https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91255
ktkachov at gcc dot gnu.org changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Last reconfirmed| |2019-07-31
Ever confirmed|0 |1
--- Comment #1 from ktkachov at gcc dot gnu.org ---
This can reproduced by copy-pasting the define_subst example from the
documentation
https://gcc.gnu.org/onlinedocs/gccint/Define-Subst-Example.html#Define-Subst-Example
into and .md file in config/arm.
The erroring out happens in has_subst_attribute in gensupport.c. Commenting out
the error in case SET_ATTR_ALTERNATIVE and just returning false allows the
build to proceed but I suppose we'd want to validate that the
set_attr_alternative contents don't contain any subst attributes so that we
don't silently reject uses of subst attributes if the backend writer mistakenly
expected them to work.
Alternatively, there's more error checking in alter_attrs_for_subst_insn that
rejects set_attr_alternative but can we rely on that to trigger appropriately?
