Comment # 6 on bug 1181104 from Stefan Seyfried

Hi Takashi,

I could not get the kernel to crash with 4.18, but I did (sometimes, this is
why I suspected a race...) get "(null)" as name in the deallocation message.

I assumed that this was the "original" bug which later (newer gcc, 4.18 did not
even build with gcc9 for me, so used gcc7, different optimization, whatever...)
turned into a crash.

And this is probably why it was impossible (for me) to bisect, apart from "it
sometimes fails only once in ten tries" being hard to make a "good/bad"
decision :-)

I'm building the patched driver right now, will be back with test results,
soon.