Where communities thrive

Join over 1.5M+ people
Join over 100K+ communities
Free without limits
Create your own community

Explore more communities

MaikKlein/ash

People

Repo info

Activity

matrixbot

@matrixbot

Ralith that would be very welcome!

Ralith the windows SDK installer has had bugs at various points

Brian Merchant

@bzm3r

ahh i see. you know what, i'm going to ask in that issue if people are still able to reproduce the problem

Brian Merchant

@bzm3r

@Ralith So, I tried to comment out some of the lines destroying objects in the Drop impl of ExampleBase, and noticed that the validation layer doesn't complain. Something similar is happening with the example here: https://github.com/linebender/piet-gpu/tree/master/piet-gpu-hal

Do you know why this might be the case?

matrixbot

@matrixbot

Ralith have you verified that it's producing output at all?

matrixbot

@matrixbot

Ralith I see output when I run the examples, and errors when I remove required destroy calls from the example

Brian Merchant

@bzm3r

@Ralith yes, I'm certain that output is being produced, since when I run the triangle example, I do get errors like this:

ERROR:
VALIDATION [VUID-vkQueuePresentKHR-pWaitSemaphores-03268 (622825338)] : Validation Error: [ VUID-vkQueuePresentKHR-pWaitSemaphores-03268 ] Object 0: handle = 0x23ade4f5df8, type = VK_OBJECT_TYPE_QUEUE; Object 1: handle = 0xedbd50000000010, type = VK_OBJECT_TYPE_SEMAPHORE; | MessageID = 0x251f8f7a | VkQueue 0x23ade4f5df8[] is waiting on VkSemaphore 0xedbd50000000010[] that has no way to be signaled. The Vulkan spec states: All elements of the pWaitSemaphores member of pPresentInfo must reference a semaphore signal operation that has been submitted for execution and any semaphore signal operations on which it depends (if any) must have also been submitted for execution. (https://www.khronos.org/registry/vulkan/specs/1.1-extensions/html/vkspec.html#VUID-vkQueuePresentKHR-pWaitSemaphores-03268)

this is how i modified the Drop impl: https://gist.github.com/bzm3r/fb299d9ff7a890b2d1508485b1cb6c6c

but no cigars in terms of seeing errors related to not having destroyed the device, surface, or debug utils loader

matrixbot

@matrixbot

Ralith @bmer you commented out the calls that would prompt it to report an error

Ralith there's no way for the validation layers to check whether you've cleaned things up on process exit because they don't hook non-vulkan APIs

Brian Merchant

@bzm3r

@Ralith sorry, can you explain further?

matrixbot

@matrixbot

Ralith @bmer the validation layers will never produce errors based on you not destroying the instance

Ralith they produce errors when you try to destroy the instance without first destroying everything else

Ralith you need to comment out a leading portion of the destroy calls; commenting out a trailing portion just means you're leaking resources, which is permitted

Brian Merchant

@bzm3r

@Ralith so one way to check to make sure I'm not leaking resources is to always at the very least, try to destroy the instance?

(I'd really like to use the validation layers to check if I am leaking resources.)

matrixbot

@matrixbot

Ralith yes, you should always destroy your instance

Brian Merchant

@bzm3r

ahh, and I think I now understand what you mean by: > there's no way for the validation layers to check whether you've cleaned things up on process exit because they don't hook non-vulkan APIs

@Ralith thank you very much, as always :)

matrixbot

@matrixbot

Ralith np

Austin Shafer

@amshafer

I'm looking at adding bindings for importing external memory, is there an example/docs of adding new extension support?

i.e. how to use the generator

Surely it's not as simple as just changing the source path in generator.rs

Austin Shafer

@amshafer

Ah, I'm an idiot. It's all scanned by vk.xml

Maik Klein

@MaikKlein

fwiw I just started to rewrite the generator. I'll push a wip draft later if you want to follow the progress. The current generator has become unmaintainable.

Austin Shafer

@amshafer

I don't actually need to modify the generator anymore. Where will the wip one be? I'm sure I'll lazily follow the contribution activity/implementation notes so I can learn how the new one works

Maik Klein

@MaikKlein

https://github.com/MaikKlein/ash/compare/wip-generator very wip atm

Antonino Siena

@nifker

May I put general questions about Vulkan here?

matrixbot

@matrixbot

Ralith I don't see why not

Antonino Siena

@nifker

Im(rather new to Vulkan) trying to get some basic stuff up but I encounter a segfault when calling VkCmdBeginRenderPass.
https://paste.gg/p/anonymous/224d26dea3de41ff937fc3006ce6bd15

matrixbot

@matrixbot

Ralith do you have the validation layers enabled?

Antonino Siena

@nifker

Oh not yet

matrixbot

@matrixbot

Ralith always a good first step

Ralith if you're on linux, another good trick is to build your video drivers with debug symbols; then the backtraces tend to be fairly self explanatory

Antonino Siena

@nifker

I have the debug symbols and I can see now that it has to do with the renderpass attachments

Antonino Siena

@nifker

it also seems the validation layers doesnt print anything to stdout

Or do I necessarily need VkDebugUtilsMessengerCreateInfoEXT?

matrixbot

@matrixbot

Ralith and specify what info you're interested in

matrixbot

@matrixbot

Ralith then you don't have the validation layers enabled

Antonino Siena

@nifker

Ok now I got it but the only error I get is this one "libSPIRV-Tools-opt.so: cannot open shared object file: No such file or directory"

not sure which library I am missing but I cant find any in my repo such as that

Antonino Siena

@nifker

Ok Ive found out my errors by the might of the debug output ^^

Antonino Siena

@nifker

I am getting this error: https://github.com/KhronosGroup/Vulkan-Docs/blob/f6a10a5af095938e2a6ac8581b7f848a4b324f46/chapters/VK_KHR_swapchain/wsi.txt#L1291
even though I created the framebuffers the same way the triangle example did - any ideas?

matrixbot

@matrixbot

Ralith that error says they need to be in a specific layout

Ralith are they in that layout?

Antonino Siena

@nifker

They should be - I specified it in the renderpass attachments - or do I need to specify that somewhere else?

Antonino Siena

@nifker

Ok I just forgot a queue_submit call :P

matrixbot

@matrixbot

Ralith those help