[Mali-G76] Fuel and other 32-bit games/dx test don't work

[Artewar67]

I don't know if this a duplicate so i decide to create separate issue.

Game "Fuel" which i want to test for #155. Don't work on my gpu, game just close after been run.

Here is logs.

Wrapper logs

vvl_2025_08_19_15_19_37.FUEL.exe.9585.txt

wrapper_cmds_2025_08_19_15_19_37.FUEL.exe.9585.txt

wrapper_log_2025_08_19_15_19_37.FUEL.exe.9585.txt

DXVK Log:

FUEL_d3d9.log

Wine log:

fuel_2025-08-19_15-19-34.txt

[leegao]

Validation Error: [ VUID-VkExportMemoryAllocateInfo-handleTypes-09860 ] | MessageID = 0x32be836e
vkBindImageMemory(): The handle type (VK_EXTERNAL_MEMORY_HANDLE_TYPE_ANDROID_HARDWARE_BUFFER_BIT_ANDROID) specified by the memory's VkExportMemoryAllocateInfo, and the VkImageCreateInfo
format (VK_FORMAT_R32_UINT)
type (VK_IMAGE_TYPE_3D)
tiling (VK_IMAGE_TILING_OPTIMAL)
usage (VK_IMAGE_USAGE_TRANSFER_DST_BIT|VK_IMAGE_USAGE_SAMPLED_BIT|VK_IMAGE_USAGE_STORAGE_BIT)
flags (VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT)
is not supported combination of parameters. vkGetPhysicalDeviceImageFormatProperties2 returned back VK_ERROR_FORMAT_NOT_SUPPORTED.
The Vulkan spec states: The bits in handleTypes must be supported and compatible, as reported by VkExternalTensorPropertiesARM, VkExternalImageFormatProperties, or VkExternalBufferProperties (https://docs.vulkan.org/spec/latest/chapters/memory.html#VUID-VkExportMemoryAllocateInfo-handleTypes-09860)
Objects: 2
    [0] VkImage 0x1c000000001c
    [1] VkDeviceMemory 0x180000000018

...

Validation Error: [ VUID-VkBindBufferMemoryInfo-memory-02726 ] | MessageID = 0x9fa0169c
(Warning - This VUID has now been reported 10 times, which is the duplicated_message_limit value, this will be the last time reporting it).
vkBindBufferMemory2(): pBindInfos[0].memory (VkDeviceMemory 0x420000000042) has an external handleType of VK_EXTERNAL_MEMORY_HANDLE_TYPE_ANDROID_HARDWARE_BUFFER_BIT_ANDROID which does not include at least one handle from VkBuffer (VkBuffer 0x480000000048) handleType VkExternalMemoryHandleTypeFlags(0).
The Vulkan spec states: If the value of VkExportMemoryAllocateInfo::handleTypes used to allocate memory is not 0, it must include at least one of the handles set in VkExternalMemoryBufferCreateInfo::handleTypes when buffer was created (https://docs.vulkan.org/spec/latest/chapters/resources.html#VUID-VkBindBufferMemoryInfo-memory-02726)
Objects: 2
    [0] VkBuffer 0x480000000048
    [1] VkDeviceMemory 0x420000000042

...

Validation Error: [ VUID-VkMemoryAllocateInfo-pNext-01874 ] | MessageID = 0x2eed5bfe
vkAllocateMemory(): pAllocateInfo->pNext<VkMemoryDedicatedAllocateInfo>.image is VkImage 0x4a000000004a but allocationSize is 7488.
The Vulkan spec states: If the parameters do not define an import operation, and the pNext chain includes a VkExportMemoryAllocateInfo structure with VK_EXTERNAL_MEMORY_HANDLE_TYPE_ANDROID_HARDWARE_BUFFER_BIT_ANDROID included in its handleTypes member, and the pNext chain includes a VkMemoryDedicatedAllocateInfo structure with image not equal to VK_NULL_HANDLE, then allocationSize must be 0 (https://docs.vulkan.org/spec/latest/chapters/memory.html#VUID-VkMemoryAllocateInfo-pNext-01874)
Objects: 1
    [0] VkImage 0x4a000000004a

# wrapper log
[001.39] lseek(0, SEEK_END) failed on the AHB fd (idx=0, fd=99), errno = 29, trying another fd 	 (__wrapper_MapMemory2KHR:458)

# wine log
00f4:trace:vulkan:thunk32_vkCreateSwapchainKHR 0x1b36f40, 0x12f230, 0, 0x13095e8
00f4:trace:vulkan:X11DRV_vkCreateSwapchainKHR 0xb400007d41fe4000 0x1000fef60 0x0 0xb400007c85118e80
00f4:err:msvcrt:_wassert (L"!status && \"vkCreateSwapchainKHR\"",L"../dlls/winevulkan/loader_thunks.c",3302)
00f4:trace:vulkan:DllMain 799C0000, 0, 00000001

[leegao]

Similar symptoms to #122

[leegao]

libvulkan_wrapper.so.zip

Can you grab logs with this build with WRAPPER_WSI_LOG_LEVEL=all set? This build adds some preliminary WSI logs

[Artewar67]

Okay i will try

[Artewar67]

Here log

wrapper_log_2025_08_22_20_39_24.FUEL.exe.30900.txt

[leegao]

[001.34] D Calling x11_surface_create_swapchain 	 (x11_surface_create_swapchain:2338)
[001.34] D In wsi_create_image 	 (wsi_create_image:854)
[001.34] D blitting 	 (wsi_create_ahardware_buffer_blit_context:145)

looks like wsi_create_image never returned (it crashed somewhere), the usual sequence should be:

[001.52] D Calling x11_surface_create_swapchain 	 (x11_surface_create_swapchain:2338)
[001.53] D In wsi_create_image 	 (wsi_create_image:854)
[001.53] D blitting 	 (wsi_create_ahardware_buffer_blit_context:145)
[001.55] D info->create_mem: 0 	 (wsi_create_image:872)
[001.56] D wsi_create_image: 0 	 (x11_image_init:1897)
[001.57] D x11_image_init: result = 0 	 (x11_surface_create_swapchain:2527)

[leegao]

WSI configuration:

[000.05] D wsi->semaphore_export_handle_types: VkExternalSemaphoreHandleTypeFlags = 10 	 (wsi_device_init:370)
[000.05] D wsi->timeline_semaphore_export_handle_types: VkExternalSemaphoreHandleTypeFlags = 10 	 (wsi_device_init:372)
[000.05] D wsi->has_import_memory_host: bool = 0 	 (wsi_device_init:374)
[000.05] D wsi->has_timeline_semaphore: bool = 1 	 (wsi_device_init:376)
[000.05] D wsi->supports_scanout: bool = 1 	 (wsi_device_init:378)
[000.05] D wsi->supports_modifiers: bool = 0 	 (wsi_device_init:380)
[000.05] D wsi->maxImageDimension2D: uint = 16384 	 (wsi_device_init:382)
[000.05] D wsi->optimalBufferCopyRowPitchAlignment: uint = 64 	 (wsi_device_init:384)
[000.05] D wsi->override_present_mode: VkPresentModeKHR = 1 	 (wsi_device_init:386)
[000.05] D wsi->force_bgra8_unorm_first: bool = 0 	 (wsi_device_init:388)
[000.05] D wsi->enable_adaptive_sync: bool = 0 	 (wsi_device_init:390)
[000.05] D wsi->force_headless_swapchain: bool = 0 	 (wsi_device_init:392)
[000.05] D wsi->force_swapchain_to_currentExtent: bool = 0 	 (wsi_device_init:394)
[000.05] D wsi->x11.override_minImageCount: uint = 0 	 (wsi_device_init:397)
[000.05] D wsi->x11.strict_imageCount: bool = 0 	 (wsi_device_init:399)
[000.05] D wsi->x11.ensure_minImageCount: bool = 0 	 (wsi_device_init:401)
[000.05] D wsi->x11.xwaylandWaitReady: bool = 0 	 (wsi_device_init:403)
[000.05] D wsi->x11.extra_xwayland_image: bool = 0 	 (wsi_device_init:405)
[000.05] D wsi->x11.ignore_suboptimal: bool = 0 	 (wsi_device_init:407)
[000.05] D wsi->sw: bool = 0 	 (wsi_device_init:410)
[000.05] D wsi->wants_ahardware_buffer: bool = 0 	 (wsi_device_init:412)
[000.05] D wsi->needs_blit: bool = 1 	 (wsi_device_init:414)
[000.05] D wsi->wants_linear: bool = 0 	 (wsi_device_init:416)
[000.05] D wsi->signal_semaphore_with_memory: bool = 0 	 (wsi_device_init:418)
[000.05] D wsi->signal_fence_with_memory: bool = 0 	 (wsi_device_init:420)
[000.05] D wsi->khr_present_wait: bool = 1 	 (wsi_device_init:422)

[leegao]

I gotta run, but I'll add more targeted logs in that function next to see where exactly the crash is

[leegao]

@Artewar67 can you enable:

WRAPPER_LOG_LEVEL=all
WRAPPER_WSi_LOG_LEVEL=all

for this build and upload the wrapper logs

libvulkan_wrapper.so.zip

[Artewar67]

Here: wrapper_log_2025_08_23_22_26_21.FUEL.exe.16553.txt

[leegao]

@Artewar67 weird it looks like that log didn't take the recent logging updates I added, try this one maybe? Same setup with WRAPPER_WSi_LOG_LEVEL=all and WRAPPER_LOG_LEVEL=all

libvulkan_wrapper.so.zip

[leegao]

libvulkan_wrapper.so.zip

Actually, this new build ^ (adds more logging for x11 side of the wsi too)

[Artewar67]

Here: wrapper_log_2025_08_24_13_26_36.FUEL.exe.10206.txt

Also, small i in WRAPPER_WSi_LOG_LEVEL is typo or not? I don't know so added bothWRAPPER_WSi_LOG_LEVEL and WRAPPER_WSI_LOG_LEVEL