Akari has quit [Quit: segmentation fault (core dumped)]
Akari has joined #zink
fahien has joined #zink
fahien has quit [Ping timeout: 480 seconds]
fahien has joined #zink
fahien has quit [Ping timeout: 480 seconds]
fahien has joined #zink
fahien has quit [Ping timeout: 480 seconds]
<zmike>
ajax: for #18780 maybe kopper_init_screen should try and create a swapchain before returning?
<zmike>
not really sure how else to resolve it
<ajax>
:/
<ajax>
that seems like solving the wrong problem
<zmike>
yeah I dunno
<ajax>
lemme see if i can repro the crash
<zmike>
worked easy for me with glxinfo
<ajax>
i don't understand what michel means there about swrast fallback
<zmike>
so the issue there is that it's running in an env where it should be using a sw driver
<zmike>
and other drivers detect this by checking for x11 dri extensions
<zmike>
but zink doesn't use that
<zmike>
so after e.g., iris fails, it tries zink and then zink succeeds
<zmike>
but zink can't create a swapchain because it's running on anv and dri3 isn't available
<zmike>
so it fails late
<ajax>
so...
<ajax>
ugh. dri2_initialize is miserable.
<ajax>
so this is kind of tricky
<ajax>
what i should do
<ajax>
is teach xvfb about glamor, on surfaceless
<ajax>
and then make it so surfaceless zink works on whatever devices it can find even if they can't "present", since there's no surfaces to present to.
<zmike>
kinda still doesn't solve this problem though?
<ajax>
sorry, i skipped over a piece
<ajax>
"make it so" includes threading enough context down to zink _display_ creation that it can know which extensions it needs to support
<ajax>
like, say, KHR_present and KHR_xcb_surface
<zmike>
ah
<ajax>
and then surfaceless/device just doesn't ask for much
<ajax>
because that's the bug here, right, is zink is trying to find a gpu to run on without knowing whether it's even connected to the same display as the frontent