The vulkan shouldn't be aware of the X driver running and it shouldn't make any difference.
Can you take a backtrace of the crash?

Thanks

FWIW, Haswell is not fully supported by our Vulkan driver.

Created attachment 143524 [details]
vulkaninfo

vulkan-smoketest randomly crashing without any error output, only show fps.

so have to wait for fully surpport.

regards.

(In reply to jydcome from comment #3)
> Created attachment 143524 [details]
> vulkaninfo
> 
> vulkan-smoketest randomly crashing without any error output, only show fps.
> 
> so have to wait for fully surpport.
> 
> regards.

The backtrace of vulkan-smoketest is what we would need to debug this issue.

howto get backtrace? vulkan-smoketest only repeat output fps.

running vulkan-smoketest --vv got:
terminate called after throwing an instance of 'std::runtime_error'
  what():  instance layer VK_LAYER_LUNARG_standard_validation is missing
Aborted

is that?

(In reply to jydcome from comment #6)
> running vulkan-smoketest --vv got:
> terminate called after throwing an instance of 'std::runtime_error'
>   what():  instance layer VK_LAYER_LUNARG_standard_validation is missing
> Aborted
> 
> is that?

Is this the crash you're running into?
If yes, then it means you're missing the validation layers.

Otherwise to get a backtrace, run the application under gdb :
gdb --args vulkan-smoketest

When it crash, just type :

bt

And past the printed lines here.

after make some packages upgrade, something interesting.

reboot then first login, run vulkan-smoketest, cpuload ~100%, got output:
545 presents in 5.00498 seconds (FPS: 108.892)
556 presents in 5.00228 seconds (FPS: 111.149)
561 presents in 5.00833 seconds (FPS: 112.013)
frames:5939, elapsedms:54985

then quit and run vulkan-smoketest what ever how many times, cpu load ~13%, got output:
48 presents in 5.0079 seconds (FPS: 9.58485)
49 presents in 5.08955 seconds (FPS: 9.62757)
frames:191, elapsedms:19724

when it crash, output:
51 presents in 5.07859 seconds (FPS: 10.0422)
49 presents in 5.02128 seconds (FPS: 9.75847)
terminate called after throwing an instance of 'std::runtime_error'
  what():  VkResult -1000001004 returned
Aborted

by (gdb) bt output:
(gdb) bt
#0  0x00007ffff672afff in raise () from /lib/x86_64-linux-gnu/libc.so.6
#1  0x00007ffff672c42a in abort () from /lib/x86_64-linux-gnu/libc.so.6
#2  0x00007ffff6d3f0ad in __gnu_cxx::__verbose_terminate_handler() ()
   from /usr/lib/x86_64-linux-gnu/libstdc++.so.6
#3  0x00007ffff6d3d066 in ?? () from /usr/lib/x86_64-linux-gnu/libstdc++.so.6
#4  0x00007ffff6d3d0b1 in std::terminate() () from /usr/lib/x86_64-linux-gnu/libstdc++.so.6
#5  0x00007ffff6d3d2c9 in __cxa_throw () from /usr/lib/x86_64-linux-gnu/libstdc++.so.6
#6  0x0000555555560d2d in ?? ()
#7  0x000055555556a8da in ?? ()
#8  0x00005555555703c5 in ?? ()
#9  0x0000555555570f30 in ?? ()
#10 0x0000555555558545 in ?? ()
#11 0x00007ffff67182e1 in __libc_start_main () from /lib/x86_64-linux-gnu/libc.so.6
#12 0x00005555555586ca in _start ()

The backtrace seems to indicate this isn't a crash in the driver, but rather the application.

crash because the app itself? i move the app window radomly will cause crash but not everytime. 

and only first login run it got over 100 fps then only ~10 fps?

The crash is an exception raised from c++ code. Our driver mostly doesn't use c++ and where it does, doesn't use exceptions.
I'm inclined to think that the app is not dealing with resize events correctly.

This doesn't look like a driver issue.