[Bug debug/84342] [8 Regression] Location views breaks cross builds of arm including gnueabihf
aoliva at gcc dot gnu.org
gcc-bugzilla@gcc.gnu.org
Tue Feb 13 11:07:00 GMT 2018
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84342
Alexandre Oliva <aoliva at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |ASSIGNED
Assignee|unassigned at gcc dot gnu.org |aoliva at gcc dot gnu.org
--- Comment #2 from Alexandre Oliva <aoliva at gcc dot gnu.org> ---
This assembler assert error is caused by a bug in the arm back-end, that makes
GCC assume it can reset the view counter at an insn because the back-end says
it has positive min length, but the insn turns out to have length zero. I
suggest you to compile that with -dA -S -fdump-rtl-final, and then figure out
what insn's length is miscomputed so as to fix the backend.
That said, there was general consensus that we don't want to rely on insn
lengths to reset view numbers for now, so there's a patch pending review that
should make this problem go latent, disabling the view counter internal reset
machinery, leaving it for the assembler alone to determine view reset points.
https://gcc.gnu.org/ml/gcc-patches/2018-02/msg00624.html
Please let me know whether you still observe the problem with that patch,
before or after it goes in. Thanks,
More information about the Gcc-bugs
mailing list