ChanServ changed the topic of #zink to: official development channel for the mesa3d zink driver || https://docs.mesa3d.org/drivers/zink.html
pbsds has quit [Quit: The Lounge - https://thelounge.chat]
pbsds has joined #zink
sinanmoh- has quit []
sinanmohd has joined #zink
_xav_ has quit [Ping timeout: 480 seconds]
_xav_ has joined #zink
pendingchaos has quit [Ping timeout: 480 seconds]
sinanmoh- has joined #zink
sinanmohd has quit [Ping timeout: 480 seconds]
pendingchaos has joined #zink
pendingchaos has quit [Read error: No route to host]
pendingchaos has joined #zink
pendingchaos has quit [Read error: No route to host]
pendingchaos has joined #zink
pendingchaos has quit []
pendingchaos has joined #zink
pendingchaos has quit [Read error: No route to host]
pendingchaos has joined #zink
<anholt> hmm, interface-blocks-api-access-members seems to have gotten slow for zink recently, across anv and radv. https://gitlab.freedesktop.org/mesa/mesa/-/jobs/39284421
<zmike> 🤔
i509vcb has joined #zink
<i509vcb> The docs for zink describe what is needed for desktop GL but do not list what gles needs? https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/drivers/zink.rst I assume that is just a documentation gap
<zmike> yes, it is
omegatron has quit [Quit: What happened? You quit!]
* anholt really wonders what zink_kopper_present_readback is supposed to do.
<zmike> zink_kopper_acquire_readback acquires the swapchain for readback, zink_kopper_present_readback restores the swapchain to its previous state
<anholt> so present is release
<anholt> but also, it seems to queue a present
<zmike> hm not quite
<zmike> yes, that's correct
<zmike> acquire has to shuffle through the swapchain images to reacquire the last-presented image
<zmike> and present re-presents it
<anholt> wait, why would you need to shuffle through?
<zmike> because there's no api to request a specific image
<anholt> what's the circumstance that you need the last-presented image, though?
<zmike> basically just piglit tests for swapchain readback
<anholt> (context: I'm trying to debug why buffer age is completely broken)
<zmike> ah you brave soul
<anholt> if there are any piglit tests assuming that reading the back buffer gives the last presented image's contents without checking buffer age, that would be broken.
<anholt> got a test name?
<zmike> hm
<zmike> I wanna say the read-front tests?
<zmike> but it's been a long time