ChanServ changed the topic of #zink to: official development channel for the mesa3d zink driver || https://docs.mesa3d.org/drivers/zink.html
omegatron has quit [Ping timeout: 480 seconds]
illwieckz has quit [Ping timeout: 480 seconds]
illwieckz has joined #zink
omegatron has joined #zink
<ajax> there's an MR for swapchain_m1 i thought. still too new to require?
<zmike> yes
<zmike> I'll add it as optional use during 23.1 cycle and require for 23.2
<zmike> or something like that
avane has quit [Ping timeout: 480 seconds]
<ajax> i feel like i'm missing something. what's hard about destroying them?
<zmike> knowing when they're no longer in use
<ajax> you can't seriously need to wait for all the images to go idle, that's madness
<ajax> they're not yours, you're just holding a reference
<ajax> that's why you asked the wsi to allocate them
<zmike> I'm not super confident that's how all implementations see it
<ajax> blah
<ajax> fine, rb
avane has joined #zink
<anholt> there's a longstanding validation error about VK objects not being destroyed by vkDestroyDevice time -- any clues on that one? Unfortunately the zink built-in validation layer stuff is not a great help for it, since we tear down validation reporting before vkDestroyDevice.
<zmike> uhhh
<zmike> there's some cases where gallium objects don't get destroyed?
<zmike> I've debugged a few of them and there's nothing to be done on the zink side
<zmike> but I haven't done much validation stomping outside of ZINK_DEBUG
<zmike> if you file a test case I can put it at the top of my list for tomorrow
* zmike is refactoring descriptor code again