ChanServ changed the topic of #zink to: official development channel for the mesa3d zink driver ||
illwieckz has quit [Quit: I'll be back!]
illwieckz has joined #zink
sinanmoh- has joined #zink
sinanmohd has quit [Ping timeout: 480 seconds]
sinanmoh- has quit []
sinanmohd has joined #zink
<zmike> ajax: nice, thanks!
<ajax> zmike: and my apologies for being hard to reach
<zmike> np, we'll get it done
<ajax> pretty sure michel won't like that patch but functionally it should be enough for you to test and see if it solves the deadlock
<ajax> if someone wants to be all Proper and add a new event type to Present and bump the protocol minor they can be my guest, i ain't about that life anymore
<zmike> I'm not sure why it would matter? is there really another case where a 0x0 present event is significant?
<daniels> you can never do a 0x0 present, because you can never have a 0x0 Pixmap, and you have to have a Pixmap to present
<zmike> that was my understanding as well
<ajax> daniels: this is sending a ConfigureNotify with a 0x0 size when the window is destroyed
<zmike> oh I misread
<daniels> ajax: PresentConfigureNotify rather than core CN, right?
<ajax> daniels: yeah
* zmike didn't misread
<ajax> xcb client libraries can't really use core events so we have to tunnel it into Present
<daniels> right
<ajax> the risk is some client might notice it and try to allocate something 0x0 or just mention the old window XID
<ajax> but... it's gone, go ahead and crash already
<daniels> right, as you say either they're going to catch BadWindow forever, or they'll catch BadMatch when they try to do their 0x0 Pixmap
<daniels> if doing the latter fixes a real bug that's old enough to sit in the front seat of the car, then cool?
<ajax> yeah. crash > deadlock here.
<daniels> ++