00:05 <alyssa> time for a "fun" rebase

00:06 <alyssa> My approach is to reorder commits according to "readiness"

00:06 <alyssa> the stuff on top becomes upstream mesa

00:06 <alyssa> that + the next block becomes asahi/mesa

00:06 <alyssa> and everything below that remains in agx/next until it can be fixed

00:07 <alyssa> unfortunately most of agx/next is quite broken right now

00:07 <alyssa> and I am 96 commits on top of main

00:09 <alyssa> ugh

00:11 <alyssa> this is tricky

00:11 <alyssa> since this code is all so broken.

00:24 <alyssa> alternatively I can just release the mess, it is -edge after all ;)

00:28 <alyssa> I mean... most of the brokenness is MSAA, buffer textures, and PBOs

00:28 <alyssa> now that I've reordered, that stuff is at the end of the series

00:28 <alyssa> oh and XFB

00:28 <alyssa> XFB is panfrost's implementation so

00:28 <alyssa> totally broken but somehow passed GLES3.1 CTS multiple times

00:31 <alyssa> which I'll push out for testing here and then we'll find out how things are

00:48 <alyssa> it's kinda batshit that i wrote a gpu driver for this computer

01:34 <alyssa> t---

01:34 <alyssa> -------

01:34 <alyssa> Just pushed a new asahi/main branch

01:34 <alyssa> In addition to the usual rebase, there's a lot of good stuff here

01:34 <alyssa> Most of it is behind the scenes, cherry-picked from agx/next

01:35 <alyssa> NV_conditional_render is checkbox implemented

01:35 <alyssa> (meaning, don't use it, but it's there.)

01:36 <alyssa> Bug fix for discard + gl_FragDepth which I think can be hit in the wild for GL

01:36 <alyssa> GL2

01:36 <alyssa> Some work to reduce shader variant jank

01:36 <alyssa> Early stuff, but hopefully there's a bit less shader compilation jank

01:36 <alyssa> Possibly broken clip distances but shh

01:37 <alyssa> Optimizations around "wait" instructions that didn't help perf as much as I was hoping but also don't seem to break anything

01:37 <alyssa> That's about it for user visible changes

01:37 <alyssa> Though as I said there are a lot of changes "under the hood" that are preparing for bigger stuff

01:38 <alyssa> Oh! and multiple render targets are now advertised

01:38 <alyssa> Code was already there, a bug fix for that + flipping the switch

01:38 <alyssa> and framebuffer fetch too

01:38 <alyssa> Please test and report regressions here :)

01:39 <alyssa> marcan: hold off on pushing the package until some of the more adventurous users here (and lina) get a chance to try this out, since it's a pretty big bump in terms of # of commits

01:39 <alyssa> but hopefully later this week

01:39 <alyssa> I also need to open MRs for this stuff, blah

01:39 <alyssa> out of time for the night though

01:39 <alyssa> cheers :)

02:03 <chadmed> alyssa: no issues here (yet) bumming around at the desktop/doing stuff on the web

09:21 <lina> alyssa: Nice!! I'll test it over the next few days ^^

09:21 <lina> Sorry for being slow on the clang-format review... though it probably didn't really have much to review ^^;;

09:39 <lina> Why is a Chromium build pulling in VK-GL-CTS... This is so cursed...

10:04 <dottedmag> lina: ANGLE test suite?

12:08 <tly> hi :) im not quite sure if im in the right channel to ask... i tried to debug the asahi mesa driver on macos because glxinfo segfaults but ran into a very weird problem: the compiled driver does not contain any debug symbols, even in "buildtype=debug" mode. the .o files that are generated dont have any debug stuff in them

12:42 <marcan> tly: The macOS build is for reverse engineering mostly, it's not really expected to work ^^

12:43 <marcan> ask alyssa and lina about the current state, but at the very least it probably will only work on 12.3

12:44 <sven> was a bit bored and didn’t feel like thunderbolt so I looked a bit into nvme/FileVault. looks like there are aes key and IV fields for aes-xts in the tcb struct

12:44 <sven> I think you can set a flag somewhere and it’ll just take that key or set another flag and it’ll assume that key is “sep entangled”

12:45 <sven> and that entanglement seems to change between boots

12:45 <marcan> sven: wrong channel?

12:45 <sven> so essentially macOS asks sep to get the volume key, sep returns that key in an entangled form and just feeds that into ans

12:45 <sven> uh. Yes

12:45 <sven> sorry 😅

12:46 <marcan> alyssa: I'll make a build a bit later and push it to asahi-dev

12:46 <marcan> also it's okay if I start tagging timestamped releases, right? otherwise we lose prior state

12:48 <tly> marcan: thanks, i think it would be cool to support macos some time in the future to work around the opengl 4.1 limit

12:48 <marcan> it's not practical because it will break every macOS update and require reverse engineering the new ABI

12:50 <tly> ah yep, thats true, i didnt think about that...

12:59 <marcan> lina: oh god I just saw your tweet, please don't make me have to package CEF... *groan*

13:56 <alyssa> marcan: yes

21:18 <alyssa> oh I was wondering why my shiny new M1 was feeling so slow

21:18 <alyssa> was running softpipe

21:18 <alyssa> ;p

21:23 <bluetail8> but you are using the gpu driver now, right?