#asahi-gpu · 2023-08-17 03:38
linaThat is totally untested, I just wrote it blind...#asahi-gpu · 2023-08-17 03:38
linaOh it might be that timeline stuff is just completely broken#asahi-gpu · 2023-08-17 03:38
i509vcbI guess this isn't tested so no idea what could be wrong there#asahi-gpu · 2023-08-17 03:38
i509vcbFlipping the bits in agxv to turn on timeline semaphores just seems to hang forever on dEQP-VK.synchronization.basic.timeline_semaphore.one_queue#asahi-gpu · 2023-08-17 03:37
lina(And then if a fence never gets signaled that can jam all of KDE)#asahi-gpu · 2023-08-17 03:37
linaThe driver is really good with memory safety stuff but there are still definitely bugs in the queuing logic/etc (plus drm_sched problems) that can cause things to get stuck I think...#asahi-gpu · 2023-08-17 03:36
linaThat sounds like a bug...#asahi-gpu · 2023-08-17 02:28
i509vcbThat seems weird#asahi-gpu · 2023-08-17 02:28
i509vcb> WorkQueue: Cannot submit, but queue is empty?#asahi-gpu · 2023-08-17 02:25
alyssaOof#asahi-gpu · 2023-08-17 02:25
i509vcbI noticed that spam when I ran dEQP-VK.synchronization.basic.binary_semaphore.chain which failed before a reboot and works after a reboot#asahi-gpu · 2023-08-17 02:24
alyssaapparently i have that too, that's neat#asahi-gpu · 2023-08-17 02:24
i509vcbWondering is there a reason why dmesg gets swamped with "Blocking due to compute queue full"?#asahi-gpu · 2023-08-17 02:15
i509vcbWayland compositor handover would be great for this type of situation...#asahi-gpu · 2023-08-17 02:13
i509vcbI think the compositor hung if I can't adjust volume, well I don't know the local IP off head so I guess I need to reboot anyways#asahi-gpu · 2023-08-17 02:11
alyssahaven't had to reboot in months, lina's kernel is rock solid <3#asahi-gpu · 2023-08-17 02:11
alyssaI regularly have to ssh in and pkill sway and that's enough to fix#asahi-gpu · 2023-08-17 02:11
alyssaIt might be#asahi-gpu · 2023-08-17 02:11
i509vcbI'm guessing this type of hang I can't fix from just trying to ssh in...#asahi-gpu · 2023-08-17 02:10
alyssawhen I was doing Mesa on macOS, i got macOS to fail in all sorts of creative ways..#asahi-gpu · 2023-08-17 02:10
alyssagreat#asahi-gpu · 2023-08-17 02:10
alyssayeah the process isolation on these things isn't.. grat#asahi-gpu · 2023-08-17 02:09
i509vcbDidn't expect running a semaphore test to do that#asahi-gpu · 2023-08-17 02:09
i509vcbWell I knew it would happen eventually that I'd cause kde to hang#asahi-gpu · 2023-08-17 02:06
i509vcbok#asahi-gpu · 2023-08-17 02:06
alyssaI'm told Collabora is working on finishing vk_meta_copy so I wasn't touching it#asahi-gpu · 2023-08-17 02:05
alyssashrug#asahi-gpu · 2023-08-17 02:04
i509vcbalyssa: looking at https://gitlab.freedesktop.org/asahi/mesa/-/blob/agxv/main/src/vulkan/runtime/vk_meta_copy.c#L558 multiple layers aren't handled yet, is this something more complicated than just run do_copy at different locations a few times or does some other part of agxv need to compensate for this?#asahi-gpu · 2023-08-16 23:09
i509vcbagxv_physical_device_try_create really makes me yearn for Rust's Result#asahi-gpu · 2023-08-16 22:07
alyssaella-0: ack#asahi-gpu · 2023-08-16 22:07
ella-0alyssa: I'm running my own kernels but go ahead and rebase#asahi-gpu · 2023-08-16 22:02
alyssauhoh#asahi-gpu · 2023-08-16 20:58
i509vcbthe fun of dEQP-VK.api.device_init.create_instance_device_intentional_alloc_fail.basic, have to effectively rewrite agxv_physical_device_try_create to make the solution bearable#asahi-gpu · 2023-08-16 16:17
i509vcbseems like what v3dv is doing is fine given that the spec seems to says it's undefined behavior to pass an invalid fd#asahi-gpu · 2023-08-16 15:43
i509vcb(radv and nvk act correctly here since a prime fd could be from one of a few memory types and the memory type bits need to be set accordingly)#asahi-gpu · 2023-08-16 15:42
i509vcbSo the question is, which is right#asahi-gpu · 2023-08-16 15:42
i509vcbThe spec doesn't actually explicitly allow or forbid this#asahi-gpu · 2023-08-16 15:42
i509vcbHowever here's the kicker#asahi-gpu · 2023-08-16 15:42
i509vcbradv and nvk will return an error correctly#asahi-gpu · 2023-08-16 15:42
i509vcbthat little pseudocode snippet should be returning VK_ERROR_INVALID_EXTERNAL_HANDLE since a file is definitely not a prime fd#asahi-gpu · 2023-08-16 15:41
i509vcbwhich seems bizzare since those cannot be prime fds which are what vulkan drivers typically consume with the dma buf handle type#asahi-gpu · 2023-08-16 15:41
i509vcbvkGetMemoryFdPropertiesKHR(device, fd)#asahi-gpu · 2023-08-16 15:41
i509vcbfd = open("~/Pictures/ferris.txt")#asahi-gpu · 2023-08-16 15:40
i509vcbAlso this means you can do something like#asahi-gpu · 2023-08-16 15:40
i509vcbYou could pass -1 (invalid fd) and it technically is ok with that function#asahi-gpu · 2023-08-16 15:40
i509vcbhowever a lot of implementations (v3dv, tu, pvr, anv, hasvk) return VK_SUCCESS regardless of the file descriptor you pass into that function#asahi-gpu · 2023-08-16 15:39
i509vcbSo vkGetMemoryFdPropertiesKHR is used primarily used for determining the memory type of a dmabuf you are importing. #asahi-gpu · 2023-08-16 15:38
alyssaI'll bump asahi/main today, and if all goes well, we'll push out the new kernel to arch/fedora users tomorrow(?) and then once the new kernel is pushed out I'll rebase AGXV#asahi-gpu · 2023-08-16 15:38
i509vcbBefore I go over to #dri-devel to try to figure out which implementations are wrong, I'll try to explain what I think is wrong with v3dv and co in that function#asahi-gpu · 2023-08-16 15:37
alyssaok, that's fine