This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: [testcase] dwarf2out ICE


>>>>> "Jakub" == Jakub Jelinek <jakub@redhat.com> writes:

> The following testcase ICEs in add_abstract_origin. I believe dwarf2out is
> confused by the same function beeing first extern inline and then defined as
> non-inline function, the ICE is when looking up die for f2.

Well, yes.  The dwarf2 backend assumes, I think reasonably, that there will
only be one inline function for each FUNCTION_DECL, so it never emits the
abstract version of the second f1, so when we go looking for it we fail.
This can be fixed by suppressing inlining of the second f1, which we should
do anyway.

I think we should move DECL_UNINLINABLE into the backend.

> Can I check the testcase in?

There is already a testcase like this: gcc.c-torture/compile/20000120-2.c.

Jason


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]