaid: echo *: it's a way to do an ls without using /bin/ls

same

media is listed

right the point of the `ls -la` is to figure out what the system thinks 'media' is (directory, file, symlink,etc). echo * won't do that.

It sounds like something is mounted at /media and the mount point is broken because the source is no longer attached.

It would be useful to run "mount" or "mount | grep media" and see what it says for it.

Reading the scrollback I think rebooting might be easier to clear any possible mounts there.

but if ls is crashing i think yeah, you should address that first

I re-read the scrollback... What was said that indicates ls is crashing? I saw the report that ls output was corrupted on the screen. But that's not crashing. Probably the listing output.

If I suspect that files being listed will be confusing to the screen I pipe the output to "less" and let it render the output safely and display it in the terminal for me.

oh

oops

ant-x, you're welcome. I have nothing against any language per se, or any software, my problem is when it starts contaminating everything

checkpoint, so, if it's getting to BSD I'm screwed... The reason I'm working on migrating to freebsd was precisely that, escaping the "invasion" of linux distros by first systemd, now rust, and who knows what's next... Kernels written in JavaScript using npm modules? Who knows!!

i see the overhead, but some of the pluses are easily identifiable with rust

It seems like the world loves things that are both slow, unsafe, and with 8475918735918937189 dependencies hahahaha

i won't argue with the dependencies but would recon that a kernel would not have very many

but rust is way more safe than c

rtprio, when it comes to memory management, sure

which realm of safety are you refering to?

My issue is with this package management thing... I mean, if programs in rust were written like programs in C, for instance, I need a library for handling HTTP(S) requests, so I link my program against libcurl and use it... But instead it seems to me that the dominant mindset around rust is to use some cargo package that will be downloaded from who

knows where and it will depend on another package who's gonna mine bitcoins in the background on my system

If only rust was used keeping the unix ecossystem around it on unix systems, well, who cares the language a dev uses to write their code, right?

right, from crates

yeah, that

All languages that do not include a comprehensive standard library and instead rely upon community contributed libraries (crates in this case) all go the same way as Wordpress modules and npm modules and Ruby gems and all of the rest. It all turns into bad code and security issues.

the "who knows where" can be a private, configured location, crates.freebsd.org for the sake of this argument

I mean... I wouldn't oppose rust software being distributed on the operating system if I knew it would just use the traditional operating system shared libraries... Then okay, use rust, use go, use npm, be free to list anything like that

rwp, yeah, that's precisely my point

Not that i.e. libcurl isn't community contributed just like everything else in open source, but it's a library that whoever distributed the operating system just checked and choosen

rtprio, yeah, but then freebsd maintainers besides maintaining freebsd packages will have to maintain rust packages?

And then they would do the same for npm packages? for go packages?

Not that I disagree with that, that would be perfect, but it seems like that would require A LOT of work

no, because libfetch or whatever would be a create instead of a .so

er, "be a crate"

well, there are no freebsd kernel npm packages, i'm not sure what you're trying to say with that

nor any javascript for that matter

rtprio, I'm not talking about freebsd, the npm kernel was just an hyperbole to describe the trend going on in linux

comparing a jump from C to Rust, to a jump from any of this to Node.js, is pretty farfetched

nearly all code now has some means of download packages and install them

when i started engineering one would download the tarball for a library to a nfs share so the build system could make use of it

these days, every time you build an app it downloads hundreds of three line functions from some repo which

And that is a bad thing.

most recently has been a highlighted to having supply chain trust issues

rtprio, precisely the problem I'm concerned about

if there was a crates.freebsd.org like suggested and everything there were approved by the community after at least some look, and every rust-based package was built only using that, well, I wouldn't have any issue running anything in rust, on the contrary, I would say "hey, it's also memory safe" and be happy knowing my system is safe

Wasn't for the practical problems that would require, like, a small community having to keep 812391823481437 small packages it would be perfect

I mean, I still don't love the idea that now every programming language would now have it's own package for handling HTTP instead of a single lib for the whole system, but at least won't be a security problem

i don't know if this is a solvable problem and i think it's only going to get worse

812391823481437 what?

is this some random.. or real number of pkgs bsd have?

and yeah rtprio I will address it

crates.freebsd.org seems like a bad idea, why not put them in ports like other stuff for freebsd? you can add explicite dependencies to other components (like psl) there and instead of keeping everything, just keep the versions that are relevant for the current tree, hopefully just one, but realistically it will be a few

codegirl, re: contaminating everything: the contamination is the result of the aggressive attudies of the authors. What with the abandomnet of Xorg in favour of Wayland, or the self-fullfiling prophecies of systemd's conquest ofLinuxes.

I like the trade-offs you get w/ rust. But that's what it is. C has a set of values. Rust has a set of values. JS/npm has a set of values. Depending on the values of that ecosytem, some use cases will be a better fit. Different considerations/accommodations to account for security/convenience concerns will sometimes need to be made to

accomplish whatever the goal is. The goals of a stable operating system aren't identical to the goals of cli applications, or gui applications, or web applications, and on and on

This is a bit weird, but I wanted to upgrade to 14.4 today. I usually compile my own kernel, so I went to /usr/src/ as usual, switched branch to releng/14.4, and pulled. Then merged my config and compiled. But after installing, it's reporting uname -r as 14.4-BETA3. Why BETA? I thought 14.4 should be RELEASE? Did I make some mistake?

Hmm.. Nevermind. I guess the easiest explanation is that I just forgot to pull before compiling, after switching branch.