Le 25/09/2015 23:41, Steve Kargl a écrit :
The attached patch has been built and regression
tested on x86_64-*-freebsd. No regression occurred.
The patch removes a conditional within an assert()
that triggers when a BLOCK construct is encountered.
What happens here, I think, is symtree is a new block-local symbol
instead of the host-associated one.
symtree is basically unused here, yet playing with the wrong one is not
right.
Can you try using gfc_get_ha_sym_tree a few lines above instead of
gfc_get_sym_tree?
A patch like that is preapproved.
Mikael
