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 [Read error: Connection reset by peer]
eukara has joined #zink
LexSfX has quit []
LexSfX has joined #zink
fahien has joined #zink
fahien has quit [Ping timeout: 480 seconds]
fahien has joined #zink
<zmike> ajax: any chance while you're rooting around in wsi/sw you can check out that glx visual ticket I filed
<zmike> that's the most common ci flake these days
<ajax> sure
<zmike> awesome, thanks
<ajax> aaaaaand now i'm nerdsniped
<zmike> you found it?
<ajax> i found something anyway
<zmike> it's weird right?
<ajax> i don't get xio error though
<zmike> hm
<ajax> glx-visuals-stencil: ../src/gallium/drivers/zink/zink_resource.c:1125: resource_create: Assertion `res->obj->dt' failed.
<zmike> what the
<ajax> not always on the same iteration, afaict, but so far it's every time i run it
<zmike> never gotten that one
<zmike> X Error of failed request: BadValue (integer parameter out of range for operation)
<zmike> Major opcode of failed request: 149 ()
<zmike> Value in failed request: 0x3401239
<zmike> Minor opcode of failed request: 4
<zmike> Serial number of failed request: 11012
<zmike> Current serial number in output stream: 11014
<zmike> I get this
<ajax> ooookay. what does 'xdpyinfo -queryExtensions | grep 149' give you on that machine?
<zmike> DRI3 (opcode: 149)
<zmike> this is on regular xorg, not xwayland
<ajax> #define X_DRI3FenceFromFD 4
<ajax> which
<zmike> ?
<zmike> huh
<ajax> is it consistently that major/minor in the error?
<zmike> yep
<zmike> every time
<zmike> wish I could repro your issue
<ajax> k. if it wasn't i might think the request stream was getting corrupted and random data was being misinterpreted as that dri3 request, and it'd just have fallen over because there was no fd sent
<ajax> but if it's consistent then probably we really are passing a bad fd across?
<zmike> seems that way
<zmike> this is on anv btw
<ajax> but, how do you get there, we literally just checked that it was valid before passing it to xcb_dri3_fence_from_fd
<ajax> it's a brand new memfd
<zmike> kusma: did you have further comments on https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17525 now that I've added the env var handling
<kusma> zmike: commented
<kusma> sorry for dropping the ball there ;)
<zmike> np np
<zmike> my cts runs will now crash more helpfully
<kusma> TBH, this sounds like it could be useful to other drivers if it was added as a CTS feature ;)
<zmike> yeah maybe I'll just rewrite it to use meson while I'm in there too
<kusma> ...
<zmike> :D
<kusma> speaking of nothing... I just read your sad trumpet noises blog post... Should we maybe document extensions that really, really, really should be supported for perf reasons also? :P
<zmike> I have a ticket for tracking this
<zmike> documentation is pending
<zmike> ish
<kusma> OK, I'm just about to start wiring up Zink on top of another Vk driver, and it would be nice to have the docs to give to $VENDOR ;)
<kusma> Right now perf isn't a big issue, but I'm sure it will become a priority pretty quicky :)
<kusma> Nice, thanks!
<zmike> and https://gitlab.freedesktop.org/mesa/mesa/-/issues/6488 was the ticket for tracking questionably documented features
<zmike> that are required
<zmike> so maybe the docs need improving on that one
<kusma> I'll take a look, thanks!
<zmike> also have you had a chance to look at hud?
<kusma> I have... completely forgotten about that!
<zmike> sad trumpet noises
<kusma> I have time for stuff like that these days, so I'll give it a stab!
<zmike> awesome
<zmike> thanks
<kusma> np :)
<kusma> OK, I guess the problem is the font is... just a box, right?
<zmike> not sure I've seen that
<kusma> Ah, missing. The black box is the background for it
<zmike> I don't get a graph
<kusma> Oh, I get a graph
<zmike> hm maybe I do get a graph
<zmike> no I don't get a graph
<zmike> or...no I get one it's just hard to see because it's maxed out at the top
<zmike> but no text
<kusma> yeah, so... texture format or texture-rect issue?
<ajax> there, _now_ i get an XIO error
<ajax> and i do it by nerfing the (basically always successful) XVisual support check in x11_surface_get_support
<ajax> so... kind of still suspecting request buffer corruption.
<ajax> yeah, and then rewriting swrastGetDrawableInfo in xcb makes it happen even less
<ajax> great! great.
<ajax> X Error of failed request: BadIDChoice (invalid resource ID chosen for this connection)
<ajax> yeah. definitely having fun over here.
omegatron has quit [Quit: What happened? You quit!]
<kusma> zmike: I have a font :)
<kusma> zmike: ...and you broke it when you switched to state-tracker lowering of texture rectangles!
<kusma> Well, broke it for Zink. Seems it was already broken for some other drivers.
<ajax> [screaming intensifies]
<zmike> I've been in border colors again, so I'm right there with you https://giphy.com/embed/n7TyooZtZn3Ko
<zmike> kusma: sounds like the st handling is broken then?
<kusma> No, this doesn't go through the st
<zmike> 🤔
<kusma> Similar to blits
<kusma> Just calls directly into the gallium driver
<ajax> why is it impossible to set a catchpoint for function return
<ajax> have the gdb developers not used C before
<zmike> that would be pretty amazing
<ajax> ugh and how did i downgrade to a non-threadfixed xlib
<ajax> zmike: !17685 helps, here
<zmike> I'm getting beachballed by gitlab
<zmike> ajax: hm now I get a different error
<zmike> X Error of failed request: BadDrawable (invalid Pixmap or Window parameter)
<zmike> Major opcode of failed request: 149 ()
<zmike> Minor opcode of failed request: 4
<zmike> Resource id in failed request: 0x3201257
<zmike> Serial number of failed request: 11186
<zmike> Current serial number in output stream: 11188
<ajax> ooh. yeah okay, same deal there i guess.
<ajax> please wait, reticulating splines
* zmike twiddles
<ajax> this is actually kinda fun, now
<ajax> you can just grep for xcb_discard_reply and be pretty sure that every time you find it it's a bug
<zmike> haha
<zmike> who would do such a thing!
Akari has quit [Quit: segmentation fault (core dumped)]
Akari has joined #zink
<ajax> zmike: can you try the zmike-tmp branch i just pushed?
<zmike> sigh
<zmike> great prank ajax
<zmike> killed my session
<ajax> !
<ajax> how the
<ajax> (sorry about that)
<zmike> probably should've tested on my test machine
<zmike> ETOOMUCHHUBRIS
fahien has quit [Quit: fahien]
fahien has joined #zink
fahien has quit []
Soroush_ has joined #zink
Soroush has quit [Read error: Connection reset by peer]
Soroush_ has quit []
Soroush has joined #zink