ChanServ changed the topic of #asahi-gpu to: Asahi Linux GPU development (no user support, NO binary reversing) | Keep things on topic | GitHub: https://alx.sh/g | Wiki: https://alx.sh/w | Logs: https://alx.sh/l/asahi-gpu
<i509vcb> I think I got the rebase done, just going through each commit and fixing compiler errors I see along the way
<i509vcb> although that's quite involved
alexaltea_ has joined #asahi-gpu
alexaltea has quit [Ping timeout: 480 seconds]
chergert has joined #asahi-gpu
<i509vcb> hmm what is TEXTURE_TYPE_XR? Not related to the rebase, just don't know
aratuk has joined #asahi-gpu
<i509vcb> I assume it's not VR related?
<aratuk> Congrats on 4.6 conformance! @alyssa, in your post it seems as though you might be conspicuously avoiding the name of "the vendor". If you don't mind my asking, did Apple ask you not to use their trademark even in describing your work, or something along those lines?
<alyssa> aratuk: Thanks, and no
<alyssa> i509vcb: imho just squash agxv
<alyssa> the history isnt' worth preserving at this point
<alyssa> vs the rebase hell
<i509vcb> Okay if you are fine with the squashing
<alyssa> I am yeah
chergert has quit [Quit: chergert]
<alyssa> I don't think preserving history is reasonable at this stage, tbh.
<alyssa> given the XML is changing externally, etc
aratuk has quit []
wrender has joined #asahi-gpu
ju has quit [Read error: Connection reset by peer]
ju has joined #asahi-gpu
<i509vcb> rebase is done
<i509vcb> let me smoke test vkcube quickly
wrender has quit [Ping timeout: 480 seconds]
<i509vcb> vkcube seems to die in the compiler
kidplayer666 has quit [Quit: Connection closed for inactivity]
KxCORP has quit [Quit: Bye!]
KxCORP has joined #asahi-gpu
<i509vcb> mismatched ppp size?
Bertrand___ has quit [Ping timeout: 480 seconds]
Bertrand___ has joined #asahi-gpu
wrender has joined #asahi-gpu
fossdd_ has quit [Ping timeout: 480 seconds]
fossdd has joined #asahi-gpu
<alyssa> i509vcb: new_ppp_update takes a # of control words, needs to be high enough
jeisom has quit [Ping timeout: 480 seconds]
<i509vcb> Clearly I'm not seeing the error, the culprit is probably somewhere here: https://gitlab.freedesktop.org/i509VCB/mesa/-/blob/rebase-10010/src/asahi/vulkan/agxv_cmd_draw.c?ref_type=heads#L657-884
<i509vcb> 4 bytes short from gdb
<i509vcb> (not shown in the branch, setting viewport_count = 1)
pthariensflame has joined #asahi-gpu
pthariensflame has quit [Quit: Textual IRC Client: www.textualapp.com]
Bertrand___ has quit [Read error: Connection reset by peer]
Bertrand___ has joined #asahi-gpu
marvin24 has joined #asahi-gpu
marvin24_ has quit [Ping timeout: 480 seconds]
tertl8 has quit [Quit: Connection closed for inactivity]
wrender has quit [Ping timeout: 480 seconds]
kidplayer666 has joined #asahi-gpu
ciara has quit [Remote host closed the connection]
ciara has joined #asahi-gpu
Bertrand___ has quit [Ping timeout: 480 seconds]
Bertrand___ has joined #asahi-gpu
possiblemeatball has quit [Quit: Quit]
jcajka has joined #asahi-gpu
mkurz_ has quit [Ping timeout: 480 seconds]
jeisom has joined #asahi-gpu
adewart has joined #asahi-gpu
jeisom has quit [Ping timeout: 480 seconds]
<alyssa> i509vcb: VIEWPORT now also has a VIEWPORT_CONTROL word, see agx_state.c (and blame if you want to know who to blame)
kidplayer666 has quit [Quit: Connection closed for inactivity]
<i509vcb> okay so now just gray output in vkcube
<i509vcb> ah no wrong order of ppp fields
kidplayer666 has joined #asahi-gpu
<i509vcb> vkcube works now
<i509vcb> I'll squash the two extra commits in like an hour
<alyssa> woo
wrender has joined #asahi-gpu
wrender has quit [Ping timeout: 480 seconds]
jeisom has joined #asahi-gpu
<i509vcb> alyssa: rebased agxv is on my rebase-10010 branch
<alyssa> :)
<alyssa> i509vcb: lmk if you want me to push that
<i509vcb> fine to push that
<alyssa> ok, will do in a bit
Guest2526 has quit [Quit: Bridge terminating on SIGTERM]
rhysmdnz has quit [Quit: Bridge terminating on SIGTERM]
rhysmdnz has joined #asahi-gpu
Jamie has joined #asahi-gpu
Jamie is now known as Guest2890
wrender has joined #asahi-gpu
wrender has quit [Ping timeout: 480 seconds]
possiblemeatball has joined #asahi-gpu
anders__2 has quit [Ping timeout: 480 seconds]
kidplayer666 has quit [Quit: Connection closed for inactivity]
possiblemeatball has quit [Quit: Quit]
chergert has joined #asahi-gpu
Bertrand___ has quit [Ping timeout: 480 seconds]
Bertrand___ has joined #asahi-gpu
possiblemeatball has joined #asahi-gpu
possiblemeatball has quit []
jcajka has quit [Remote host closed the connection]
Bertrand___ has quit [Ping timeout: 480 seconds]
wrender has joined #asahi-gpu
anders2 has joined #asahi-gpu
wrender has quit [Ping timeout: 480 seconds]
chergert has quit [Quit: chergert]
wrender has joined #asahi-gpu
cyrinux has quit []
cyrinux has joined #asahi-gpu
wrender has quit [Ping timeout: 480 seconds]
anders_2 has joined #asahi-gpu
chergert has joined #asahi-gpu
anders2 has quit [Ping timeout: 480 seconds]
Dementor99 has joined #asahi-gpu
Dementor9 has quit [Ping timeout: 480 seconds]
wrender has joined #asahi-gpu
Dementor99 has quit [Ping timeout: 480 seconds]
Dementor9 has joined #asahi-gpu
chergert has quit [Quit: chergert]
possiblemeatball has joined #asahi-gpu
mkurz_ has joined #asahi-gpu
john-cabaj has quit [Remote host closed the connection]
john-cabaj has joined #asahi-gpu
darkapex has quit [Remote host closed the connection]
darkapex has joined #asahi-gpu
ciara has quit [Ping timeout: 480 seconds]
<i509vcb> Not sure where I'd get the values to upload in sysvals to give load_attrib_clamp_agx something to load other than 0
<i509vcb> agxv_CmdBindVertexBuffers2 seems like where I'd do that?
chergert has joined #asahi-gpu
<alyssa> i509vcb: ultimately vkCmdBindVertexBuffers2, but it should be in the vk_graphics_state
<alyssa> but agxv doesn't advertise robustness yet so you can just hack that out
<alyssa> (or set the clamp to infinity or whatever)
<i509vcb> Hmm setting clamp to infinity I guess shows more problems still exist in agxv
wrender has quit [Ping timeout: 481 seconds]
wrender has joined #asahi-gpu
<i509vcb> https://gist.github.com/i509VCB/6c8e60fe7cdd996f6642469beee3248e triangle example in sascha willem's examples, did set clamp to infinity and output is just black
<i509vcb> I have no idea what I should be looking for in agx decode dumps
<alyssa> well I'm first going to look at the VS because we're talking about it
<alyssa> oh uh can you build with disasembler
<alyssa> cherrypick "HACK: Wire dougallj's disassembler into wrap+compiler" from my next branch and setup applegpu
<alyssa> Depth bias/scissor
<alyssa> Scissor: 454
<alyssa> do you really upload 455+ scissors?
<i509vcb> no...
<i509vcb> unless the example is doing that for some reason
<alyssa> worth looking into
wrender has quit [Remote host closed the connection]
wrender has joined #asahi-gpu
<i509vcb> Scissor count seems to increase every few draws
<i509vcb> yup that's a memory leak, the number of scissors is never trimmed and it's in a dynarray
<i509vcb> same as depth bias
<alyssa> first vertex shader instruction is the clamp:
<alyssa> 0: 12014a421a4a403a icmpsel ult, r0, r5, u18, r5, u18
<alyssa> from the preamble, u18 is equal to
<alyssa> 584: 0501200d08c03200 device_load 0, i32, xy, r0_r1, u0_u1, 130, signed
wrender has quit [Ping timeout: 480 seconds]
<alyssa> which is plausible
<alyssa> looks plausible, idk
<alyssa> try stuff? or throw me a branch
<i509vcb> telling lowering to output nir_imm_int(b, UINT32_MAX) just results in black, so it's probably not that
<i509vcb> (for the clamp value)
wrender has joined #asahi-gpu
<chergert> i notice this assertion when testing the new GTK renderer for GNOME 46 and the tab overview used in various apps
<chergert> ../src/mesa/state_tracker/st_gen_mipmap.c:114: st_generate_mipmap: Assertion `pt->last_level >= lastLevel' failed.
<chergert> does that sound like a driver bug or an app/library bug?
* chergert is building mesa here so can definitely try different things to test stuff
zzywysm has quit [Quit: Textual IRC Client: www.textualapp.com]
zzywysm has joined #asahi-gpu
wrender has quit [Ping timeout: 480 seconds]