03:08 jenatali: karolherbst: What is it this time?
07:14 karolherbst: jenatali: "volatile global uchar4* p_var = &a_var[1];" becomes OpSpecConstantOp %_ptr_CrossWorkgroup_uchar PtrAccessChain %a_var %ulong_4
07:14 karolherbst: soo.. &a_var[4]; instead
07:15 karolherbst: not sure if it's broken in LLVM IR or in the translator, but looks like something up with vectors
07:15 karolherbst: but anyway... I have some of the tests passing now :)
07:15 karolherbst: just all vectors and non char tests aren't
07:20 karolherbst: *sigh*
07:20 karolherbst: "@p_var = addrspace(1) global ptr addrspace(1) getelementptr (i8, ptr addrspace(1) @a_var, i64 2), align 8"
07:21 karolherbst: opaque pointer strikes again
07:22 karolherbst: fixed by a7a36de0c88622bf810a021816e82c4a673174fc in the translator...
07:22 karolherbst: probably
07:23 karolherbst: not backported :')
07:36 karolherbst: mhh.. I think there is more to this...
18:45 Lyude: hm, I notice that drm_atomic_get_existing_plane_state() doesn't actually seem to get a pointer to the atomic state from __drm_plane_states.old_state or .new_state, but .state. What exactly is/was .state used for?
18:45 Lyude: (wondering since vkms uses this function in a few spots)
20:48 vsyrjala: please don't use drm_atomic_get_existing_plane_state(). we need to nuke it. one should only use the get_{new,old} stuff
20:49 vsyrjala: i wrote some preliminary patches to clean up a bunch of code in random drivers around this stuff, but haven't had time to polish/post them
20:52 sad_plan: hello. archive.mesa3d.org returns a duplicate end-entity certificate when using bearssl to fetch a tarball. would this be something someone here would be able to look into, or should we look someplace else?
20:53 sad_plan: it was reported back in 2020 aswell https://lists.freedesktop.org/archives/sitewranglers/2020-January/012973.html
20:59 dwfreed: probably incorrect apache configuration
21:17 sad_plan: I dunno, I just know someone needs to fix it. I just gotta get that infomation to said individual :p
21:23 alanc: I think the right people are on #freedesktop but I don't know how many are around on a holiday weekend
21:26 sad_plan: ill ask over the weekend. thanks
23:33 kode54: is there an equivalent of RADV_DEBUG=zerovram for other GPUs?