running diag(NULL, 1) makes R crash, without an error of warning. I don't know if it is windows specific though. This behavior was also present in 3.2.4.
system x86_64, mingw32
svn rev 70478
version.string R version 3.2.5 (2016-04-14)
nickname Very, Very Secure Dishes
I can confirm this, and the fact that it's easy to fix. The code used the LENGTH() macro without checking that it was applicable. (You need to use the length() function on NULL to get zero.)
Have committed fix to R-devel; soon to the beta.