ChanServ changed the topic of #asahi-gpu to: Asahi Linux: porting Linux to Apple Silicon macs | GPU / 3D graphics stack black-box RE and development (NO binary reversing) | Keep things on topic | GitHub: https://alx.sh/g | Wiki: https://alx.sh/w | Logs: https://alx.sh/l/asahi-gpu
SSJ_GZ has quit [Ping timeout: 480 seconds]
millenialhacker has joined #asahi-gpu
millenialhacker has quit [Ping timeout: 480 seconds]
capta1nt0ad has quit [Remote host closed the connection]
<alyssa>
< Pass type: Translucent
<alyssa>
---
<alyssa>
> Pass type: unknown 3 (XXX)
<alyssa>
lina: >.>
<alyssa>
you were right to revert the pass type commit then
<jannau>
alyssa: that has the broken stride check in agx_resource_from_handle()
<jannau>
I think the patch works fine the Xorg problems I saw were unrelated
<alyssa>
oh right yes
<alyssa>
need to rebase asahi/main
scardracs has quit []
<alyssa>
will let my MRT stuff land first
<alyssa>
it's in CI now
<jannau>
dcp only is not bad on 5120x2160@60 + 1920x1080@60 (plasma/wayland). I think the broken vblank might help
<jannau>
it sees 2 * 60 vblanks per second (for dcp and dcpext) and thinks it runs at 120Hz
<jannau>
I guess that might be just the confused "show fps" effect
user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
user982492 has joined #asahi-gpu
user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
karolherbst has joined #asahi-gpu
<Dcow_>
jannau: what 5120x2160 monitor do you use?
Dcow_ has quit []
<yuka>
> Couldn't compile vertex program (Default, blit_screen) : 0:1(10): error: GLSL 1.50 is not supported. Supported versions are: 1.10, 1.20, and 1.00 ES
<alyssa>
every complaint that $FOO doesn't work delays $FOO working by a month
<alyssa>
as a general rule
<yuka>
I did not expect it to work, but I'm happy that I now know what the obstacles are
<yuka>
(and fixed some of them)
<yuka>
it came with a bundled jemalloc that did not like 16K page size
<alyssa>
(similarly every complaint that $FOO is slow with $COMMUNITY_BACKED_OPEN_SOURCE_GPU_DRIVER causes $FOO to be slow for another month on said driver)
<yuka>
so I replaced that
<jannau>
lol, drm_fb_dma_get_gem_addr() already adjusts for the x/y offset. that makes dealing with planes src rectangle a little annoying
<jannau>
I don't think we can't avoid converting it to int since we need to know width and height in case of scaling
<jannau>
s/can't/can/
<alyssa>
context?
<jannau>
X11 multi-display support
<jannau>
it allocates a single fb and one display is offsetted
<jannau>
dcp uses drm_fb_dma_get_gem_addr() to set the surf address for dcp and we set the src rectangle also based on the offset
<alyssa>
right, ok
<alyssa>
good luck :3
<jannau>
"req->swap.src_rect[l].x = req->swap.src_rect[l].y = 0;" works for me