@polter-rnd Good catch! A call to munmap() was indeed missing, causing the accumulation of mmap'ed objects you've detected.
It turns out that I had wrongly interpreted the Wayland documentation of function wl_shm_pool_destroy() that reads

The mmapped memory will be released when all
buffers that have been created from this pool
are gone.

I thought it meant the Wayland library does the munmap() call. Apparently, that's incorrect. The Wayland client program is expected to call munmap(). This requires to add a means to detect when all buffers from a pool are gone. That's hopefuly in place in the commit mentioned above.

Please, close issue if you confirm the memory leak is fixed.

—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you are subscribed to this thread.Message ID: <fltk/fltk/issues/730/1572259805@github.com>