types/wlr_linux_dmabuf_v1.c:526: feedback_compile: Assertion `table_len > 0' failed. #1218
Comments
|
I'm getting the same error with polaris card in nested mode, but embedded works fine. |
sharkautarch
added a commit
to sharkautarch/gamescope
that referenced
this issue
Mar 27, 2024
sharkautarch
added a commit
to sharkautarch/gamescope
that referenced
this issue
Mar 27, 2024
Haven't tested this (besides making sure it compiles w/o errors), but see if this branch/fork I made works: (clone from my fork, and then do |
I've build it but it has error connecting to wayland: `No CAP_SYS_NICE, falling back to regular-priority compute and threads. Performance will be affected. xdg_backend: libdecor: GTK cannot connect to Wayland compositor` In DMESG: |
sharkautarch
added a commit
to sharkautarch/gamescope
that referenced
this issue
Mar 27, 2024
Ok, I know there was another issue related to gamescope wayland backend having problems w/ libdecor |
|
now testing it with vkcube I'm gettiing the same error as on generic one (I've checked branch and its correct) ./gamescope -- vkcube |
¯_(ツ)_/¯ |
Tried your branch |
|
Your GPU is too old to support modifiers. I need to add a non-modifier fallback. |
|
I am having the same issue with the 3.14.3 version and the latest git:
|
|
Downgrading to 3.14.2 for now. To help people searching to find this issue: 3.14.2: after 82e3632 : |
This was referenced May 6, 2024
|
Does latest git work for you? 4ca3986 should hopefully fix it. |
|
I cloned the project, did the checkout of that commit and built the binary, but I still get this: |
That commit works for me: |
|
I'll attach the entire terminal log, in case I got something wrong during compile or during the switch to that 4ca3986: |
Thanks! Nested works on my old rx560, tried embedded - didn't work |
Installed gamescope from AUR via the package Still getting this with While running it via TTY bash login shell with Downgrading back to 3.14.2 works for me without error. |
Funny enough, even eliminating 82e3632 change completely does not fix the issue. I.e. applying diff --git a/src/wayland_backend.cpp b/src/wayland_backend.cpp
index c5b33cf..2fcb1ff 100644
--- a/src/wayland_backend.cpp
+++ b/src/wayland_backend.cpp
@@ -1273,22 +1273,8 @@ namespace gamescope
void CWaylandBackend::GetPreferredOutputFormat( VkFormat *pPrimaryPlaneFormat, VkFormat *pOverlayPlaneFormat ) const
{
- VkFormat u8BitFormat = VK_FORMAT_UNDEFINED;
- if ( SupportsFormat( DRM_FORMAT_ARGB8888 ) )
- u8BitFormat = VK_FORMAT_B8G8R8A8_UNORM;
- else if ( SupportsFormat( DRM_FORMAT_ABGR8888 ) )
- u8BitFormat = VK_FORMAT_R8G8B8A8_UNORM;
-
- VkFormat u10BitFormat = VK_FORMAT_UNDEFINED;
- if ( SupportsFormat( DRM_FORMAT_ABGR2101010 ) )
- u10BitFormat = VK_FORMAT_A2B10G10R10_UNORM_PACK32;
- else if ( SupportsFormat( DRM_FORMAT_ARGB2101010 ) )
- u10BitFormat = VK_FORMAT_A2R10G10B10_UNORM_PACK32;
-
- assert( u8BitFormat != VK_FORMAT_UNDEFINED );
-
- *pPrimaryPlaneFormat = u10BitFormat != VK_FORMAT_UNDEFINED ? u10BitFormat : u8BitFormat;
- *pOverlayPlaneFormat = u8BitFormat;
+ *pPrimaryPlaneFormat = VK_FORMAT_A2B10G10R10_UNORM_PACK32;
+ *pOverlayPlaneFormat = VK_FORMAT_B8G8R8A8_UNORM;
}
bool CWaylandBackend::ValidPhysicalDevice( VkPhysicalDevice pVkPhysicalDevice ) const
Still produces the same result for me |
|
Well, ditching the Wayland backed and switching back to SDL backend (as it was in 3.14.2) works for me. diff --git a/src/main.cpp b/src/main.cpp
index 88c4c7c..972b2ab 100644
--- a/src/main.cpp
+++ b/src/main.cpp
@@ -762,7 +762,7 @@ int main(int argc, char **argv)
if ( eCurrentBackend == gamescope::GamescopeBackend::Auto )
{
if ( g_pOriginalWaylandDisplay != NULL )
- eCurrentBackend = gamescope::GamescopeBackend::Wayland;
+ eCurrentBackend = gamescope::GamescopeBackend::SDL;
else if ( g_pOriginalDisplay != NULL )
eCurrentBackend = gamescope::GamescopeBackend::SDL;
else
Although, it's a dirty solution. Users should not need to patch the source to do that. We need a CLI option to explicitly set the backend. Lines 6 to 14 in 932a7bb Presumably, also deprecate existing |
Worked for me too, here is the output of `gamescope vkcube' after using your patch on the commit 932a7bb: |
|
I made a PR with new CLI option. #1321 |
|
using --backend=sdl works for me |
|
this is fixed for me too when explicitly setting sdl, it would be nice if backend=auto would detect the card and set it automatically to sdl with these cards |
Something like this, and/or maybe also detecting such an issue and falling back to SDL instead of throwing a segfault. Is that possible? |
This was referenced Aug 14, 2024
korewaChino
added a commit
to korewaChino/gamescope
that referenced
this issue
Aug 20, 2024
…to` is used. This works around ValveSoftware#1218 by making use of the new backend option added in ValveSoftware#1321, but adds a check to automatically fall back to the SDL backend if the current GPU does not support Vulkan DRM modifiers.
korewaChino
added a commit
to korewaChino/gamescope
that referenced
this issue
Aug 20, 2024
…to` is used. This works around ValveSoftware#1218 by making use of the new backend option added in ValveSoftware#1321, but adds a check to automatically fall back to the SDL backend if the current GPU does not support Vulkan DRM modifiers.
|
are there plans to get the wayland-backend (or drm) on polaris-gpus (without modifiers and without doing the workaround with the sdl-backend)? |
|
I do not have a polaris GPU to look into this with. I have tested on other devices without modifiers and forced them off and it seems to work fine, so I don't know what the issue is. The only thing I can think is that Other than that, it really just needs someone with a Polaris GPU to go and actually debug the issue. |
Sure. What exactly we need to debug? |
|
Had a look at this and it seems like forcing Seems like This makes this check return from the function, and Running this on kwin if that makes any diffrence. Test log |
|
What hardware are you using? And can you send the output of drm_info? |
|
I pushed some stuff to master that might help, can you give it a try? |
Wayland backend works now! But SDL backend crashes instead lol. |
|
Tried the changes, and seems to work pefectly for Wayland backend. As mentioned above it doesnt seem to work for sdl. I believe the reason for this is other backends return an empty span for supported modifiers while the new changes in backend.h checks if it contains invalid, without a check for Hardware infoCPU: amd 3700x I have 3 screens connected and running, and a fourth screen with a forced EDID through kernel parameter which is turned off. drm_infoBackend wayland logBackend SDL log |
|
Mh I'm now having that error, which I didn't before. I've installed gamescope with the gamescope-git AUR package (current version is from the latest commit to this date, d87f6d9). I'm not on Wayland.
damn 🤣 edit: the 3 recent commits didn't change a thing you can build from 7a27374, last commit that works on X, which broke with 0110109 |
|
@awsms @misyltoad |
|
Issue occurs for me on commit |
|
@MaxandreOgeret it's commit ddf0d76. Gamescope is broken since 0110109, read my previous post |
|
Can confirm, 0110109 works in Wayland on my RX580. SDL backend also works along with this patch. |
ProjectSynchro
added a commit
to ProjectSynchro/copr-gamescope-git
that referenced
this issue
Oct 5, 2024
ProjectSynchro
added a commit
to ProjectSynchro/copr-gamescope-git
that referenced
this issue
Oct 5, 2024
|
Still broken as of 06a6225 BTW |
|
Of course, because #1548 is still not merged. |
|
Still an issue on X11 |
see :
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
since update to 3.14.2 I'm facing this error every time I running something through gamescope, it crashes immediately. It worked fine with 3.13.16
Here is an example with vkcube:
No CAP_SYS_NICE, falling back to regular-priority compute and threads.
Performance will be affected.
xdg_backend: Seat name:
xdg_backend: PreferredMetadata: Red: 0.64 0.33, Green: 0.3 0.6, Blue: 0.15 0.06, White: 0.3127 0.329, Max Luminance: 100 nits, Min Luminance: 0 nits, Max Full Frame Luminance: 100 nits
ATTENTION: default value of option vk_khr_present_wait overridden by environment.
ATTENTION: default value of option vk_khr_present_wait overridden by environment.
vulkan: selecting physical device 'AMD Radeon RX 570 Series (RADV POLARIS10)': queue family 1 (general queue family 0)
vulkan: physical device does not support DRM format modifiers
wlserver: [backend/headless/backend.c:67] Creating headless backend
vulkan: supported DRM formats for sampling usage:
gamescope: types/wlr_linux_dmabuf_v1.c:526: feedback_compile: Assertion `table_len > 0' failed.
The text was updated successfully, but these errors were encountered: