01:37 fdobridge_: <S​id> bind
02:31 fdobridge_: <S​id> @karolherbst I did a thing that solved my problem while continuing to let me use hdmi audio
02:31 fdobridge_: <S​id> ```
02:31 fdobridge_: <S​id> # Unbind audio controller to allow runpm to work.
02:31 fdobridge_: <S​id> SUBSYSTEM=="pci", DRIVER=="snd_hda_intel", ATTR{vendor}=="0x10de", ATTR{device}=="0x1aeb", RUN+="/bin/sh -c 'echo 0000:01:00.1 > /sys/bus/pci/drivers/snd_hda_intel/unbind'"
02:31 fdobridge_: <S​id>
02:31 fdobridge_: <S​id> # Bind audio controller on HDMI connection to have working HDMI audio
02:31 fdobridge_: <S​id> ACTION=="change", SUBSYSTEM=="drm", RUN+="/bin/sh -c 'echo 0000:01:00.1 > /sys/bus/pci/drivers/snd_hda_intel/bind'"```
02:32 fdobridge_: <S​id> ...wait am dumb
02:32 fdobridge_: <S​id> first rule is run every time second is run too
02:32 fdobridge_: <S​id> hang on
02:39 fdobridge_: <S​id> I know I could handle this with a script but I wanna keep it pure udev 😅
02:57 fdobridge_: <a​irlied> @gfxstrand just fyi the push sizing in begin rendering is undersized
02:57 fdobridge_: <a​irlied> not sure if that is just on your mod branch or not
02:58 fdobridge_: <g​fxstrand> 😭
02:59 fdobridge_: <g​fxstrand> That might just be the modifier branch. There were merge conflicts with the sizing and I haven't dug that far into it.
04:44 HdkR: /6/4
09:06 fdobridge_: <m​arysaka> I have a patch for it on the mesh shader MR too
14:19 fdobridge_: <d​adschoorse> does nv not have vec2 fmin/fmax?
14:27 fdobridge_: <m​arysaka> I know that it exists on Ampere+ as `HMNMX2`, but couldn't get it to be generated for some reasons
14:27 fdobridge_: <m​arysaka> probably will have to fuzz nvdisasm to find the opcode at some point
14:29 fdobridge_: <m​arysaka> There is also F2FP.PACK_AB for fp32x2 -> fp16x2 but haven't digged too much into that
14:36 fdobridge_: <m​henning> Looks like the opcode is
14:36 fdobridge_: <m​henning> 0x240 hmnmx2__RRR
14:36 fdobridge_: <m​henning> 0x840 hmnmx2__RIR
14:36 fdobridge_: <m​henning> 0xa40 hmnmx2__RCR
14:36 fdobridge_: <m​henning> From my RE work here https://gitlab.freedesktop.org/mhenning/re/-/blob/main/opclass/opclass86?ref_type=heads
14:36 fdobridge_: <m​arysaka> oh right you did that I totally forgot!
14:36 fdobridge_: <m​henning> Looks like the opcode is
14:36 fdobridge_: <m​henning> `0x240 hmnmx2__RRR`
14:36 fdobridge_: <m​henning> `0x840 hmnmx2__RIR`
14:36 fdobridge_: <m​henning> `0xa40 hmnmx2__RCR` (edited)
14:38 fdobridge_: <m​arysaka> ``0x832 hmul2__RI`` is also a thing, I guess I will fix HMUL2 immediate form tomorrow then
14:38 fdobridge_: <m​arysaka> ``0x832 hmul2__RI`` is also a thing, I guess I will fix HMUL2 immediate form tomorrow then, it was failing quite badly for me (edited)
19:15 Lyude: I -think- I figured out the types I need for exposing all of the various modesetting objects :), it seems my first step is going to be writing up ww lock wrappers and then come up with a new variant of LockedBy<> for fallible locking schemes