[Bug middle-end/64592] New: [5 regression] tramp3d EH unwind tables are 50% bigger with mainline compared to GCC 4.9

[Bug middle-end/64592] New: [5 regression] tramp3d EH unwind tables are 50% bigger with mainline compared to GCC 4.9

[hubicka at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64592

            Bug ID: 64592
           Summary: [5 regression] tramp3d EH unwind tables are 50% bigger
                    with mainline compared to GCC 4.9
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: middle-end
          Assignee: unassigned at gcc dot gnu.org
          Reporter: hubicka at gcc dot gnu.org

root@okamzik:/aux/hubicka# size tramp3d-50-Os-2 -A | grep eh_
.eh_frame                122932   4554928
.eh_frame_hdr             11972   4677860
root@okamzik:/aux/hubicka# size tramp3d-49-Os -A | grep eh_
.eh_frame                 79580   4560264
.eh_frame_hdr             12572   4639844

[Bug middle-end/64592] [5 regression] tramp3d EH unwind tables are 50% bigger with mainline compared to GCC 4.9

[hubicka at ucw dot cz]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64592

--- Comment #2 from Jan Hubicka <hubicka at ucw dot cz> ---
> accumulate-outgoing-args default setting change?
Nope, 4.9 already had -mno-accumulate-outgoing-args with -Os.

[Bug middle-end/64592] [5 regression] tramp3d EH unwind tables are 50% bigger with mainline compared to GCC 4.9

[hubicka at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64592

--- Comment #3 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
With help of Jakub we managed to figure out that on my setup it is because GCC
5.0 defaults to no .cfi directives, while GCC 4.9 uses them.
This seems to be because cfi working advance test, but we do not see ho this
test changed recently.

It may be helpful to update binutils on tester machines at opensuse.org

[Bug middle-end/64592] [5 regression] tramp3d EH unwind tables are 50% bigger with mainline compared to GCC 4.9

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64592

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
What binutils version do you have out of interest?

[Bug middle-end/64592] [5 regression] tramp3d EH unwind tables are 50% bigger with mainline compared to GCC 4.9

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64592

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |WORKSFORME

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Must be some Debian specific bug.  I've tried Fedora binutils 2.22 as well as
some earlier ones and cfi advance configury test on the trunk passes just fine.
In any case, I think the problem is analyzed and is caused by buggy assembler.