On 2/21/20 1:56 PM, José Rui Faustino de Sousa wrote:
Since the cfi.n pointer is uninitialized in some infrequent situations (using -static-libgfortran seems to do the trick) the pointer seems to contain garbage and a segmentation fault is generated
Hmm, that sounds like papering over a real bug. At a glance, I do not see anything in the test case which is undefined behaviour – and if it is not undefined behaviour, it should work with all optimization options and with libgfortran linked both statically and dynamically.
Tobias PS: I am woefully aware that there a still several patches to be reviewed.
