x10z has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<OscarL>
x512[m]: re: mailing list: that was the case when *I* was on those lists (before 2006 :-D). In any case... hope you're considering "applying for commit access (+vote rights)", if the "On the initiative of existing members of the team" part of the rules are flexible, as it seems they are :-D
x10z has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
x10z has joined #haiku
valerius_ has quit [Quit: ZNC 1.8.2+deb2+b1 - https://znc.in]
valerius_ has joined #haiku
tuaris has quit [Read error: Connection reset by peer]
<andreasdr[m]>
Hi there.
v_harkonnen has quit [Ping timeout: 480 seconds]
AlaskanEmily has quit [Remote host closed the connection]
x10z has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
x10z has joined #haiku
freddietilley has joined #haiku
Begasus has joined #haiku
<Begasus>
g'morning peeps
x10z has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<Molnija>
my soul feels compressed
<zdykstra>
Morning Begasus!
<Begasus>
Hi there zdykstra :)
v_harkonnen has joined #haiku
<andreasdr[m]>
Hi Begasus (@_oftc_Begasus:matrix.org), Molnija, zdykstra
<Molnija>
does anyone plan to grow maize this year in this version of this universe
x10z has joined #haiku
<Begasus>
Hi andreasdr[m] Molnija
flag_ has joined #haiku
flag has quit [Ping timeout: 480 seconds]
<x512[m]>
@_oftc_OscarL:matrix.org: I do not consider getting commit access to Haiku because it is not really needed for me and I was not welcomed by Pulkomandy.
<x512[m]>
I need to post patches to mailing lists for some Haiku-related patches upstreaming.
xet7 has quit [Ping timeout: 480 seconds]
<andreasdr[m]>
Hi x512
<andreasdr[m]>
How is it going?
<coolcoder613>
Hi Begasus, andreasdr[m]
<Begasus>
Hi coolcoder613 :)
Begasus_32 has joined #haiku
<andreasdr[m]>
Hi coolcoder613 (@_oftc_coolcoder613:matrix.org)
<OscarL>
yesm but has at least one non-test related .py file in it.
<Begasus>
that got me going on frozen bubble :)
* OscarL
will try with just a simple cmd:python -> cmd:python3
<OscarL>
(or just commenting it out :-P)
<Begasus>
maybe it can be updated, not sure how far it is from upstream :)
<OscarL>
only see 1.8.11
<Begasus>
Today we are releasing GCompris version 4.0.
<Begasus>
something to look into :)
<Begasus>
(node:94340) MaxListenersExceededWarning: Possible EventEmitter memory leak detected. 11 close listeners added to [TLSSocket]. Use emitter.setMaxListeners() to increase limit
<Begasus>
that's still there :)
<OscarL>
oof... 12 MiB for this alien thing :-/
<Begasus>
npm notice New minor version of npm available! 10.1.0 -> 10.4.0
<Begasus>
that too (once nodejs is merged) :)
<Begasus>
bias
coolcoder613 has joined #haiku
x10z has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<botifico>
[haikuports/haikuports] Begasus dd86ef1 - npm bump version (#10128)
mmu_man has quit [Ping timeout: 480 seconds]
<OscarL>
"npm: too cool for 32 bits"
<Begasus>
we never know as it needs nodejs :P
<OscarL>
yeah... I meant that one... /me doesn't pays much attention to the JS world :-P
<Begasus_32>
note: '#pragma message: The simdutf library is designed for 64-bit processors and it seems that you are not compiling for a known 64-bit platform. All fast kernels will be disabled and performance may be poor. Please use a 64-bit target such as x64, 64-bit ARM or 64-bit PPC.
<dovsienko>
if there is a quick solution to implement pcap_inject(), it would be nice to have it done, otherwise it would be useful to expand the comment in the existing stub function to say why it cannot be done
<Begasus>
fmemopen.c:367:16: error: 'off64_t' undeclared (first use in this function); did you mean 'off_t'? ... seen it around ...
<dovsienko>
after that the space for immediate improvements can be considered exhausted and the next move would be back-porting
<Begasus>
k, this one is out of my leage :) error: expected expression before 'fmemc_get_buf_addr_t'
mmu_man has joined #haiku
<OscarL>
sweet satan, these tests make watching paint dry seems like a rollercoaster in comparison.
zard has joined #haiku
<Begasus>
lol
<Begasus>
fwiw, when the take too long I tend to nuke them too OscarL :)
v_harkonnen has joined #haiku
<OscarL>
86/122... Too late to kill it now. Sunken cost fallacy be damned!
<OscarL>
indeed, and then wonder why I have no sound :-)
<Begasus>
heh
<Begasus>
ok, ktorrent not so good, taking too much system resources here
<Begasus>
can't win them all :P
HaikuUser has joined #haiku
HaikuUser has quit []
HaikuUser has joined #haiku
HaikuUser has quit []
mr_lou has quit [Quit: Leaving]
mmu_man has quit [Ping timeout: 480 seconds]
nephele has joined #haiku
<dovsienko>
waddlesplash: I have done some tests on sending raw packets, and it seems Haiku network stack does not support such a feature now. could you double-check my results before I cement this with a code in pcap_inject()?
<waddlesplash>
not even with sockaddr?
<waddlesplash>
I mean, sockaddr specified in sendto
<dovsienko>
you mean sendto()?
v_harkonnen has joined #haiku
<waddlesplash>
yes
<dovsienko>
that would expect the packet data not to have the protocol header, correct?
<waddlesplash>
hm, maybe so
<waddlesplash>
dovsienko: what error do you get on send() after bind()?
<waddlesplash>
from what I can see, that path should actually work
<dovsienko>
but it would not match the definition of pcap_inject at https://www.tcpdump.org/manpages/pcap_inject.3pcap.html and the implementation in all other modules that support it, so it would be a problem to use that
<waddlesplash>
no, I said what error do you get with send not sendto
<dovsienko>
Linux uses a separate AF for such sockets (AF_SOCKET), I do not understand if it would be semantically correct to reuse AF_LINK for the purpose
<waddlesplash>
ah, maybe so
<waddlesplash>
I am not really an expert about these things
tqh has joined #haiku
bbjimmy has quit [Quit: Vision[]: i've been blurred!]
<dovsienko>
I found socket_send() in src/add-ons/kernel/network/stack/net_socket.cpp, which seems to be the implementation because it returns B_BAD_VALUE if send() gets the size argument above SSIZE_MAX:
mmu_man has joined #haiku
<waddlesplash>
that's the overall implementation of send() yes
<waddlesplash>
all send() goes through that
<dovsienko>
so my code can trigger this by making the packet length invalid:
<dovsienko>
try_direct(): send() returned -1
<dovsienko>
send(): Bad address
<dovsienko>
(arguably, it should complain about the length, not the address)
<dovsienko>
then the B_BAD_VALUE in socket_send() turns to EFAULT in the errno after send() returns -1
<waddlesplash>
no
<dovsienko>
oh wait
<waddlesplash>
EFAULT will be from 1 of 2 places:
<waddlesplash>
1. syscall entry, if you passed a buffer+length that isn't inside userspace
<dovsienko>
I oversized the packet length, but the actual buffer is still a short sequence of bytes, so perhaps there was a page fault trying to take that from userspace to the kernel, correct?
<waddlesplash>
2. buffer copy, if the buffer isn't actually there and it faulted when trying to read
<waddlesplash>
yes
<dovsienko>
let me confirm the result with a bigger buffer
bbjimmy has joined #haiku
<dovsienko>
apparently, it is not so easy to oversize SSIZE_MAX on a 64-bit host, so let's not do it
<Begasus>
afk
<waddlesplash>
dovsienko: why would you want to??
<dovsienko>
I wanted to confirm that's the actual function
<dovsienko>
so, with the correct arguments to send() it returns -1 with "Destination address required" (EDESTADDRREQ)
<dovsienko>
this happens both when I use bind() on the socket, feeding it the address returned by ioctl(SIOXGIFADDR) directly from struct ifreq, and when I use setsockopt() with SO_BINDTODEVICE
<dovsienko>
(in the latter case feeding it ifr_index from struct ifreq filled by SIOCGIFINDEX)
<dovsienko>
both methods fail on all permutations of {AF_LINK, AF_INET, AF_INET6} × {SOCK_RAW, SOCK_DGRAM} unless socket() fails first with "Address family not supported by protocol family"
<nephele>
> I wanted to confirm that's the actual function
<dovsienko>
12 combinations in total, none work. sometimes socket() succeeds, but ioctl() fails before the next step
<nephele>
It's much easier to add some debug prints in the kit, rebuild it, and put it in lib/ next to your test programm
<dovsienko>
it is likely there's a proper debugger as well, but I am trying to keep it simple for now
<dovsienko>
but thank you for the suggestion
<Halamix2>
Begasus: I've tried to build nodejs20 on x86 VM under VirtualBox, but all I got was a critical VM error :P
<dovsienko>
waddlesplash: would you like to have the test program?
<waddlesplash>
yes please
<nephele>
dovsienko: yes, called "Debugger" :D
<nephele>
it's a GUI programm, always installed. and you can attach it to running programms
<waddlesplash>
nephele: we are talking about kernel issues at the moment
<dovsienko>
which in turn can use pcap_inject() when available, which also tries to abstract a number of differences away
<waddlesplash>
I guess sending to AF_LINK would make sense to not have ethernet header
<waddlesplash>
and then AF_LINK adds it
<waddlesplash>
and AF_PACKET or AF_RAW or something would be logical to get things with headers
<dovsienko>
the way it seems to me, socket_send() ought to take into account the fact bind() or setsockopt() has been called, and use the associated interface
<waddlesplash>
it does that
<waddlesplash>
socket_send is extremely generic, it doesn't know anything about what's going on at lower levels
<waddlesplash>
literally all invocations of send() go through socket_send, whether it's for AF_UNIX or TCP or whatever else
<waddlesplash>
the "destination address required" is probably coming from a lower level after socket_send has passed the buffer off
<waddlesplash>
or, sendto() is causing a behavioral difference here
<Begasus>
Hi Halamix2, on bare metal it went fine as far as I let the build go ahead (untill korli merged) :)
<dovsienko>
the if() at line 1323 requires the destination address data
<Begasus>
Halamix2, bumped npm after that so that should be on par again too
<dovsienko>
which sendto() supplies and send() does not
<waddlesplash>
dovsienko: you missed the block right above that.
<OscarL>
Begasus: Mmm, think I'll need to change that "subversion$secondaryArchSuffix" to just "subversion" after all :-D
<OscarL>
(on the git PR :-D)
<Begasus>
not going to fix the parse errors on libpcre2, too many for my brain :/
<Begasus>
OscarL, can't use "cmd:svn"?
<Begasus>
hp should complain on what is missing there
<OscarL>
it needs a bunch of stuff from the subversion package (perl modules, libs)
<OscarL>
using "cmd:svn" gives the wrong impression that it needs the command line program.
<Begasus>
that was an example :) could be that it "requires" another cmd from in there
<OscarL>
doesn't needs it for building, so `git_svn_x86` will just refuse to install until I fix it :-D
<Begasus>
ps, you need to clean some branches :P
<OscarL>
stop snooping around!
<Begasus>
I said requires, not requires_build
<Begasus>
I only did "git fetch oscarl" :P
<OscarL>
welp... there was your mistake :-P
<Halamix2>
Small question about bare metal: Haiku is not like Windows, it won't self-destruct if I install x64 and x86 on one PC (separate partitions of course)?
<Begasus>
if they can handle it there shouldn't be an issue Halamix2
<OscarL>
Halamix2: I have it setup like that.
<Begasus>
got one like that on an external disc :)
* Begasus
does a "git checkout --track oscarl" ...
<OscarL>
Halamix2: even using same partitions/installs in both bare-metal, and from VirtualBox. Shouldn't cause issues.
<Halamix2>
OscarL: I had enough critical VM errors trying to compile stuff I'll go with bare metal for now :P
<OscarL>
weird.
<OscarL>
except for the difference on speed (and availalbe RAM)... works for me most of the time in my old Phenom II X4.
<OscarL>
man... who would now! .recipe changes work better if you SAVE them to disk!
<Begasus>
rofl
<OscarL>
good thing I at least rememberd to use both "-G" and "-F" this time :-D
* Begasus
whistels ...
<Begasus>
or rather mumbles ... forgot -G again
<OscarL>
yes! success!
<Begasus>
sure?
<OscarL>
got git 2.43.2 working against subversion 1.14.3.
<OscarL>
`git svn -h` didn't explode at least.
<Begasus>
heh
<Begasus>
which should go first?
<OscarL>
let me check with strace, to see if it loads the proper libs (it should, because I've nuked alien_svn already :-D)
<Begasus>
not merging yet ... but still
<Begasus>
running git build on 32bit (not updated subversion)
<OscarL>
hang on. need to push small change in the git .recipe first.
<Begasus>
tl :P
<Begasus>
but I can do a -F also :)
<OscarL>
welp, you can dp hp -G -F later then too :-D
<Begasus>
need to do a -F anyway as I commented out subversion$secondaryArchsuffix atm :P
Maturi0n_ has joined #haiku
<OscarL>
`strace perl /boot/system/lib/x86/git-core/git-svn | grep 'system/lib'` <<< looks good to me.
<OscarL>
force pushed already.
mr_lou has joined #haiku
<Begasus>
k, checking here on 64bit, that's a "bit" faster :)
<OscarL>
just in case... PR order shouldn't matter much, except for the alien_svn removal (should be last).
<Begasus>
makes sense :)
Maturi0n has quit [Ping timeout: 480 seconds]
<Begasus>
bugger ...
<Begasus>
no complaints about missing subversion in the recipe :/
<OscarL>
it is a runtime dependency... from a perl module. no way hp will detect that.
<Begasus>
fix it? :P
<OscarL>
even if I knew how... I'm not touching perl :-D
<Begasus>
k, seems to be good here
<Begasus>
biab
<Halamix2>
Welp, desktop PC doesn't let me to boot x86 pendrive for whatever reason (although I have secure boot enabled cause win11)
<Halamix2>
Laptop it it then
<Begasus>
Halamix2, on this laptop I can't boot 32bit either
<Begasus>
OscarL, on 64bit all seems to be ok
x10z has joined #haiku
<Begasus>
gitk still crashing on resizing
<OscarL>
glad to hear Begasus (not about that crash, thou!)
<Begasus>
it's an issue for quite some time OscarL
<OscarL>
again that SDL/OpenGL related crash, no?
<Begasus>
git seems to be ok on 32bit too
<Begasus>
yeah OscarL, although I can't find anything related there in the recipe?
<OscarL>
gitk uses tcl/wish... which in Haiku uses some implementation of wish that relies on SDL2.
<nosycat>
undroidwish, yeah.
<OscarL>
(that's what I remember at least :-P)
<Begasus>
ah right "wish"
v_harkonnen has quit [Ping timeout: 480 seconds]
<OscarL>
nosycat: I never seem able to remember that name :-D
<nosycat>
Can't blame you.
<Begasus>
food ... bbl
<OscarL>
off here as well. Later folks!
OscarL has quit [Quit: Gone with the wind]
<Halamix2>
Is it possible to install Haiku 32-bit on UEFI system or am I just doing something wrong?
<nephele>
it is possible
<nephele>
but only if your efi is also 32bit firmware
<Halamix2>
ah, that's the issue then
<nephele>
installing the 32bit version on a 64bit firmware currently iirc does not work
<nephele>
you may try to install it with CSM if your board supports that
<Halamix2>
I have some PCs that should be compatible; nto as fast as my main PC though. I wondered if installing it on a compatible PC and moving the drive, then setting up GRUB would work
x10z has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
xet7 has joined #haiku
<Halamix2>
Wait, if that would ever work then I could be able to boot the installer from GRUB already
<nephele>
No, it would not
<nephele>
Yes indeed :D
<nephele>
there is a forum topic about this problem, you could drop in there and check if there was any progress
<nephele>
booting the 32bit version from a 64bit efi loader that is
<nephele>
the opposite problem iirc has already been adressed. For some older macbooks with a 64bit processor and a 32bit efi
<nephele>
(I'm assuming you want the 32bit version for BeOS compat or something)
<Halamix2>
Something like that, I want 32bit version to test if package willb uild
<nephele>
in that case you could also spin it up in qemu under linux if you want
<Halamix2>
I just like breaking things and doing it the overly complicated way, I guess :P
<nephele>
That's fine, kindred spirit ;)
<nephele>
If you want you can tackle that efi loader problem though
x10z has joined #haiku
tombhadAC has joined #haiku
itaipu has joined #haiku
Begasus_32 has quit [Quit: Vision[]: Gone to the dogs!]
nosycat has quit [Quit: Leaving]
gouchi has joined #haiku
x10z has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<Begasus>
closing down, cu peeps!
Begasus has quit [Quit: Vision[]: i've been blurred!]
diver has quit [Quit: Leaving.]
nephele has quit [Quit: Vision[]: i've been blurred!]
coolcoder613_32 has quit [Quit: Vision[]: i've been blurred!]
illwieckz has quit [Quit: I'll be back!]
illwieckz has joined #haiku
diver has joined #haiku
xerexexe has joined #haiku
HaikuUser has joined #haiku
<HaikuUser>
Hallo
<zeldakatze[m]>
I've just installed Haiku on an old laptop (a TravelMate 290 built in 2002 which I had lying around) and I'm impressed at how well it runs with Haiku, given how weak the system is, feels like the laptop was built in 2012 instead of 2002.