<pairisto[m]>
how do you get the module name for a driver? for instance, tun.cpp's module name?
illwieckz has quit [Ping timeout: 480 seconds]
Maturi0n_ has joined #haiku
Maturi0n has quit [Ping timeout: 480 seconds]
illwieckz has joined #haiku
<phcoder>
@waddlesplash, @augiedoggie : fixed PS/2 on Chromebook. Turns out that draining isn't the problem. When I compared closely with other implementations it turns out that none checks return value of reset. I followed Linux and checked that a) reset command sends at all b) get_device_id is sane or c) set_leds is sane
<phcoder>
It's too late now, going to sleep. Will send code later
<waddlesplash>
hopefully this will still crash and give you something intelligible
<HaikuUser>
do i run that in my terminal?
<waddlesplash>
yes
HaikuUser2 has joined #haiku
<HaikuUser>
this time it when i run that command it crashes at an earlier point
HaikuUser2 has quit []
<HaikuUser>
but it seem to be at the same instruction, 0f05 syscall
<waddlesplash>
got a report?
<waddlesplash>
apologies, I'm in and out here. if you use my handle I will probably notice your messages more reliably
<HaikuUser>
yea, i got a report
frkzoid has joined #haiku
<waddlesplash>
the fault instruction will still be the syscall, but the message should be different
<waddlesplash>
because, hopefully the debug allocator caught the problem at its source
<HaikuUser>
is the message at the thread state?
<waddlesplash>
it should be at the line saying Call (...)
<HaikuUser>
Call (thread 3392 tried accessing address 0xea63939000 which is a guard page (base: 0xea63938ff0, size: 8, alignment: 16, allocated by thread: 3392, freed by thread: -1))
<waddlesplash>
yes, so this is an out of bounds read or write
<waddlesplash>
the debug allocator has done its job, this is the real problem :)
<waddlesplash>
so, where was the program executing before it hit this problem?
<waddlesplash>
is it in your code, or in a HaikuPorts library
<HaikuUser>
in my code
<waddlesplash>
well, then you have a bug
<waddlesplash>
hopefully, you can figure out what the problem is from here (without seeing any code I can't really say much more)
<HaikuUser>
okay, thank you
<waddlesplash>
the "base" and "size" fields may be helpful
<waddlesplash>
likely you either have an allocation that's too small, or you are using it incorrectly somehow
<HaikuUser>
i think i might know the problem, give me a second
<HaikuUser>
yea, it's fixed, thank you
<B2IA>
(AGMS) Butler message for ysei_-_ppc Netscape for BeOS PPC? All I could find were skins for Mozilla and CL-AMP.
HaikuUser has quit [Quit: Vision[]: i've been blurred!]
_0xdd has quit [Quit: WeeChat 3.5]
mmu_man has quit [Ping timeout: 480 seconds]
kinkinkijkin has quit [Quit: Leaving]
_0xdd has joined #haiku
phcoder has quit [Quit: Connection closed for inactivity]
linuxmaster has joined #haiku
<pairisto[m]>
I am a little confused about how drivers are published to /dev. Is it when you call open() on the driver itself, like if you were to call open on a driver in /boot/system/add-ons/kernel/drivers/dev/net?
<pairisto[m]>
So I am using the legacy API and I thought I was giving it the right installed devices? though since its a virtual device driver is giving it itself not good? like the way I have it should make the pseudo-file in /dev/net (as of current).
<OscarL>
"/boot/system/non-packaged/add-ons/kernel/drivers/dev/net/tun" << is that the binary you get when to do "jam tun_driver" ?
<OscarL>
(btw, AFAIK, for legacy drivers, the binary goes in "add-ons/kernel/drivers/bin", and then a symlink pointing to the binary is placed under /boot/system/non-packaged/add-ons/kernel/drivers/dev/[whatever]/[you_need]")
Begasus has joined #haiku
<Begasus>
g'morning peeps
<OscarL>
hi Begasus.
<Begasus>
Hola OscarL !
<Begasus>
checked out cudatext now? ;)
<OscarL>
haven't update yet :-D
* OscarL
will do so now.
<OscarL>
I have to remove all traces of my hackedup package (the one that used the SYS:ENV + LD_PRELOAD trick, and also removed lib:libpython from REQUIES) :D
<OscarL>
found it: "cudatext-1.195.0.6-666-x86_64.hpkg" revision a bit cursed :P
<Begasus>
:P
<OscarL>
"cudatext_source" maybe that should be disabled?
<OscarL>
Seems to work!
<Begasus>
nice! :)
<OscarL>
Noticed that at exit... it outputs: "Exception at 00000181DF99CA30: EAccessViolation:"
<OscarL>
(Only on 64 bits... was the same with the previous 3.7 version, BTW)
<OscarL>
but doesn't seems to affect normal operation.
<Begasus>
Yeah, seems to work as expected :D
<Begasus>
playing around with iaito now, gonna check a build on 64bit to see if it's more stable there (less crashes)
<OscarL>
good luck with that :-)
<Begasus>
looks promising so far, most things seem to work
<Begasus>
pancake (upstream) is debugging a package I linked him to see if he can find the cause
Begasus_64 has joined #haiku
<OscarL>
good to have some help from upstream.
<Begasus_64>
sure, seeing my coding skills are less then zero :)
<OscarL>
remember: debugging is harder than coding, so don't beat yourself over bugs you didn't even created :-D
<OscarL>
pairisto[m]: does "api_version" appears if you run "readelf -s /boot/system/non-packaged/add-ons/kernel/drivers/dev/net/tun | grep api_version" ? (if not, then either you have the wrong binary there, or something's is really off with it)
dryphb has quit [Ping timeout: 480 seconds]
<OscarL>
for reference, you can compare with "readelf -s /system/add-ons/kernel/drivers/bin/tty | grep api_version", for example.
freddietilley has joined #haiku
HaikuUser has joined #haiku
CPYou has joined #haiku
HaikuUser is now known as mrjones
CPYou has quit []
<OscarL>
Buh! Begasus's .recipe lexer for CudaText is not included! I want my money back! (Oh... wait...).
<OscarL>
Begasus_64: here you have an easy one to pump your commit number! "app-accessibility/brltty/patches/brltty-git.patchset" << patchset with no .recipe
<Begasus_64>
OscarL, check the menu option Lexers :P
<Begasus>
it isn't the first one there without a recipe OscarL
<Begasus>
mostly kept for historical reasons (not to loose the work)
<OscarL>
ok. /me puts down the broom.
<Begasus_64>
tried if you could get a recpe for it? :)
<Begasus_64>
get/create*
<OscarL>
nah... scipy left me wishing for a time out :-D
<Begasus_64>
heh, same here :P
<pairisto[m]>
OscarL: I am doing it on linux so I am not quite sure about readelf since its not on pkgman. While I can't compile it using `jam tun_driver`, looking through all the tun files in the built version of the system they are all the same using `comm -3 [tun in /system] [tun in /boot]`?
<Begasus_64>
pairisto[m], readelf is included in default install afaik
<OscarL>
"nm" can also show symbols, not sure what's available on Linux.
<OscarL>
readelf seems to be part of binutils on Linux (and also on Haiku :-D)
<pairisto[m]>
makes sense as I am using QEMU
<pairisto[m]>
to spin something up and test
<Begasus_64>
right
<OscarL>
you said "While I can't compile it using `jam tun_driver`" then how are you generating the binary you get under " /boot/system/non-packaged/add-ons/kernel/drivers/dev/net/tun" (again, seems a weird place to me, that should be a symlink to /boot/system/non-packaged/add-ons/kernel/drivers/bin/tun_driver, if I'm reading your jamfile correctly.
<pairisto[m]>
sorry, I thought you meant using jam tun_driver with the kernel addon portion on linux itself
<OscarL>
Might be a tad easier to just do things inside the VM? At least for legacy API drivers, that you can just replace the binary without the need for reboots... worked for me when doing my silly drivers (it87, amd_temp, poke).
<OscarL>
Heck, I use just the makefile engine to build them out of tree :-D
<pairisto[m]>
I have to cahnge the SubDir part for that right?
<OscarL>
for the "poke" driver (that one is included in Haiku's sources... I just do "jam <driver>poke", and it ends up inside the "generated/*" folder. I don't change anything. (but I install it manually. because I know nothing better :-D)
<OscarL>
(not that I've touch that one regularly... wrote it in 2005, and fixed a bug on it last year, I think :-D)
<Begasus_64>
fresh coffee :D
<pairisto[m]>
fair enough
korli has joined #haiku
<korli>
"jam <driver>poke" won't work :) needs to be: jam "<driver>poke"
<pairisto[m]>
OscarL: alright confirmed that api_version is not there
<Begasus_64>
hi korli! :)
<korli>
hi Begasus_64
<Begasus_64>
long time no see on IRC :)
<OscarL>
pairisto[m]: see? I even got that one wrong. Do as korli says :D
<OscarL>
Hi korli! nice to "meet" you :-)
<korli>
hi OscarL, we met already :)
<Begasus_64>
k, same behaviour on 64bit for Iaito
<Begasus_64>
hope upstream can debug that :)
<OscarL>
On IRC as well? (Think I lost my memory pills again :-D). Anyway... thanks a lot for all your work korli. Much appreciated!
<korli>
hi pairisto[m], getting along with building and testing a driver?
<Begasus_64>
he's haikuports master OscarL :)
<korli>
OscarL: long ago maybe on IRC
<pairisto[m]>
sort of not really? it builds but it doesn't run
<OscarL>
I see. Thanks for your patience too. Sorry for being too slow most of the time. :-D
<pairisto[m]>
as in it doesn't make the pseudo-file when open is used
<korli>
pairisto[m]: do you have a dev entry under /dev/ when using find /dev -name tun
<korli>
or whatever name you used
<pairisto[m]>
nope :/
<korli>
it can't be opened when it doesn't exist
<pairisto[m]>
in /dev yes but the bin is there /boot/system/add-ons/kernel/drivers/bin
<korli>
ok the net_server probably tries everything under /dev/net
<Habbie>
dryphb, i installed it on an SSD that already had two other OSes
<Habbie>
dryphb, just read the installation guide carefully
<Begasus_64>
and ... same thing with Qt5 on Iaito
<mrjones>
dryphb, You need some free space for a partition to Haiku.
<korli>
pairisto[m]: you can use "ifconfig delete /dev/net/tun"
Begasus_64 has quit [Quit: Vision[]: i've been blurred!]
<pairisto[m]>
yes
<pairisto[m]>
the interface is gone
<korli>
then check that the driver is working a expected with "cat /dev/net/tun"
nephele has joined #haiku
Begas_VM has joined #haiku
dryphb[m] has joined #haiku
<dryphb[m]>
Hello!
<pairisto[m]>
kinda, since I currently don't have anything in the read and it just returns B_OK I think its stuck infinitely but it is reading I think? at least when cat wasn't giving me anything and I tried tail it gave tail: memory exhausted
<dryphb>
[09:27] mrjones: dryphb, You need some free space for a partition to Haiku.
<dryphb>
I have it, but will a clean SSD help?
<korli>
pairisto[m]: yes
<pairisto[m]>
so reading is fine
<pairisto[m]>
* so reading is fine?
<dryphb>
Sweet, time to boot into my Debian live media & install gparted to nuke my SSD
<dryphb>
And yes, it's a known good 1tB drive
<mrjones>
dryphb, it will be easier with a whole disk for Haiku. But it is not needed. You can have multiple OS'es installed.
<korli>
pairisto[m]: you need some two buffer queues, one for the tun app side, one for the tun stack side, each read call wil take some buffer off each queue, each write call will put a buffer on the another queue
<mrjones>
dryphb, adjust the size of existing partitions if there are any to create empty space. Then create the partition during installation of Haiku.
<pairisto[m]>
korli: was thinking of something like that for reading and writing already through possibly a DoublyLinkedList to act a s a queue for the app side and the user-space side unless haiku already has a queue data structure?
<pairisto[m]>
* korli: was thinking of something like that for reading and writing already through possibly a DoublyLinkedList to act as a queue for the app side and the user-space side respectively unless haiku already has a queue data structure?
<korli>
util/queue.h has a queue
<korli>
but you need to know the buffer size of each buffer
* mrjones
is idle: doing non-computer stuff
<pairisto[m]>
are you talking about for a fixed_queue?
<korli>
no a simple queue
<pairisto[m]>
* talking about needing to know the buffer size for a
<korli>
possible net_buffer_module_info *gBufferModule; can be used to allocate and free buffers
<pairisto[m]>
probably will need that but are you referring to int count or are you referring to the buffer size to know where the next element in the queue is?
<korli>
I mean the buffer size for the read calls
<Begas_VM>
OscarL got me sidetracked again :P
<korli>
packets can be of different sizes
<OscarL>
"WARNING: The script pylsp is installed in '/packages/python3.9-3.9.17-1/.self/non-packaged/bin' which is not on PATH." <<< I think we might need to patch our "pip" packages to fix that spurious warning. (and maybe the same with the annoying "WARNING: Running pip as the 'root' " one :-D )
<OscarL>
Begas_VM: what did I do now? :-D
<pairisto[m]>
got it, forgot that C/C++ isn't python when it comes to reads lol
<OscarL>
Begas_VM: exactly, I guess it is trying to literally search for "/packages/python3.9-3.9.17-1/..." in PATH and it is not finding it :-D
<Begas_VM>
iaito recipe is as good as it gets for now
<pairisto[m]>
alright well I gotta get to bed for now since I've been up for about 20 hours so I'll catch you guys later. thx again korli and OscarL for all the help!
<Begas_VM>
ah
<OscarL>
re brltty: good, good. :-P
<OscarL>
later pairisto[m]. Sleep well!
<korli>
pairisto[m]: fine! please try to switch the driver to C++, it may help then to reuse some class like .src/add-ons/kernel/network/protocols/tcp/BufferQueue.h
<pairisto[m]>
got it
<Begas_VM>
g'night pairisto[m]
<nephele>
OscarL: you can just ignore the warning
<OscarL>
I know.
<nephele>
it's basically unfixable, it's a design flaw because of the existance of symlinks ;)
<OscarL>
but casual users might get confused.
<OscarL>
if its unfixable... then we can just patch it away :-P
<nephele>
I'm not sure how it determines what dir was used. but that would be /system/non-packaged anyhow
<nephele>
it should really use ~/config/non-packaged instead
nephele has quit [Quit: Vision[]: i've been blurred!]
<OscarL>
no. ~/config/non-packaged is for "pip install --user"
nephele has joined #haiku
freddietilley has quit [Quit: WeeChat 3.8]
<OscarL>
cool, got python-lsp-server working with CudaText. Too bad that I keep getting some Access Violations (on 64 bits at least).
* mrjones
is idle: BRB
* mrjones
has returned
freddietilley has joined #haiku
<OscarL>
(something's is not entirely right with Lazarus/Qt on 64 bits Haiku it seems)
<Begas_VM>
issues OscarL?
<OscarL>
some, yes. For example "ERROR: Exception in CudaText for cudatext._timer_proc_callback_proxy: Access violation"
mrjones has quit [Quit: Vision[]: i've been blurred!]
<Begas_VM>
when building cudatext with lazarus?
<OscarL>
(while using some of the features related to the python LSP client)
<OscarL>
I'm just using CudaText, not building it.
<Begas_VM>
ah :)
<Begas_VM>
lazarus is only needed to build, shouldn't interfere between qt and cudatext?
<Begas_VM>
other then that, the crashes on iaito might be related? :)
<OscarL>
by "lazarus" I mean.... the GUI components that are written in ObjectPascal... and those use Qt (on Haiku at least) as a backend.
<OscarL>
so we have... FCL->Qt->Haiku API.
<OscarL>
So... CudaText uses FCL code (that comes from Lazarus), and that calls Qt code, that then calls our API. And there's seems to be something not entirely right regarding, judging from the Access violations and "Object reference is Nil." I get on 64 bits.
<Begas_VM>
how do you trigger it?
<OscarL>
Open CudaText from terminal. just close it, and there you have at least one.
<OscarL>
the more you use menues, plugings, etc, the more they appear on stdout.
<Begas_VM>
nothing shown here
<Begas_VM>
KERN: runtime_loader: Cannot open file libiconv.so (needed by /boot/system/apps/CudaText): No such file or directory
<Begas_VM>
from syslog
<OscarL>
some also are reported inside CudaText (as ""ERROR: Exception in CudaText for cudatext._timer_proc_callback_proxy: Access violation", after I installed the LSP client plugin, and configured it to use the one for Python)
<Begas_VM>
OTE: LSP: no server configs loaded from "~/config/settings/cudatext/settings"
<Begas_VM>
NOTE* :)
<OscarL>
https://bpa.st/4UGYY << this one I get everytime I open and close CudaText from Terminal (with only an "untitled" doc)
<Begas_VM>
in Terminal: Data set on unsupported clipboard mode. QMimeData object will be deleted.
<Begas_VM>
weird, not getting any of those on 32bit
<OscarL>
are you trying on 32 bits? because I havee seen those exceptions only on 64 bits. /me reboots to 32 just to check again.
<Begas_VM>
LSP client installed
<OscarL>
not that weird, I think. might be just some "int vs pointer" issue.
<Begas_VM>
you're the expert there :)
<OscarL>
"expert"
* OscarL
dies laughing
<Begas_VM>
not yet! :P
AlaskanEmily has quit [Remote host closed the connection]
<OscarL>
(while on 32 bits) pkgman could suggest to install "package_x86" if you ask for "package" and the later doesn't existes, but the former do.
<OscarL>
*should
HaikuUser has joined #haiku
<Begasus>
nitpicking :)
HaikuUser2 has joined #haiku
HaikuUser2 has quit []
<OscarL>
"notpicking" the correct _x86 :-P
<Begasus>
k, nitpicking_x86 then :P
<Begasus>
what about when both are available?
<OscarL>
install the one the user requested.
<Begasus>
well, you requested package, not package_x86
<OscarL>
more and more packages are x86 only, so...
<Begasus>
right, but not all :)
<OscarL>
doesn't changes the fact that if you ask for "package", it is not available, but "package_x86" is... it would make sense to just install that.
<OscarL>
and yes... I know... "patches welcomed" :-P
freddietilley has quit [Quit: WeeChat 3.8]
<OscarL>
"This plugin does not support setting windows masks". You get that one too Begasus ? I *think* what one comes from Qt (and not from something in CudaText)
<Begas_VM>
You could always file an issue at haikuports :) (I'm not touching qt) :P
freddietilley has joined #haiku
<OscarL>
yeah... neither do I.
<OscarL>
damn, just seeing that I'm on "scipy-1.10.1" brach made me wanna turn off the PC :-D
<OscarL>
*branch
<Begas_VM>
lol
HaikuUser has quit [Quit: Vision[]: i've been blurred!]
<OscarL>
gah! "pip3.9 install python-lsp-server" failed while tying to compile some code for a .whl file. not smart enough to "setarch x86" (neither I nor pip, that is :-P)
<Begas_VM>
going strong today :)
nephele has quit [Quit: Vision[]: i've been blurred!]
<OscarL>
some of us need to be on the wrong side of the Bell curve, I guess :-D
mmu_man has joined #haiku
<OscarL>
Cool. LSP for python works OK on CudaText 32 bits. (no "Access violation" or other issues so far).
<nekobot>
[haiku/haiku] ddc88ea85a63 - docs: remove TODO item for ARM Accessed and Modifed page flags
HaikuUser has joined #haiku
HaikuUser has quit []
The_Ringmaster has joined #haiku
tqh has quit [Quit: Leaving]
<pairisto[m]>
quick question, I want to add BufferQueue.h from src/add-ons/kernel/network/protocols/tcp but I am not currently sure what to add to the Jamfile for jam to link it with my file. the file that I am adding BufferQueue.h is in src/add-ons/kernel/drivers/network/tun
<The_Ringmaster>
is there a way to force haiku to use a certain graphics driver like vesa for example?
<PulkoMandy>
Yes, in the kernel settings file or from the boot menu
<The_Ringmaster>
where is that file?
<The_Ringmaster>
I found a vesa file under kernel/drivers, I'm guessing that's it
The_Ringmaster has quit [Quit: Vision[]: i've been blurred!]
Maturi0n has joined #haiku
Maturi0n_ has quit [Ping timeout: 480 seconds]
The_Ringmaster has joined #haiku
<The_Ringmaster>
yeah so that vesa file did work just set it to 1920 1080 32 and it worked upon reboot, nice!
The_Ringmaster has quit [Quit: Vision[]: i've been blurred!]
remowo has joined #haiku
<remowo>
Evening everyone ...
<remowo>
would it be wise to even think of installing haiku on asus-meepc1001
<erysdren>
good evening remowo
HaikuUser has joined #haiku
The_Ringmaster has joined #haiku
<HaikuUser>
gimp installed from haiku depot wont start
<HaikuUser>
thanks all for help in advance
Diver has quit [Quit: Leaving.]
<HaikuUser>
hello is there a way to start gimp it does not show in applications menu
HaikuUser2 has joined #haiku
phcoder has quit [Quit: Connection closed for inactivity]
HaikuUser2 has quit []
kai has joined #haiku
kai is now known as Guest3522
<Guest3522>
Hello! I'm trying out the system and have a (probably) noobish question, how do I make /boot/system writable? I'm trying to install a library by `make install`
<The_Ringmaster>
gimp starts fine here from the app menu on the latest nightly
HaikuUser has quit [Quit: Vision[]: i've been blurred!]
<BlueSky76>
Guest3522: You can't make /boot/system writable, it is designed to be read-only, because of Haiku's quite unique packaging system. There are the non-packaged folders in /boot/system or /boot/home that you can use. They are writable.
Vidrep_64 has joined #haiku
Vidrep_64 has quit []
<Guest3522>
BlueSky76 thanks! didn't notice I had to point cmake to set the install directory as /boot/system/non-packaged that helped a lot!
freakazoid332 has joined #haiku
<The_Ringmaster>
has gcc been upgraded to version 12 yet?
jason123santa has left #haiku [Error from remote client]
<BlueSky76>
Guest3522: Great. Glad that it worked :-)
frkzoid has quit [Ping timeout: 480 seconds]
frkazoid333 has quit [Ping timeout: 480 seconds]
Udi has quit [Quit: Vision[]: i've been blurred!]
frkazoid333 has joined #haiku
Udi has joined #haiku
<Udi>
can you recommend IRC client for IOS (in case the nightly build breaks my Haiku install :)
<The_Ringmaster>
igloo is an irc client for ios.. it might work
<waddlesplash>
there's a webirc instance for oftc I'm pretty sure
<Udi>
thanks!
The_Ringmaster has quit [Quit: Vision[]: i've been blurred!]
<pairisto[m]>
I want to add BufferQueue.h from src/add-ons/kernel/network/protocols/tcp but I am not currently sure what to add to the Jamfile for jam to link it with my file. the file that I am adding BufferQueue.h is in src/add-ons/kernel/drivers/network/tun. I put SubDirHdrs [ FDirName $(HAIKU_TOP) src add-ons kernel network protocols tcp ] ; in the Jamfile and I get -iquote src/add-ons/kernel/network/protocols/tcp when compiling but it's still
<pairisto[m]>
not including it as I get undefined reference errors?
jason123santa has joined #haiku
phcoder has joined #haiku
<phcoder>
@waddlesplash: I can confirm: OpenBSD fails the same way
dryphb has quit [Remote host closed the connection]