stikonas has quit [Remote host closed the connection]
kaspter has joined #panfrost
karolherbst has quit [Read error: Connection reset by peer]
<alyssa>
icecream95: Oh yeah. Forgot about that bug.
* alyssa
whistles
<HdkR>
Yes...Never reported ever
<HdkR>
:)
<alyssa>
Well. No point in hiding AFBC support if it's already broken anyway.
<alyssa>
And should seriously figure out who to poke to get that fix landed and backported :|
<alyssa>
I don't think Mesa has a reasonable way to workaround
camus has joined #panfrost
kaspter has quit [Ping timeout: 272 seconds]
camus is now known as kaspter
jernej has quit [Ping timeout: 272 seconds]
tgall_foo has joined #panfrost
davidlt has joined #panfrost
chrisf has quit [Remote host closed the connection]
chrisf has joined #panfrost
<daniels>
alyssa: yeah I know, there seemed to be no movement towards root-causing it or getting a fix applied, so I ordered a RockPi to look at it myself
<icecream95>
On the topic of not yet reported problems, there are a number of issues with checksumming that I really should fix before Mesa 21.1…
karolherbst has joined #panfrost
paulk-leonov has joined #panfrost
paulk-leonov has quit [Remote host closed the connection]
paulk-leonov has joined #panfrost
paulk-leonov has quit [Remote host closed the connection]
paulk-leonov has joined #panfrost
alpernebbi has joined #panfrost
raster has joined #panfrost
stikonas has joined #panfrost
davidlt has quit [Ping timeout: 240 seconds]
nlhowell has joined #panfrost
camus has joined #panfrost
kaspter has quit [Ping timeout: 240 seconds]
camus is now known as kaspter
camus has joined #panfrost
kaspter has quit [Ping timeout: 240 seconds]
camus is now known as kaspter
nlhowell has quit [Quit: WeeChat 3.0]
nlhowell has joined #panfrost
davidlt has joined #panfrost
nlhowell has quit [Remote host closed the connection]
nlhowell has joined #panfrost
archetech has joined #panfrost
hexdump0815 has quit [Quit: Connection closed]
nlhowell has quit [Remote host closed the connection]
nlhowell has joined #panfrost
stikonas has quit [Remote host closed the connection]
stikonas has joined #panfrost
camus has joined #panfrost
kaspter has quit [Ping timeout: 240 seconds]
camus is now known as kaspter
<alyssa>
daniels: It's the YTR issue I already reported and sent a fix for, it just never got landed and I don't know how to kernel
stikonas_ has joined #panfrost
stikonas has quit [Ping timeout: 272 seconds]
alpernebbi has quit [Quit: alpernebbi]
ggardet has joined #panfrost
guillaume_g has quit [Ping timeout: 272 seconds]
stikonas_ is now known as stikonas
robmur01_ is now known as robmur01
<daniels>
alyssa: well we can YOLO it in, but it's just sat there for so long with no-one ever seeming quite sure which part was correct (who's advertising the wrong modifier) that you can argue it's now ABI, and maybe the fix is adding an XOR flag
<daniels>
or just deciding that rockchip-drm defines the ABI by virtue of having actually shipped with it on by default
<robmur01>
Is there a precedent set by other consumers? If I'm reading right, at least meson and mali-dp seem to both require and explicitly advertise the modifier, which points at rockchip being plain inconsistent and wrong
<macc24>
is mali-dp even used anywhere?
<robmur01>
yes
<narmstrong>
I need to try the same setup on the Amlogic GXM & G12A/G12B to see if we have the same regression with afbc, but I suppose @chewitt would have complained
<narmstrong>
on amlogic G12A, YTR is optional and can be disabled, and on GXM it's mandatory like for RK
<alyssa>
daniels: The rockchip code is obejctively wrong
<alyssa>
The fix is objectively correct
<alyssa>
That it's sat there means *nobody has ever used the upstream rockchip code*
<alyssa>
Rockchip needs to be fixed, full stop.
<alyssa>
There's no ABI to break when nothing could ever consume it in the ifrst place.
<chewitt>
I'm not aware of any AFBC issues, but then I also wouldn't know how they'd manifest themselves
<daniels>
heh, it looks like panfrost is also objectively wrong and will break when YTR is flipped on
<daniels>
everyone's a winner
<chewitt>
or how to check if AFBC is being used
<chewitt>
since RPi4 got HDR working I've mostly being using that not my half-working Amlogic boards
<alyssa>
daniels: what's the panfrost bug?
<daniels>
alyssa: YTR is only ever set on the RT side, not on tex
<robmur01>
interesting... I can repro the psychedelic-tiles behaviour with Sway if I switch inputs on my monitor (or physically unplug/replug), but then if I switch VTs and back it comes good again
<alyssa>
daniels: So everyone is on the same page that it's just Rockchip that's horribly broken? ;)
<daniels>
trying to confirm
<alyssa>
glad someone on the team understands modifiers ;)
<daniels>
VOPs in later gens have an explicit YTR flag in the HW reg which is set depending on the value of the flag in the modifier
<daniels>
earlier gens appear to have no explicit flag and also do not explicitly set it in the modifier
<alyssa>
3399 is earlier gen?
<daniels>
yeah
ggardet is now known as guillaume_g
karolherbst has quit [Ping timeout: 240 seconds]
<robmur01>
earliest, even (as far as AFBC goes) :)