ChanServ changed the topic of #zink to: official development channel for the mesa3d zink driver || https://docs.mesa3d.org/drivers/zink.html
eukara has quit [Ping timeout: 480 seconds]
eukara has joined #zink
eukara_ has joined #zink
eukara has quit [Ping timeout: 480 seconds]
LexSfX has quit [Ping timeout: 480 seconds]
LexSfX has joined #zink
ced117 has quit [Ping timeout: 480 seconds]
ced117 has joined #zink
LexSfX has quit [Ping timeout: 480 seconds]
LexSfX has joined #zink
<ajax> zmike: so that undying gears problem
<zmike> yeah
<ajax> i take it the wsi thread is stuck waiting for a present event that's not coming?
<zmike> no
<zmike> uhh
<zmike> maybe
<zmike> let me check
<zmike> no
<zmike> looking at this, the driver is just drawing more frames endlessly but has nothing to draw to
<ajax> well. for glx that error condition is called GLXBadCurrentDrawable and you can get it from any of WaitX WaitGL or most likely SwapBuffers
<ajax> do we just not raise that ever?
<zmike> seems not?
<zmike> you should be able to repro it with glxgears on zink+nv
<ajax> woof no we do not.
<zmike> but you have to actually close the window
<ajax> oh good. we only generate that error's deprecated cousin BadCurrentWindow, in the "i'm running on osx and pointed at X11.app" accel path
<ajax> so that's excellent.
<zmike> useful
<ajax> my nv machine is not easily powered on atm, happen to have a backtrace handy?
<zmike> uhh
<zmike> backtrace for what?
<ajax> gears, when it's refusing to die.
<zmike> just fire up my MR and click the close button on glxgears
<zmike> #13 0x0000555555557735 in draw ()
<zmike> #14 0x0000555555557148 in main ()
<zmike> then it's inside mesa
<zmike> spinning away in futility
<ajax> datura:~/git/zink% zink glxgears
<ajax> MESA: error: zink: failed to update swapchain capabilities
<ajax> X connection to :0 broken (explicit kill or server shutdown).
<ajax> KILL 0x562792597c40
<ajax> and then back to the prompt
<zmike> that's on nv?
<ajax> no, like i said, not easily accessible right now
<zmike> oh I misread
<zmike> yeah that's what happens on every other driver
<zmike> I guess different x11 wsi
<ajax> yeah
<zmike> maybe we want something in the kopper frontend that can notice when the swapchain is killed and simulate an x error?
<ajax> yeah, killed swapchains should no-op for everything but actual wsi drawable action, whichever one we hit from SwapBuffers needs to thread an error return back up
<ajax> s/action/inter&/
<zmike> zink_kopper_update is actively returning false with my MR
<zmike> so then it's up to the frontend to handle that
<ajax> i can do that. one sec...
Akari` has joined #zink
Akari has quit [Ping timeout: 480 seconds]
<zmike> ajax: what branch is this
<zmike> oh
<zmike> duh
<ajax> !16038 rebased to main plus a thing
<zmike> ajax: no change
<ajax> hmph
cheako has joined #zink