<botifico>
[haikuports/haikuports] OscarL 8a3f3ee - python3.12: update to version 3.12.2. (#10111)
<Begasus>
there :)
<OscarL>
thanks :-D
* OscarL
waves at Anarchos
coolcoder613_ has quit [Ping timeout: 480 seconds]
<OscarL>
k... off to run some errands. See you later folks!
OscarL has quit [Quit: "I'll be back!"]
<Begasus>
later!
<Begasus>
tl :)
kescher has quit [Quit: Bye]
kescher has joined #haiku
pinaraf_ has joined #haiku
coolcoder613 has joined #haiku
pinaraf has quit [Read error: Connection reset by peer]
<Anarchos>
Begasus i made some progress on my math verifier : a lot of refactoring
<Anarchos>
Begasus ideas more clear :)
mmu_man has quit [Ping timeout: 480 seconds]
<Begasus>
Anarchos moving forward then :)
erysdren has joined #haiku
TheEricExperiment has joined #haiku
coolcoder613 has quit [Ping timeout: 480 seconds]
coolcoder613 has joined #haiku
mmu_man has joined #haiku
<Anarchos>
Begasus yes
jmairboeck has quit [Ping timeout: 480 seconds]
walkingdisaster has joined #haiku
arti has quit [Quit: No Ping reply in 180 seconds.]
<Begasus>
LabPlot picked up the post for it at mastodon client :)
Kokito has joined #haiku
arti has joined #haiku
<Kokito>
Howdy
<Kokito>
I am about to update my nightly build and was wondering if the Noto font issue has been fixed.
<Begasus>
Hi Kokito
<Begasus>
no :)
<Kokito>
Hi Begasus!
<Begasus>
so if you update best do it in Terminal
<Begasus>
"pkgman full-sync && shutdown -r"
<Kokito>
Thank you!
<Begasus>
np (someone mentioned that on the forum) :)
erysdren has quit [Quit: Konversation terminated!]
coolcoder613 has quit [Ping timeout: 480 seconds]
<Begasus>
k, back on track on the PR's
coolcoder613 has joined #haiku
<Begasus>
afk
coolcoder613_ has joined #haiku
coolcoder613 has quit [Ping timeout: 480 seconds]
<dovsienko>
waddlesplash: the other bug fixes to pcap-haiku.c have been committed, I am looking into loopback packets typing and framing now
Kokito has quit [Quit: Vision[]: i've been blurred!]
Kokito has joined #haiku
coolcoder613_ has quit [Ping timeout: 480 seconds]
Kokito has quit []
Kokito has joined #haiku
coolcoder613 has joined #haiku
<Begasus>
got around to it Kokito?
<Kokito>
Yeah, no problem. As it turns out, I was using a different font on my system. I did have to change the font settings in Vision though. All is good now!
<Kokito>
It is my favorite font with support for Japanese glyphs
<Begasus>
ah, good to go then :)
<Begasus>
do we have that in the depot?
<Kokito>
I don' know. Let me check.
HaikuUser has joined #haiku
<Kokito>
Yes, it is in the Haiku Depot. It's called M+
<Begasus>
ah :) thanks for checking
coolcoder613 has quit [Ping timeout: 480 seconds]
HaikuUser has quit [Quit: Vision[]: i've been blurred!]
coolcoder613 has joined #haiku
coolcoder613 has quit [Ping timeout: 480 seconds]
andreaa72 has joined #haiku
andreaa72 has left #haiku [#haiku]
<Kokito>
Begasus, M+ font on Depot is dated 10/8/17. I actually instaled the latest version manually.
<Kokito>
Just copied the font files to /boot/system/non-packaged/data/fonts
coolcoder613 has joined #haiku
<Begasus>
that works too Kokito, I use it all the time to check out things
<Begasus>
I'm having mplus-062.recipe in the tree?
coolcoder613 has quit [Ping timeout: 480 seconds]
HaikuUser has joined #haiku
coolcoder613 has joined #haiku
phcoder has quit [Quit: Connection closed for inactivity]
andreaa72 has joined #haiku
andreaa72 has left #haiku [#haiku]
kescher has quit [Quit: Bye]
coolcoder613 has quit [Ping timeout: 480 seconds]
HaikuUser has quit [Quit: Vision[]: i've been blurred!]
GregC has joined #haiku
phcoder has joined #haiku
<phcoder>
Hello. I'm unable to log into my account at review.haiku-os.org. I just see a single word "Forbidden" after entering my password. Who can help me?
coolcoder613 has joined #haiku
Kokito has quit [Quit: Vision[]: i've been blurred!]
kescher has joined #haiku
coolcoder613 has quit [Ping timeout: 480 seconds]
coolcoder613 has joined #haiku
eno_dk[m] has joined #haiku
coolcoder613 has quit [Ping timeout: 480 seconds]
v_harkonnen has quit [Ping timeout: 480 seconds]
v_harkonnen has joined #haiku
coolcoder613 has joined #haiku
<andreasdr[m]>
Hi there.
<Begasus>
Arrr andreasdr[m]
<andreasdr[m]>
Arrr Begasus!
coolcoder613 has quit [Ping timeout: 480 seconds]
<Begasus>
still fighting with kdevelop and kio :)
<Anarchos>
Begasus only the braves can succeed when the app name begins by 'k' :)
<phcoder>
I try to build PPC port but libgcc-kernel.a misses restgpr symbols. Easy to fix but it gets downloaded as part of gcc_syslibs_devel*-ppc.hpkg. Any pointers on rebuilding this package?
<waddlesplash>
phcoder: that'll be one for kallisti5
<waddlesplash>
the gerrit issue I mean
<Begasus>
heh Anarchos :)
<waddlesplash>
phschafft: those packages are/were built as part of the --bootstrap process
<waddlesplash>
err. sorry, phcoder not phschafft
<waddlesplash>
autocomplete behaving oddly...
<Begasus>
Anarchos, most of the things are working fine, only this one ...
pinaraf has joined #haiku
<phschafft>
waddlesplash: thanks for the correction. now I'm de-confused. ;)
pinaraf_ has quit [Ping timeout: 480 seconds]
<Begasus>
crash report (syslog) when launching kdevelop: https://bpa.st/OHCA
coolcoder613 has joined #haiku
<Anarchos>
Begasus i am unable to solve my highlight.js trouble on tiddlywiki :(
<Begasus>
:/
<Begasus>
haven't look at it anymore Anarchos
<waddlesplash>
dovsienko: sounds good. your changes in libpcap also look ok to me (at least from the Haiku side of things.) let me know if you run into any oddities or need anything else
zard has joined #haiku
<Anarchos>
Begasus look at what ?
<Begasus>
tiddlywiki
pinaraf has quit [Ping timeout: 480 seconds]
coolcoder613 has quit [Ping timeout: 480 seconds]
v_harkonnen has quit [Ping timeout: 480 seconds]
v_harkonnen has joined #haiku
coolcoder613 has joined #haiku
<Anarchos>
Begasus i can't remember how i installed it !
coolcoder613 has quit [Ping timeout: 480 seconds]
po||ux has quit [Quit: bye, bye!]
<Begasus>
Anarchos, would have to look again also on how I checked it
coolcoder613 has joined #haiku
<dovsienko>
waddlesplash: taking the packets as they come just works: "listening on loop, link-type RAW (Raw IP), snapshot length 262144 bytes"
<dovsienko>
the only remaining problem is to tell interfaces that do this from interfaces that don't
<waddlesplash>
check IFT_*?
<waddlesplash>
IFT_LOOP and IFT_TUNNEL will send only IP packets with no eth header, then obviously IFT_ETHER will have it
<waddlesplash>
you can check that on the interface, or you can just check the sockaddr received with the packet, I thinkl
<waddlesplash>
*think
<dovsienko>
I tried using SIOCGIFADDR as suggested in the bug report, also SIOCGIFTYPE, but both return an invalid interface type 0 instead of what I would expect (IFT_ETHER or IFT_LOOP)
<PulkoMandy>
Which one? There is a glitch when updating, but after a reboot it should be fine, I think?
<dovsienko>
i.e. I see both iifreq.fr_type == 0 and sockaddr_dl.sdl_type == 0
<dovsienko>
both for Ethernet and the loopback interface
<dovsienko>
the matter is, the framing needs to be known at the device activation time, at the packet reading time it is too late
<waddlesplash>
well, somehow ifconfig gets this information correctly
<waddlesplash>
and displays it
<waddlesplash>
so clearly it can be fetched
<dovsienko>
that's a good point, let me look into the source
<dovsienko>
BNetworkAddress::LinkLevelFrameType() looks the best fit, the only problem being that libpcap does not use C++ for portability reasons
<dovsienko>
(at least that's the method I would try first). alright, let me double-check the ioctl()s in my working copy...
xet7 has quit [Remote host closed the connection]
<waddlesplash>
dovsienko: those classes all just return stuff from the sockaddrs
<waddlesplash>
no need to invoke them in pcap
<waddlesplash>
that function just returns sdl_e_type. not sure if that's filled in or not here
<waddlesplash>
looks like ethernet sets it but maybe others don't
<waddlesplash>
dovsienko: yeah the kernel doesn't fill in sdl_e_type for SIOCGIFADDR
<waddlesplash>
just sdl_type
<dovsienko>
I see sockaddr_dl.sdl_e_type == 0
<waddlesplash>
yes
<waddlesplash>
and looking at the data structures the kernel fills SIOCGIFADDR from there's not really any good way to get sdl_e_type here
<waddlesplash>
we'd have to add new APIs to the deframe modules probably to have it export that info
<waddlesplash>
so just sdl_type is what you get
<waddlesplash>
hm. however the kernel does appear to know header_length here
<waddlesplash>
not sure if there's a good way to return that info, or what use it would be (in the absence of knowing what kind of header it is)
nosycat has joined #haiku
<dovsienko>
something is odd with my code, let me reduce the ioctl() to a small test program, and then if it still does not work on my system, I will send you the program so you can double-check on your system, fine?
coolcoder613 has joined #haiku
<waddlesplash>
sure
<waddlesplash>
but if ifconfig works for you then clearly there's some way to make this work on your system :)
<waddlesplash>
dovsienko: note that we do have an strace, though it looks like it's not especially useful for this case
<dovsienko>
that was the cause of the problem, the socket must be AF_LINK. on this note, I wonder of SIOCGIFTYPE was supposed to be the correct shortcut to sockaddr_dl.sdl_type
<dovsienko>
(trying to do a SIOCGIFTYPE on an AF_LINK socket returns EINVAL)
<dovsienko>
SIOCGIFTYPE "works" on an AF_INET socket, but ifr_type is always zero, so this possibly could be a failure to fail
<dovsienko>
my command of ioctl() is a bit rusty though
coolcoder613 has quit [Ping timeout: 480 seconds]
<dovsienko>
long story short, I can reproduce the happy path that results in sockaddr_dl.sdl_type being IFT_LOOP for the loopback and IFT_ETHER for the Ethernet interfac
bjorkintosh has joined #haiku
coolcoder613 has joined #haiku
phcoder has quit [Quit: Connection closed for inactivity]
<dovsienko>
also, SIOCGIFSTATS works on an AF_INET socket, but not on an AF_LINK one, which does not immediately make sense
coolcoder613 has quit [Ping timeout: 480 seconds]
mr_lou has quit [Read error: Connection reset by peer]
mr_lou has joined #haiku
coolcoder613 has joined #haiku
coolcoder613 has quit [Ping timeout: 480 seconds]
walkingdisaster has quit [Quit: Vision[]: i've been blurred!]
<waddlesplash>
indeed link doesn't implement that
<waddlesplash>
might make sense to add, not sure
coolcoder613 has joined #haiku
<dovsienko>
alright, my working version is not entirely dissimilar from the one korli suggested, but it has a comment and more error checks: pcap-haiku.c | 27 +++++++++++++++++++++++++--
Anarchos has quit [Quit: Vision[]: i've been blurred!]
<dovsienko>
this works on your development snapshot, and on beta4 loopback capture does not deliver any packets in the first place, so this should not require any backward compatibility workarounds, as far as I understand
<dovsienko>
waddlesplash: do you expect any of your recent changes to make it into beta4 or it will be beta5?
<phcoder>
kallisti5: can you help me with my account?
<phcoder>
waddlesplash: do you have a link to docs?
<waddlesplash>
dovsienko: beta5. however since beta4 doesn't support loopback capture it doesn't much matter does it?
x10z has joined #haiku
<waddlesplash>
phcoder: for bootstrap? uh, not sure we have much. they'll be in-tree in docs, maybe in docs/develop (which is rendered online here: https://www.haiku-os.org/docs/develop/)
<waddlesplash>
if we have any that is
coolcoder613 has joined #haiku
v_harkonnen has quit [Read error: Connection reset by peer]
tuaris has joined #haiku
v_harkonnen has joined #haiku
coolcoder613 has quit [Ping timeout: 480 seconds]
<dovsienko>
waddlesplash: I can certainly live without loopback capture in beta4, and if it stays this way, there will be less back-porting to do
<dovsienko>
DLT_RAW seems to be the best fit for this problem space, thank you for making these changes
<dovsienko>
since recvfrom() also fills the sockaddr, I am thinking of adding extra AF checks around there to make sure only IPv4 and IPv6 makes it through
coolcoder613 has joined #haiku
<dovsienko>
the new code will need a test run on beta4 to make sure non-loopback capture still works
<dovsienko>
also there are compiler warnings from a long time ago that require attention (and possibly another commit)
<waddlesplash>
dovsienko: it'll stay that way at this point. beta4 is now over a year old and we are working towards b5
<waddlesplash>
dovsienko: what wound up being wrong with korli's change, as tcpdump didn't work for him?
HaikuUser has joined #haiku
HaikuUser has quit []
<dovsienko>
once all is clear, the next logical step would be to back-port the entire lot into libpcap-1.10, so when the next stable release is out, it works on the development branch and continues to work on beta5 after it becomes available
<dovsienko>
so either the kernel or libpcap would need to prepend that
<dovsienko>
then tcpdump would find the 4-byte header and figure the rest out
coolcoder613_ has joined #haiku
<dovsienko>
however, both are superfluous to the problem space, especially DLT_NULL with its host-dependent byte order
<dovsienko>
also the ioctl() used a non-existent variable, and I started my changes using the AF_INET socket, which did not work
coolcoder613 has quit [Ping timeout: 480 seconds]
<dovsienko>
in the days before IPv6 it would be a perfect fit to use DLT_IPV4: "Raw IPv4; the packet begins with an IPv4 header."
<dovsienko>
in an IPv6-only world the perfect fit would be DLT_IPV6: "Raw IPv6; the packet begins with an IPv6 header."
mmu_man has quit [Ping timeout: 480 seconds]
<dovsienko>
however, DLT_RAW is a nice superset of the two without any additional headers: "Raw IP; the packet begins with an IPv4 or IPv6 header, with the version field of the header indicating whether it's an IPv4 or IPv6 header."
<dovsienko>
so you strip the Ethernet header(s) from an IPv4 or IPv6 packet and it is immediately consumable as DLT_RAW
<dovsienko>
(or you don't prepend Ethernet headers in the first place and avoid so much unnecessary complication)
<dovsienko>
as to why Linux loopback uses DLT_EN10MB, I do not know, but it would not be surprising if that was mostly for historic reasons (i.e. it was just easier to implement this way a long time ago)
<dovsienko>
because tunnels, which most certainly were implemented later than loopback, use other link types:
<dovsienko>
# tcpdump -i gre0 -L
<dovsienko>
Data link types for gre0 (use option -y to set):
<dovsienko>
LINUX_SLL (Linux cooked v1)
<dovsienko>
# tcpdump -i tunl0 -L
<dovsienko>
Data link types for tunl0 (use option -y to set):
<dovsienko>
RAW (Raw IP)
coolcoder613_ has quit [Ping timeout: 480 seconds]
coolcoder613 has joined #haiku
mmu_man has joined #haiku
nosycat has quit [Quit: Leaving]
coolcoder613 has quit [Ping timeout: 480 seconds]
coolcoder613 has joined #haiku
Al2O3 is now known as HappyValentines
HaikuUser has joined #haiku
HaikuUser has quit []
<dovsienko>
waddlesplash: I am adding some finish to my working copy, let me do that, test and commit it, then I am going to update the bug report.
coolcoder613 has quit [Ping timeout: 480 seconds]
coolcoder613_ has joined #haiku
gouchi has joined #haiku
gouchi has quit [Remote host closed the connection]
coolcoder613_ has quit [Ping timeout: 480 seconds]
x10z has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<waddlesplash>
dovsienko: ok
<waddlesplash>
dovsienko: oh yeah, I think DLT_RAW is what I used in the code I wrote for our tcp_shell (test harness to run the tcp stack in userland) when writing pcap files
v_harkonnen has quit [Read error: Connection reset by peer]
v_harkonnen has joined #haiku
<Begasus>
Some progress before I leave, still crashes when creating/opening a (new) project though https://0x0.st/HnMt.95_1.png (KDevelop with the 5.114.0 framework)
<Begasus>
cu peeps!
Begasus has quit [Quit: Vision[]: i've been blurred!]
zard has quit [Quit: leaving]
coolcoder613 has joined #haiku
nephele has joined #haiku
oldkenobi has quit [Quit: ...]
nephele has quit [Quit: Vision[]: i've been blurred!]
<waddlesplash>
Anarchos: I think your comment is correct
jmairboeck has quit [Quit: Konversation terminated!]
<Anarchos>
waddlesplash i would prefer them to accept it ... Otherwise i fear they will never merge it.
coolcoder613 has joined #haiku
<dovsienko>
packet capture is working on my OpenVPN tunnel using DLT_RAW, so "tcpdump -n -i tun/0 -c 10" shows the ICMP packets the remote end (Linux) is sending, but there is no response
<dovsienko>
which is consistent with the routing table at Haiku end not having a host route for the inner remote address
<dovsienko>
which may or may not be related to ifconfig displaying "inet addr: 10.x.x.25, Mask: 10.x.x.18" rather than a point-to-point style address
<dovsienko>
or could be a lack of Haiku support in OpenVPN
nephele has quit [Quit: Vision[]: i've been blurred!]
x10z has joined #haiku
coolcoder613 has quit [Ping timeout: 480 seconds]
v_harkonnen has joined #haiku
phcoder has quit [Quit: Connection closed for inactivity]
<dovsienko>
after I add a host route, there's an ICMP echo reply to every ICMP echo request *inside* tun/0, but on the Ethernet interface the Haiku host is still not sending UDP/1194 back
<dovsienko>
alright... I tried a few configuration tweaks, but it seems the packets Haiku sends out of tun/0 either do not reach the OpenVPN process, or it discards them
coolcoder613 has quit [Ping timeout: 480 seconds]
<dovsienko>
FWIW, my working copy allows tcpdump to see incoming IPv4 and IPv6, as well as outgoing IPv4 on the tunnel interface using DLT_RAW, so this will be a working part of the commit, and OpenVPN (or tun?) issues can be debugged as and when necessary
<dovsienko>
once in a long time Haiku OpenVPN seems to send an occasional UDP/1194 packet to the Linux host, which I presume is keep-alive, but no payload is making through from Haiku to Linux, so it seems the problem is somewhere between the tun interface and OpenVPN process on Haiku
HaikuUser has joined #haiku
HaikuUser has quit []
<waddlesplash>
dovsienko: ah if you got the older version of the patch set I did, it broke openvpn initially
<waddlesplash>
the second version fixed it
<dovsienko>
is there an easy way to get a newer version then?
<dovsienko>
waddlesplash: yep, on "hrev57582_7403_3 Feb 15 2024 19" traffic is making both ways using the same combination of configuration and manual host route
coolcoder613 has quit [Ping timeout: 480 seconds]
<dovsienko>
so OpenVPN in tun mode is fine with my working copy of libpcap, if you needed this information for the patch review. let's test OpenVPN tap mode for the sake of completeness...
DKnoto_W has joined #haiku
coolcoder613 has joined #haiku
DKnoto has quit [Ping timeout: 480 seconds]
coolcoder613 has quit [Ping timeout: 480 seconds]
<dovsienko>
waddlesplash: OpenVPN in tap mode works, as far as ping is concerned, by in pcap-haiku.c recvfrom() occasionally returns with bytesReceived > 0, but fromLength == 0
<dovsienko>
when this happens on a tunnel that has just been brought up, it tends to be that sockaddr.sa_len == 0 and sockaddr.sa_family == 0, so my filter discards these packets as invalid for DLT_EN10MB, even though the packet length is 100-200 bytes
<dovsienko>
let me see what these discarded packets looks like without the sockaddr data verification...
DKnoto_W has quit [Quit: Leaving]
DKnoto has joined #haiku
<dovsienko>
that's multicast. looks like a bug to me: when multicast arrives on a TAP interface, recvfrom() zeroes both src_addr and addrlen
<dovsienko>
in fact, this is not specific to multicast, but to any incoming packet
coolcoder613 has joined #haiku
<dovsienko>
the ARP request and the ICMP echo requests from the Linux end all come with: fromLength == 0
<dovsienko>
in addition to that, most of broadcast and multicast arrive with sockaddr.sa_len == 0 and sockaddr.sa_family == 0, then after a minute or so they are more likely to have properly filled sockaddr_dl
<dovsienko>
possibly there's an initialization missing somewhere
<dovsienko>
let's add this to the bug report for now
coolcoder613 has quit [Ping timeout: 480 seconds]
<PulkoMandy>
Yes, multimast support is not done yet
HaikuUser has joined #haiku
HaikuUser is now known as BabaZula
<dovsienko>
I mean, you get the packet data if you turn a blind eye on the sockaddr, but that's a sign of a bug somewhere. I am going to omit the sanity check for now, but it would be nice to have it eventually
coolcoder613 has joined #haiku
BabaZula has quit []
louis77 has left #haiku [Using Circe, the loveliest of all IRC clients]