Memory leaks in nvidia driver

[TheComet]

I am unsure who to contact or where to report, so I'll start here. Maybe you can refer me.

On my journey to learning the vulkan API I have discovered memory leaks in my graphics driver. See this thread for details on how to reproduce it:
https://forums.khronos.org/showthread.php/13307-Memory-leaks-in-vulkan?p=41530

A quick summary: I do the simplest thing possible, namely create and immediately destroy a vulkan instance. By keeping track of malloc() and free() calls I noticed that 45 calls to malloc() cannot be matched to their respective calls to free(). Further, by creating two instances (instead of one) and then destroying them again, I find that the memory leak count remains at 45 instead of going up to the expected 90. This tells me that there must be some kind of global state not being cleaned up.

A report of each leak with backtraces can be found here: http://pastebin.com/td5iX7Z0

I have reproduced this issue on two separate machines, both running gentoo linux.

System 1: nVidia GTX 580 using x11-drivers/nvidia-drivers-375.20-r1, kernel 4.4.26, x86_64
System 2: nVidia GT 520M using x11-drivers/nvidia-drivers-375.20-r1, kernel 4.4.21, x86_64

A second person has reproduce the issue (see thread) on his W10 machine with no leaks.

Who can I report this to? How can I help fix this?

[roarinelk]

Write a mail with detailed steps on how to reproduce it to linux-bugs@nvidia.com and/or post in [1]
The nvidia devs don't respond immediately, but they usually do respond eventually.


[1] https://devtalk.nvidia.com/default/board/98/linux