austriancoder changed the topic of #etnaviv to: #etnaviv - the home of the reverse-engineered Vivante GPU driver - Logs https://freenode.irclog.whitequark.org/etnaviv
pcercuei has quit [Quit: dodo]
gbisson has quit [*.net *.split]
shoragan has quit [*.net *.split]
shoragan has joined #etnaviv
Net147 has quit [Quit: Quit]
Net147 has joined #etnaviv
Net147 has quit [Client Quit]
Net147 has joined #etnaviv
pcercuei has joined #etnaviv
mntmn has joined #etnaviv
<mntmn> time to test-drive the latest mesa on gc7000l...
<mntmn> now building with !7603 !9310 !9780 !9871
<mntmn> anything else i should test?
<marex> mntmn: neverball
<marex> mntmn: btw have you ever used the devcoredump with etnaviv ?
<marex> austriancoder: any hint on what to look for in the decoded traces ?
<mntmn> marex: will test. yes, i have used that but like forever ago, so my memory of that is near 0
<austriancoder> marex: that link is dead
<marex> mntmn: the patches are now in, so it should work
<marex> austriancoder: which link ?
<marex> oh
<mntmn> marex: yep, i saw
<austriancoder> marex: maybe send me the dumb via mail
<marex> austriancoder: I would like to learn it though :)
<marex> austriancoder: anyway, see email
<austriancoder> marex: maybe its time for a blog post.. so everybody can read up the knowledge
<marex> austriancoder: I think that's long overdue in fact
<austriancoder> marex: fine.. I will prep something - but now I am out for a short bike tour as the weather here is so wonderful
<marex> austriancoder: no worries, have fun
<mntmn> ok, at least one of those 4 patches made glxinfo segfault and cause really funky corruption. i'm now on mesa master without any extra MRs.
<mntmn> marex: neverball looks good!
<mntmn> incl. sprites
<marex> mntmn: reflective surfaces are still broken (the NEVERBALL text in main menu) I think
<marex> mntmn: good thing about the point sprites, thanks for checking
<mntmn> i have a good new pathological case
<mntmn> it was already there in my last mesa build, but is still there on master: dropdowns in firefox cause a flickery mess
<mntmn> wayland-native firefox, that is
<marex> mntmn: hold on
<marex> mntmn: try applying weston MR570
<marex> (if you are on weston that is)
<mntmn> i'm using sway :D
<marex> mntmn: see the MR anyway, and MR568
<marex> mntmn: it might just be related
<mntmn> something about keyboard?!
<mntmn> marex: the bug that i'm seeing is def. an etnaviv problem. i'll make short video
<marex> mntmn: yeah, on keyboard-less devices, the chromium dropdown menus were randomly placed with weston , it worked on devices with keyboards , problem was in top level surface focus
<mntmn> marex: i think that is extremely unrelated
<marex> ok
<mntmn> one sec, i'll show you
<mntmn> it's this typical double buffer confusion that can happen from time to time, but in this case it's 100% reproducible
<mntmn> will now try some MRs one by one
<mntmn> austriancoder: hey, last time i checked !9871 i probably made a mistake. it absolutely works. fixes alpha test in kicad 3d viewer for example
<mntmn> marex: maybe not a etnaviv bug after all, just looks like one https://bugzilla.mozilla.org/show_bug.cgi?id=1552020
<mntmn> > This issue was identified as a damage tracking issue in sway for subsurfaces attached to xdg_popups
<mntmn> upgraded sway to latest master and it is indeed fixed.
<mntmn> nice.
<mntmn> interesting, firefox with accelerated layers now works on latest mesa (before, there was a shader compiler crash). but introduces new interesting flickering.
<marex> mntmn: now that looks like my glmark2 cat pictures corruption :)
<marex> mntmn: oh ...