<kallisti5[m]>
<Begasus> "don't let kallisti5 hear you :P" <- *powers up* Yup! ... *powers down*
<OscarL>
heh
* OscarL
runs (an even larger) "get -r" one las time. Just in case.
AlaskanEmily has joined #haiku
<OscarL>
copied 13856 files... no issues. FWIW, my VM is reading from a physical partition (VBox has it attached via an AHCI controller, without using host I/O cache).
<waddlesplash>
OK
<waddlesplash>
then it's probably virtio scsi indeed
<OscarL>
Devices preferences shows it under: Mass storage Controller-> SCSI Disk Drive (HARDDISK)
<TmTFx>
Is it just my machine or even yours haiku_nightly have broken "tar" command from commandline? In beta4 it works fine, in nightly it creates an output file of 299bytes @.@
TmTFx has quit [Quit: Vision[]: i've been blurred!]
TmTFx has joined #haiku
* OscarL
boots one last time today, just to check tar.
<OscarL>
jeez, 100% (or watever it is instead of minimum) brightness is killing me :-D
<TmTFx>
*_+
<TmTFx>
I looked at md5sum of both packages (in beta4 and nightly) and they look the same
<OscarL>
TmTFx: tar --version --> 1.35?
<TmTFx>
yes
<OscarL>
which parameters are you using?
<TmTFx>
tar cjf temp.tar.bz2 HQ\ Orchestral\ Soundfont\ Collection\ v2.1.2.sf2
<TmTFx>
with minus or without it won't work
walkingdisaster has joined #haiku
<OscarL>
"tar cjf temp.tar.bz2 Code" worked for me (Code being a dir with... you can guess).
<TmTFx>
let me try with dirs
<TmTFx>
not even with dirs here... ~> tar -cjf test.tar.bz2 Projects
<TmTFx>
created a file of -rw-r--r-- 1 user root 513 Lui 30 06:57 test.tar.bz2
<waddlesplash>
no errors?
<TmTFx>
513bytes
<TmTFx>
oh
<TmTFx>
no
<TmTFx>
sorry checked wrong dir
<TmTFx>
with dir works
<waddlesplash>
well, I tried with just one file and it worked ok
<OscarL>
yup, works for me with sigle files too.
<TmTFx>
T_T
<TmTFx>
why not for me
<OscarL>
pkgman full-sync? and see what it wants to change (if any) ?
<TmTFx>
it seems that the 400MB file became a file of 46bytes
<TmTFx>
it's not a tar problem...
<TmTFx>
not even cp
<TmTFx>
I'm gonna find those lost bytes
<TmTFx>
sorry for losing time ^^'
<waddlesplash>
your disk may be corrupted
<waddlesplash>
run a checkfs
<TmTFx>
tar is ok! It was just the file overwritten by a command, my fault. Thank you for your efforts
<OscarL>
mmm, not sure if placebo effect, or that last change related to "power-saving" on the scheduler is really noticeable.
<OscarL>
I have the impression that cpu freqs spike up less than before (watching via ActivityMonitor).
<OscarL>
won't help much with battery life, given the screen is brighter than a morning sun but... I'll take it :-D
<TmTFx>
same here
<OscarL>
the one time I get a CPU with working frequency scaling on Haiku... can't use touchpad, screen blinds me, and wifi works perfect 1 day every 3 months :-D
Begasus has joined #haiku
<OscarL>
(thus why I still use Haiku on VBox on that netbook)
<OscarL>
up already Begasus?
<Begasus>
g'morning peeps and OscarL :)
<Begasus>
almost up for 2 hours :)
rexbinary_ has joined #haiku
<Begasus>
you're still up, or had a nap in between?
<OscarL>
nah... still up. seems *you* had just a nap instead of a full nigth of sleep :-D
rexbinar- has joined #haiku
<OscarL>
(or I lost again the sense of time... happens :-D)
<Begasus>
shorter then usual, up as usual though :)
<OscarL>
re: boot issues related to serial output... I remember being unable to boot my VM if I left the serial port loop cable connected (Windows COM port, redirected to the VM).
rexbinary has quit [Ping timeout: 480 seconds]
<Begasus>
I hardly boot into Windows lately, and it's been even longer since I done that :)
<Begasus>
guess the python PR's can be merged, no comments there :)
rexbinary_ has quit [Ping timeout: 480 seconds]
<OscarL>
they "look not that ugly to me" at least :-P
<OscarL>
+1 on all of them from my side :-D
<Begasus>
yeah, I know they don't follow you recent changes, but not going for a full re-write on them, good enough for now :)
<Begasus>
hmm ... should update neochat to latest beta here :)
<Begasus>
Feature Add an option to block invites from users who you may not knowExternal link, to help users being spammed by bad actors on Matrix. This is could also be useful if you’re only using your Matrix account to communicate with friends & family. 24.08
<OscarL>
Begasus: no worries on not following my "style". I'm just learning as I go, so my "style" also changes with time. And if something bothers me enough... either I'll propose a change, or ask for help (still will be staying away from rebuilding qt stuff, for example :-D)
<botifico>
[haikuports/haikuports] Begasus 4e49ead - assimp, bump version (#10760)
<Begasus>
k, branches cleaned up localy :)
<OscarL>
woohoo!
<Begasus>
now cleaning up local recipes in my "working" fork
<OscarL>
re that "block invites" feature... reminds me of /me blocking Telegram "users" (bots mostly) from time to time. (because I have a public @OscarL handle there, despite only using Telegram with my close family/friends)
AlaskanEmily has quit [Remote host closed the connection]
<Begasus>
all on par :)
SLema has quit [Ping timeout: 480 seconds]
<Begasus>
got an account on Telegram, only to try it out, wonder if I can still log in :P
<OscarL>
Remeber that. We exchanged some "hello there" when you tried it :-D. Ny default, Telegram accounts expire after 6 months of inactivity, IIRC (unless you change settings, up to a 1 year).
<Begasus>
heh
<OscarL>
yeah, options are 1, 3, 6 months or a year.
<Begasus>
probably expired then :)
<Begasus>
IRC will do for here, matrix mainly to get in touch with KDE dev's
<Begasus>
should add a connection to libera.chat in Vision ...
<OscarL>
same. I just use IRC here, to bother (with my non-stop nonsense) all those [m] lurker users we have :-D
<Begasus>
heh
<Begasus>
someone has to entertain them :P
<OscarL>
can't find other reason for them to be here, really :-P
<Begasus>
141 users in the channel, atleast 2 that are awake :)
dby has quit [Quit: Vision[]: i've been blurred!]
* OscarL
passes the hat, just in case.
<Begasus>
heh
vdamewood has joined #haiku
<Begasus>
k, libera.chat login fixed :)
axeld has joined #haiku
<Begasus>
The following configuration files were considered but not accepted:
<Begasus>
bugger ... time to boost that one I guess :P
<Begasus>
moin axeld
<axeld>
Moin Begasus
<Begasus>
not seeing you pop in the channel that much :)
<Begasus>
k, libquotient_qt5 updated, now libquotient_qt6
<axeld>
Begasus: Indeed, I don´t find much time lately
<Begasus>
QCRITICAL: TestOlmAccount::claimMultipleKeys() Can't use password login at localhost:1234 - check that the homeserver is running
<Begasus>
I'm not running a homeserver, stupid test :P
<Begasus>
can relate axeld, pasted that stage here :)
<Begasus>
pasted/past* (?) :)
<Begasus>
biab :)
DKnoto has joined #haiku
<axeld>
I guess it´ll take some more time until I´m there...
<OscarL>
holy molly! AxelD in da house!
<OscarL>
not sure when I'll get another "live" chance to say this so... thanks A LOT AxelD for all your amazing BeOS/OpenBeOS/Haiku related work over the years! Much appreciated!
<axeld>
OscarL: Thanks, though I wish it could be more nowadays
<OscarL>
wish I could either send you some of me free time, or get some of your skills :-D. In any case... thanks a lot, man. I've been reading your commits logs for longer than I dare to remember.
DKnoto_W has quit [Ping timeout: 480 seconds]
<Begasus>
second that OscarL
<OscarL>
at least he is no longer locked in a basement :-D
<OscarL>
(mmm, or should we...)
<Begasus>
then this would be just another quiet channel :P
<OscarL>
I mean... should we lock him up in a basement again, as BGA and co used to do back then hoping for more AxelD awesomeness? :-D
<Begasus>
didn't work in the case of BGA, I guess he's still stuck there with the others :P
<Begasus>
k, doggies, biab
<OscarL>
enjoy!
freddietilley has joined #haiku
tuaris1 has quit [Read error: Connection reset by peer]
Begasus has quit [Quit: Vision[]: i've been blurred!]
dovsienko has joined #haiku
<dovsienko>
hello all
<dovsienko>
waddlesplash: the KDL prompt for the most recent occurrence of EndpointManager::Unbind() (#17058) is available in case you need it
scops has joined #haiku
Anarchos has quit [Ping timeout: 480 seconds]
<waddlesplash>
dovsienko: hmm
<waddlesplash>
dovsienko: can you try "dump_endpoints"?
<waddlesplash>
this won't list our socket though because it's not in the hash
<waddlesplash>
but we still might get something interesting
<dovsienko>
waddlesplash: Unknown command "dump_endpoints". Enter "help" to get a list of all supported commands.
<waddlesplash>
oh sorry, tcp_endpoints
<waddlesplash>
and then, tcp_endpoint <address from the KDL message>
<dovsienko>
this time the VM has the virtual serial port, so I can copy and paste from terminal as usual
<dovsienko>
waddlesplash: do you want it in a private message or in the ticket?
<waddlesplash>
paste in ticket
mmu_man has joined #haiku
<dovsienko>
for the second command which specific address from KDL does it need?
<waddlesplash>
"bound endpoint 0x.... not in hash"
juanjo has joined #haiku
<dovsienko>
done. is 206.189.242.242 the Haiku update server by any chance?
<waddlesplash>
yes
<waddlesplash>
dovsienko: no, the second command needed to be different
<dovsienko>
it would be a good practice to give it a correct PTR record
<waddlesplash>
note "tcp_endpointS" vs "tcp_endpoint" (no S)
<dovsienko>
pardon my ignorance, done
HaikuUser has joined #haiku
HaikuUser has quit []
<waddlesplash>
lol, that command doesn't print address :|
<waddlesplash>
well at least we can see it's in state syn-received not a closed state
<waddlesplash>
so that's interesting
<dovsienko>
arguably, the first command could complain about an unexpected argument. also, when typing "help" on the serial console, the pager "press any key" prompt is on the GUI screen only
<waddlesplash>
there is a way to disable the pager if you are actually using the serial console as default
<waddlesplash>
might just be a kernel settings file thing, not sure
<dovsienko>
let's focus on the roughest issue at hand
<waddlesplash>
yeah
<waddlesplash>
ok, so, let's see if we can dump the raw bytes at the address offset
<dovsienko>
I could TCP-forward the serial connection to you if that helps
<waddlesplash>
I think there's only one more bit of useful info
<waddlesplash>
looks like we need offsets 24 and 32 respectively
<waddlesplash>
so: dl 0x...+24
<waddlesplash>
(just put +24 inline, the KDL will compute the address for you)
<dovsienko>
I was just going to ask :-D
<waddlesplash>
then with the first address dumped there, do "dw <addr>+32"
<waddlesplash>
or maybe "db" (bytes)
<waddlesplash>
since this will be the sockaddr_in
zard has quit [Quit: leaving]
<dovsienko>
please check
<waddlesplash>
well ... that doesn't look like a sockaddr_in
<dovsienko>
could be a stale pointer somewhere
<waddlesplash>
shouldn't be
<waddlesplash>
this is a KDEBUG build, freed memory gets overwritten with 0xdeadbeef
_-Caleb-_ has left #haiku [#haiku]
<dovsienko>
it could have been reallocated to something else later
<waddlesplash>
it is possible I suppose
<waddlesplash>
it is also possible I just got the offsets wrong somehow
<dovsienko>
I can keep the VM in this state for a few more days if necessary
<waddlesplash>
hm
<dovsienko>
(in case you need time to think about it)
<waddlesplash>
well, I don't know
<waddlesplash>
if that is really the socket address then it could explain what's going on here: it's memory corruption
<waddlesplash>
so the socket really is in the endpoint hash, but the endpoint hash does not find it because the port has been overwritten
<waddlesplash>
the TCP endpoint should have a reference to the socket and does not release it until a time-wait state or something like it
<waddlesplash>
so that should not be the cause of the problem
Anarchos has joined #haiku
<waddlesplash>
well, I see another assertion we can add to the reference counter
<waddlesplash>
not sure if that will help
<waddlesplash>
dovsienko: you can also try "socket <second addr>" (i.e. the addr fetched from "dl")
<dovsienko>
do you mean 0xffffffff80c6cf18?
bjorkintosh has quit [Quit: "Every day, computers are making people easier to use." David Temkin]
<waddlesplash>
yes
<dovsienko>
done
<waddlesplash>
for some reason it looks like we do not have a KDL command to give info for a specific allocation
<waddlesplash>
the memory manager has a routine for it but it's not wired to a command
<waddlesplash>
well, admittedly there is no "debug" variant that ignores locking
<waddlesplash>
yeah, that doesn't look like a socket
<waddlesplash>
dovsienko: try dl 0x..+16 [first address, of the endpoint]
<waddlesplash>
see if the address there is also 0x...cf18
<dovsienko>
do you need another command fed into the prompt?
<waddlesplash>
yeah, try "dl" again, with the endpoint address, and +16
<waddlesplash>
if the address dl spits out in the data is 0x...cf18 that will be interesting
<waddlesplash>
if it's not that will be even more interesting
<waddlesplash>
other than that I don't have any more ideas, I will see about adding a few more asserts
<waddlesplash>
does this reproduce any more reliably?
<waddlesplash>
if so, I will make a test build
<dovsienko>
posted to the ticket. this is the first time it reproduced in a couple months, but I have not been using SSH connections much on this VM
<waddlesplash>
OK
<waddlesplash>
dovsienko: interesting, that's not the same address
<waddlesplash>
could you dump it as a socket?
<waddlesplash>
wait
<waddlesplash>
no, it is the same, I misread
<waddlesplash>
OK, well, at least it's consistent; it's the socket that got trashed somehow
<dovsienko>
are you sure you do not want the serial console?
<waddlesplash>
yeah, that's all I have to test at the moment really
<waddlesplash>
more interesting than serial console would be GDB but it's not helpful on a non-debug build like this
<waddlesplash>
looks like net_socket is 480 bytes, so it will be in the block allocator for size 512
SLema has joined #haiku
<waddlesplash>
dovsienko: hmm ... actually. maybe if I check a few things here serial console access could be interesting after all
<waddlesplash>
that READ FAULT is at 0xccccc
<waddlesplash>
that's what we fill memory with before it's returned from the allocator
<waddlesplash>
so, that does point to the socket getting freed and then reused
<waddlesplash>
as opposed to overwritten
<dovsienko>
let me rig this up...
<waddlesplash>
hmm
<waddlesplash>
dovsienko: eh. actually, never mind
<waddlesplash>
if it's uninitialized memory then the only thing we could determine from here is what specifically reusde it
<waddlesplash>
which is not helpful information
<waddlesplash>
something freed the socket too early, and that's all there is too it
<waddlesplash>
I'll add more asserts though
SLema has quit [Read error: No route to host]
HaikuUser has joined #haiku
nipos has joined #haiku
<nipos>
Hello there. Does Haiku have some limit how much memory it does support?
<HaikuUser>
I don't remember :-)
<nipos>
Boots and works fine here if I enable the limit to 4GB memory, otherwise it's stuck at the bootscreen with no icons lighting up
<waddlesplash>
are you on 32bit or 64bit?
<phschafft>
nipos: 32 or 64 bit?
<HaikuUser>
It certainly can handle more than that.. is this the 32 bit version?
<nipos>
64bit
<waddlesplash>
hmm, interesting
<nipos>
I know it can handle more than 4GB, but maybe not 132GB?
<waddlesplash>
it should be able to
<waddlesplash>
but we may have a bug. do you have any way to get serial out on this machine?
HaikuUser has quit [Quit: Vision[]: i've been blurred!]
<dovsienko>
nipos: how much RAM does the host have without the limit? is it a hardware box or a VM?
axeld has quit [Ping timeout: 480 seconds]
<dovsienko>
BTW, about 1 time out of 10 Haiku boot gets stuck in my VM, then after a reset it usually goes well
<nipos>
Serial output should be easy over the iRMC, but I never tried. Let's test this...
<waddlesplash>
dovsienko: what light does it get stuck on?
SLema has joined #haiku
<dovsienko>
it is difficult to remember, but if it manages to get through any, it is about 1/3 of them. at most
<nipos>
Anyone know how I access Serial over LAN?
<waddlesplash>
dovsienko: so, "disk" icon?
<dovsienko>
install ipmitools or some such, identify the main command, open the man page and search for "sol" (serial over LAN)
<waddlesplash>
or the "devices" icon
<waddlesplash>
serial out will give better info, but if you are using virtio-scsi or virtio-block it may be virtio stalls
<waddlesplash>
I reported a ticket about this recently, it reproduces much more easily under the guarded heap
<waddlesplash>
hopefully korli will take a look
<dovsienko>
waddlesplash: I do not remember, it has been like this for a long time so I stopped paying attention
<dovsienko>
nipos: alternatively, plug an actual serial cable to a Linux host, then you can make it remotely accessible using ser2net and SSH forwarding
<nipos>
I don't have any serial cable, neither do I have any Linux stuff here, but ipmitool looks right. Currently trying to find out how to use it
<dovsienko>
waddlesplash: I will try to remember next time when booting, but do not hold your breath. what is the ticket you reported?
<SLema>
Hm, I had this issue recently but intermittlently
<waddlesplash>
or rather, it's one of the things I expected
<SLema>
Like I reboot and it's ok. Connection fails in Vision too.
<waddlesplash>
dovsienko: looks like the user I am running as just has no permissions to create directories in tmp for whatever reason, that's the problem
<waddlesplash>
we should add a chmod to the tmpfs mount I guess
<waddlesplash>
SLema: that sounds unrelated
SLema has quit [Quit: Vision[]: i've been blurred!]
SLema has joined #haiku
<nipos>
I tried disabling Non uniform memory access (NUMA) hoping that it may change something, but it didn't. Any other ideas?
<waddlesplash>
ultimately this is just a kernel bug somehow
<waddlesplash>
nipos: do you not get traces from the bootloader?
<nipos>
What do you mean?
<waddlesplash>
the paste you sent begins with "Welcome to kernel debugger output"
<nipos>
Well, that's all I get from Haiku I think
<waddlesplash>
the bootloader should also send serial logs
erysdren has joined #haiku
<waddlesplash>
I see this is beta4, can you try a nightly build?
<nipos>
This is everything from power on until it hangs, nothing from the bootloader
<waddlesplash>
ok
<nipos>
Still had a Beta4 USB Stick laying around. Sure I can test a nightly, but need to download it first. It's probably better to continue tomorrow. Thanks anyway :)
diver1 has quit [Quit: Leaving.]
_-Caleb-_ has joined #haiku
DKnoto has quit [Ping timeout: 480 seconds]
talos has quit [Ping timeout: 480 seconds]
coolcoder613 has quit [Quit: No Ping reply in 180 seconds.]