Re: [PATCH] Make calls in virtual thunks local if possible

Re: [PATCH] Make calls in virtual thunks local if possible

[John David Anglin]

This patch has broken the hppa-unknown-linux-gnu port.

/home/dave/gcc-3.4/objdir/gcc/xgcc -shared-libgcc -B/home/dave/gcc-3.4/objdir/gcc/ -nostdinc++  -L/home/dave/gcc-3.4/objdir/hppa-linux/libstdc++-v3/src -L/home/dave/gcc-3.4/objdir/hppa-linux/libstdc++-v3/src/.libs -B/home/dave/opt/gnu/hppa-linux/bin/ -B/home/dave/opt/gnu/hppa-linux/lib/ -isystem /home/dave/opt/gnu/hppa-linux/include -isystem /home/dave/opt/gnu/hppa-linux/sys-include -shared -nostdlib /home/dave/opt/gnu/lib/crti.o /home/dave/gcc-3.4/objdir/gcc/crtbeginS.o  .libs/allocator-inst.o .libs/codecvt.o .libs/complex_io.o .libs/concept-inst.o .libs/ctype.o .libs/demangle.o .libs/ext-inst.o .libs/fstream-inst.o .libs/functexcept.o .libs/globals.o .libs/io-inst.o .libs/ios.o .libs/istream-inst.o .libs/limits.o .libs/locale.o .libs/locale-inst.o .libs/localename.o .libs/misc-inst.o .libs/ostream-inst.o .libs/sstream-inst.o .libs/stdexcept.o .libs/streambuf-inst.o .libs/string-inst.o .libs/strstream.o .libs/valarray-inst.o .libs/wstring-inst.o .libs/codecvt_members.o .lib!
 s/collate_members.o .libs/ctype_members.o .libs/messages_members.o .libs/monetary_members.o .libs/numeric_members.o .libs/time_members.o .libs/basic_file.o .libs/c++locale.o -Wl,--whole-archive ../libmath/.libs/libmath.a ../libsupc++/.libs/libsupc++convenience.a -Wl,--no-whole-archive  -L/home/dave/gcc-3.4/objdir/hppa-linux/libstdc++-v3/src -L/home/dave/gcc-3.4/objdir/hppa-linux/libstdc++-v3/src/.libs -lm ../libmath/.libs/libmath.a -lm ../libsupc++/.libs/libsupc++convenience.a -lm -L/home/dave/gcc-3.4/objdir/gcc -L/home/dave/opt/gnu/hppa-linux/bin -L/home/dave/opt/gnu/hppa-linux/lib -L/home/dave/opt/gnu/lib -L/home/dave/opt/gnu/lib/gcc-lib/hppa-linux/3.4 -L/home/dave/opt/gnu/lib/gcc-lib/hppa-linux/3.4/../../../../hppa-linux/lib -L/home/dave/opt/gnu/lib/gcc-lib/hppa-linux/3.4/../../.. -lgcc_s -lgcc -lc -lgcc_s -lm -lgcc_s -lgcc -lc -lgcc_s   /home/dave/gcc-3.4/objdir/gcc/crtendS.o /home/dave/opt/gnu/lib/crtn.o  -Wl,-O1 -Wl,--version-script=libstdc++-symbol.ver -Wl,-soname -W!
 l,libstdc++.so.6 -o .libs/libstdc++.so.6.0.0
/home/dave/opt/gnu/bin/ld: BFD 2.14.90 20030602 internal error, aborting at ../../src/bfd/elf32-hppa.c line 3864 in elf32_hppa_relocate_section

/home/dave/opt/gnu/bin/ld: Please report this bug.

collect2: ld returned 1 exit status

I'm not entirely sure what's wrong but it appears that ld can't handle
the following situation (from io-inst.s):

        .section        .gnu.linkonce.t._ZNSdD1Ev,"ax",@progbits
	.align 4
	.weak   _ZNSdD1Ev
	.type   _ZNSdD1Ev, @function
.LFB1804:
_ZNSdD1Ev:

	...

        .set    .LTHUNK1,_ZNSdD1Ev

	...

        .data
	.align 4
.LTHN0:
	.word P'.LTHUNK1

The P' is a procedure descriptor (plabel) constructor.  Using a procedure
descriptor for the branch from the thunk was necessary under hpux.  This
prevents the linker from adding a stub between the branch and its target.

This technique was carried over to linux.  It might be possible to
change linux to use a simple pc-relative branch in the thunk.  The
stubs inserted by the linux linker don't mess with the return pointer
(except in the multispace model which the kernel doesn't currently
support).  Possibly, we won't get a stub at all if the branch is local.

However, it can be seen that the thunk follows the function.  As a result,
we can't be certain that a pc-relative branch will be able to reach the
start of the function.  It would be better to have the thunk just before the
function if possible.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc-cnrc.gc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)

Re: [PATCH] Make calls in virtual thunks local if possible

[Jason Merrill]

On Wed, 25 Jun 2003 11:46:20 -0400 (EDT), "John David Anglin" <dave@hiauly1.hia.nrc.ca> wrote:

> However, it can be seen that the thunk follows the function.  As a result,
> we can't be certain that a pc-relative branch will be able to reach the
> start of the function.  It would be better to have the thunk just before the
> function if possible.

Indeed.  Jakub, can you make that change?

Jason

Re: [PATCH] Make calls in virtual thunks local if possible

[Jakub Jelinek]

On Wed, Jun 25, 2003 at 11:46:20AM -0400, John David Anglin wrote:
> This technique was carried over to linux.  It might be possible to
> change linux to use a simple pc-relative branch in the thunk.  The
> stubs inserted by the linux linker don't mess with the return pointer
> (except in the multispace model which the kernel doesn't currently
> support).  Possibly, we won't get a stub at all if the branch is local.

The output_mi_thunk target hook should for binds_local_p function create
a tail call to function. For
void foo (void) __attribute__((visibility ("hidden")));
void bar (void)
{
  foo ();
}
tail call GCC must emit a sequence which will work no matter how far foo
from bar's end is, if it is in the same shared library and output_mi_thunk
should emit the same. If this is impossible on certain arches,
then can_output_mi_thunk can just bail out and let the generic code handle
it.

> However, it can be seen that the thunk follows the function.  As a result,
> we can't be certain that a pc-relative branch will be able to reach the
> start of the function.  It would be better to have the thunk just before the
> function if possible.

How would it help? You can have many thunks, so if the pc-relative branch
is severely limited, you can as well not be able to branch accross the
other thunks.

	Jakub