AlienSoldier has quit [Quit: Vision[]: i've been blurred!]
<thomaslewis>
I'm getting some odd behavior when I press CapsLock. The system hangs for a couple of seconds, then dumps everything I've typed all at once. It also happens when I turn it off. Any idea where to look for the cause?
<waddlesplash>
wow, never heard of that one before
<waddlesplash>
you should probably just file a ticket and attach the usual info
<thomaslewis>
It's not one particular application.. Maybe I should update (I'm on hrev56148).
<jessicah>
no recent changes should affect the keyboard
<thomaslewis>
How is the CapsLock toggle actually handled (or where)?
<jessicah>
input server
<thomaslewis>
It's fine for a while after the system starts, but after 15-20 mins, it starts that really annoying behavior. I think it's just the input thread that's hanging.
<jessicah>
iirc, something like set_modifier_keys or such
<jessicah>
which, iirc, involves sending a BMessage... can't quite recall
<jessicah>
oh, that might be something different... seems like IS_SET_KEYBOARD_LOCKS is the magic sauce
* thomaslewis
goes to grep the source
<jessicah>
which sends a message from input server to input server addon with B_KEY_LOCKS_CHANGED
<jessicah>
so yeah, it involves a little bit of messaging
<thomaslewis>
Yeah. It's in a MessageReceived() override in InputServer.cpp. It sends a message to HandleSetKeyboardLocks().
<jessicah>
which sends to the keyboard add-on, which updates internal state... it grabs a lock, which is the only reason I can think of that it'd be taking a while
<jessicah>
that, or messaging itself has become borked
<thomaslewis>
Looks like the definitionn of HandleSetKeyboardLocks is in InputServer.cpp (line 841).
<thomaslewis>
Looks like the definition of HandleSetKeyboardLocks is in InputServer.cpp (line 841).
<jessicah>
which messages the addon
<jessicah>
see src/add-ons/input_server/devices/keyboard/KeyboardInputDevice.cpp:552
<jessicah>
I would probably start with adding timing to the function, might show if the problem is in updating the keymap state
<thomaslewis>
Looks like it treats a keymap change and a lock change as the same thing (which it technically is (I guess)).
<thomaslewis>
LEDs aren't getting updated, either (even though the add-on specifically calls KeyboardDevice::_UpdateLEDs()).
<jessicah>
at all, or just sometimes?
<thomaslewis>
Once the weird delay starts, the LED state is locked to whatever it was when the state change lag starts.
<jessicah>
that would imply that fFD has been unset
<thomaslewis>
At first, I thought my laptop was overheating (it uses the CapsLock LED to warn of heat events).
<jessicah>
or the ioctl failed
xet7 has joined #haiku
<jessicah>
:o maybe that's confusing the driver
<jessicah>
depending how it changes the LED
<jessicah>
couldn't add an extra LED? :p
<thomaslewis>
Well, it didn't seem to be overheating, and the UEFI should only hijack the OS ioctl if it's indicating a weird status.
<thomaslewis>
It's Dell, so...
* thomaslewis
shrugs.
<jessicah>
well, UEFI runtime services aren't present in Haiku, so UEFI can't affect Haiku after boot
<jessicah>
it'd have to be done in some other way if it works in Haiku
<jessicah>
you could comment out the `_UpdateLEDS` if you wanted, to see if it's something in there going haywire, in addition to adding timing debug to syslog
<thomaslewis>
Looks like it's the USB keyboard watcher... Wait a sec. I'm using a Logitech trackball with the Unified Receiver. I wonder if Haiku is trying to update the "USB keyboard" the receiver exposes even though it doesn't exist...
<jessicah>
it could be a possibility too
<jessicah>
who knows with all these homebrew usb wireless dongles
<thomaslewis>
Okay, so even with the receiver unplugged, the Input preferences shows an AT keyboard and a USB keyboard. That makes me wonder which one is actually the keyboard embedded in the laptop shell.
<thomaslewis>
A new keyboard showed up when I plugged it in. So Haiku thinks I now have an AT Keyboard, PS/2 Touchpad, USB Keyboard, USB Mouse, USB Tablet, USB Keyboard, USB Mouse, USB Tablet.
<thomaslewis>
So it stands to reason that if I unplug the receiver, reboot, and connect the trackball via Bluetooth, the issue should stop because the add-on won't be trying to update non-existent devices.
<jessicah>
it's a theory :) and not surprising the number of USB HID showing up... I see some crazy stuff even in Windows
<jessicah>
I have one mouse, one keyboard, but two mice in windows, and 10 USB HID devices :p
<thomaslewis>
Well, the bluetooth server couldn't see my trackball.
skirst has joined #haiku
OrangeBomb has quit [Quit: Slacking off]
kikadf_ has joined #haiku
kikadf is now known as Guest1703
kikadf_ is now known as kikadf
nephele has quit [Ping timeout: 480 seconds]
Guest1703 has quit [Ping timeout: 480 seconds]
OrangeBomb has joined #haiku
ablyss_ has quit [Quit: Leaving]
<waddlesplash>
well, that was a lot in one day
<waddlesplash>
hopefully tomorrow I'll start finding solutions to the regressions
<waddlesplash>
jessicah: you dig out that 8265 device? would be good to test, and if it doesn't work, to have a dev with the device to try fixes :-p
<jessicah>
not yet
<jessicah>
it's my desktop system :)
<waddlesplash>
I've gotten a 9xxx device but it's also my desktop :-p
<waddlesplash>
I'll reboot to Haiku in a few days and see what happens
<waddlesplash>
hopefully by then I'll have figured out what ails kim1963's hardware for iprowifi4965
<thomaslewis>
I found a bunch of PDFs with the Logitech HID++ protocol specs. How difficult are Haiku USB drivers to write?
ablyss has joined #haiku
HaikuUser has joined #haiku
HaikuUser has quit []
Bolin has joined #haiku
Bolin has quit []
<waddlesplash>
thomaslewis: you probably only need a small patch to the existing driver.
<waddlesplash>
first determine precisely where the hang is occurring
<waddlesplash>
then determine why and how we get there, etc.
vdamewood has joined #haiku
<thomaslewis>
It's due to how the Unifying receiver reports devices if the DJ Enumerator isn't available (it's a bit of SW that specifies what types of devices are paired to each of the 6 or 7 possible device IDs in the receiver). If the enumerator isn't present, it presents all HID interfaces for all the devices.
<thomaslewis>
So, the receiver is registering my trackball as a keyboard, mouse, and tablet (touchpad).
<thomaslewis>
As I understand it (from the Logitech docs), the Enumerator sits between the receiver and the HID drivers.
<thomaslewis>
I need to slow down, though. I need to finish learning how to work with the Haiku API, then *maybe* think about modifying the usb_hid driver or writing the enumeration layer.
ablyss has quit [Read error: Connection reset by peer]
ablyss has joined #haiku
rennj has quit [Ping timeout: 480 seconds]
rennj has joined #haiku
skirst has quit []
skirst has joined #haiku
Begasus has joined #haiku
<Begasus>
g'morning peeps
<netpositive>
morning
<Begasus>
morning netpositive
Begasus_32 has joined #haiku
skirst has quit []
B2IA has quit [Ping timeout: 480 seconds]
HaikuUser has joined #haiku
HaikuUser has quit []
Atomozero has joined #haiku
DKnoto has quit [Ping timeout: 480 seconds]
DKnoto has joined #haiku
Bolin has joined #haiku
eroux has joined #haiku
AlaskanEmily has quit [Remote host closed the connection]
skirst has joined #haiku
Atomozero has quit [Quit: Vision[]: i've been blurred!]
skirst has quit []
skirst has joined #haiku
mmu_man has joined #haiku
Bolin has quit [Quit: Vision[]: i've been blurred!]
wicknix has quit [Remote host closed the connection]
wicknix has joined #haiku
_paul0 has joined #haiku
paul0 has quit [Ping timeout: 480 seconds]
MajorBiscuit has joined #haiku
Begasus has quit [Ping timeout: 480 seconds]
Begasus has joined #haiku
vdamewood has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
mmu_man has quit [Ping timeout: 480 seconds]
Letterus[m] has joined #haiku
mmu_man has joined #haiku
tqh has joined #haiku
mmu_man has quit [Ping timeout: 480 seconds]
mmu_man has joined #haiku
tuaris has quit [Read error: Connection reset by peer]
<trungnt2910[m]>
Somehow Haiku's Debugger does not show disassembly for functions without a name.
<trungnt2910[m]>
Any way to work around this?
Gaspare has joined #haiku
mmu_man has quit [Ping timeout: 480 seconds]
x10z has joined #haiku
Gaspare has quit [Quit: Gaspare]
B2IA has joined #haiku
<kallisti5[m]>
<trungnt2910[m]> "image.png" <- The application needs to be built with debugging symbols enabled
<kallisti5[m]>
generally, gcc -g
<kallisti5[m]>
there's also a way to install _debug packages if they exist with symbols... but I don't know much about that
<trungnt2910[m]>
I doubt this is related in any way to debugging symbols.
<trungnt2910[m]>
What I'm trying to do is to view the disassembly of an unnamed function (for example, code that is the result of a C++ lambda).
doppler has quit [Ping timeout: 480 seconds]
doppler has joined #haiku
<kallisti5[m]>
ah yeah, i see what you're saying
<kallisti5[m]>
none of the functions in the backtrace in the top right pane show asm?
<kallisti5[m]>
at minimum the assembly should be shown
<trungnt2910[m]>
No.
<trungnt2910[m]>
They don't show any disassembly on the source view
<trungnt2910[m]>
For other functions (with a defined name), the disassembly shows normally.
MajorBiscuit has quit [Quit: WeeChat 3.5]
mmu_man has joined #haiku
frkazoid333 has joined #haiku
mmu_man has quit [Ping timeout: 480 seconds]
B2IA has quit [Quit: Vision[]: i've been blurred!]
B2IA has joined #haiku
eroux has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<B2IA>
(bbjimmy_64) ok
mmu_man has joined #haiku
<mbrumbelow[m]>
Updated to 56177... wifi dead.
<mbrumbelow[m]>
Filing bug report with Syslog. No wireless networks in the list.
mmu_man has quit [Ping timeout: 480 seconds]
tuaris has joined #haiku
<waddlesplash>
mbrumbelow[m]: driver?
<waddlesplash>
trungnt2910[m]: it's a bug, there's a ticket about this somewhere iirc
Gaspare has joined #haiku
<trungnt2910[m]>
waddlesplash: Where, can't search from Google (I found something about "better disassembly", upvoted it, but it's not the same issue).
<trungnt2910[m]>
I'm amazed at how you can dig up issues from 8 years ago.
<waddlesplash>
well, first of all, I vaguely remembered the issue, so that helps
<waddlesplash>
it might have even been in my browser history if I remembered what keywords to use
<waddlesplash>
but really I just know how we organize the bugtracker and so I can quickly find it if it's there to be found
<waddlesplash>
we are pretty diligent about proper usage of components. thousands of tickets get unmanageable without them :)
eroux has joined #haiku
<waddlesplash>
(I don't remember it from 8 years ago though, I've run into it more recently than that)
<trungnt2910[m]>
> The symbol isn't actually flagged as a function in this case. Consequently, NextSymbol() skips it. Not sure of a good way around that though.
<trungnt2910[m]>
Well isn't it a bad idea to run the disassembler based on symbols?
<waddlesplash>
we need to know where the function begins to start disassembling
<waddlesplash>
we can disassemble starting at PC but that won't be very useful now will it?
<waddlesplash>
so symbols are used for this, usually
<waddlesplash>
however we clearly need an alternative mechanism for when we aren't in a symbol indeed. Not sure what that mechanism should be, I'm not really an expert here
<waddlesplash>
looking at what LLDB does, if you can figure that out, may be of use. I'd bet something like looking at the calling instruction and seeing where the first jump was to
<trungnt2910[m]>
If that's the case the Debugger should have some kind of fallback.
<trungnt2910[m]>
If no code is available, fall back by disassembling according to the addresses on the stack trace?
<waddlesplash>
again: we don't want to start disassembling at the program counter, we want to start at the beginning of the function
<waddlesplash>
also the disassembler in Debugger is *really* slow, I keep meaning to look into that, it's very annoying
jmairboeck has joined #haiku
<trungnt2910[m]>
Also, how much work was it to bring that old version of GDB to Haiku and how much will it take to bring something similar but newer (like LLDB?) to Haiku?
<waddlesplash>
a lot of work, you have to write a whole backend
<waddlesplash>
anyway I attempted a port of a much newer GDB using the old port's code
<waddlesplash>
it got pretty far, but it fails to start applications or something, I can't remember
<waddlesplash>
the code's in HaikuPorts if you want to look at it
<trungnt2910[m]>
Haiku's Debugger has a nice UI and stuff but the "nice" UI is the reason why I disliked it. It's so slow....
<waddlesplash>
the slowness is mostly the disassembler I think lol
<waddlesplash>
at least that's the immediate slowness of switching functions
<trungnt2910[m]>
waddlesplash: No offense but is that the one which has been given a one star rating?
<waddlesplash>
probably not, it should be disabled?
<waddlesplash>
if someone enabled it then they should not have
<waddlesplash>
it could be the old gdb 6.8
<trungnt2910[m]>
Hmmm so I found the wrong GDB.
<waddlesplash>
well, gdb 6.8 did work, it's just very old
<waddlesplash>
anyway, Debugger's slow, but can't investigate that because the profiler is currently mostly broken, which is probably a kernel timer interrupts problem, but that only manifests on some systems and not others ...
<waddlesplash>
maybe that'll be the next "yak shave" after I fix all the regressions I caused in WiFi :-p
<trungnt2910[m]>
I'll have a look at the GDB in HaikuDepot again and I'll try it.
Gaspare has joined #haiku
m199 has joined #haiku
Gaspare has quit [Quit: Gaspare]
<mbrumbelow[m]>
waddlesplash: Does not seem to be the driver.
<waddlesplash>
mbrumbelow[m]: no, what's the driver?
floof58 has quit [Remote host closed the connection]
<waddlesplash>
it's probably a driver problem
<mbrumbelow[m]>
LOL ok.
<waddlesplash>
after all the drivers are what changed in recent days...
mmu_man has joined #haiku
floof58 has joined #haiku
HaikuUser has joined #haiku
HaikuUser has quit []
<mbrumbelow[m]>
would that be something like: idualwifi7260?
<mbrumbelow[m]>
See ticket: 17782
<waddlesplash>
mbrumbelow[m]: correct
eroux has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Anarchos>
rennj i see the slider moving on the window, but the screen brightness doesn't change. With the mouse, it changes as due.
rennj has quit [Quit: Leaving]
rennj has joined #haiku
<rennj>
woops
<rennj>
closed the window..i got broken mouse button needs cleaning or replacing
* Anarchos
: no trouble rennj
<rennj>
walmart only sells cheap mice
<Anarchos>
rennj you can try in Terminal : Screen & sleep 1;hey -o Screen SET Value OF View 2 OF View 0 OF View 0 OF Window 0 TO 256 ;sleep 1; kill Screen
<rennj>
In 1991 Steve Jobs' company commissioned an head-to-head programming competition to show how much faster and easier it was to program a NeXT computer vs a Sun workstation.
<Anarchos>
augiedoggie i wasn't aware of this command. Anyway i don't understand why the scripting does not work
<augiedoggie>
i haven't looked at that part
<Anarchos>
augiedoggie tell me if you want the command line i used with hey to script Screen
AlaskanEmily has joined #haiku
<augiedoggie>
i'm not sure if i can help too much, i don't have that slider in my screen preferences because my displays don't support that