[PATCH v3 2/5] Make libc symbols hidden in static PIE

[H.J. Lu]

On Fri, Jan 15, 2021 at 4:41 PM H.J. Lu <hjl.tools@gmail.com> wrote:
>
> On Fri, Jan 15, 2021 at 2:42 PM H.J. Lu <hjl.tools@gmail.com> wrote:
> >
> > On Fri, Jan 15, 2021 at 7:28 AM H.J. Lu <hjl.tools@gmail.com> wrote:
> > >
> > > On Fri, Jan 15, 2021 at 6:27 AM Szabolcs Nagy <szabolcs.nagy@arm.com> wrote:
> > > >
> > > > The 01/15/2021 05:43, H.J. Lu wrote:
> > > > > On Fri, Jan 15, 2021 at 3:25 AM Szabolcs Nagy <szabolcs.nagy@arm.com> wrote:
> > > > > > options:
> > > > > >
> > > > > > (1) fix extern hidden pie calls on i686 (by making
> > > > > >     them the same as default vis pie calls so ifuncs
> > > > > >     work).
> > > > > >
> > > > > > (2) annotate ifuncs (avoid hidden for them, ifuncs
> > > > > >     cannot appear in early start code anyway because
> > > > > >     of IRELATIVE): this can be difficult to maintain.
> > > > > >
> > > > > > (3) annotate early object accesses to be hidden so
> > > > > >     RELATIVE relocs are avoided. (most targets want
> > > > > >     all objects to be hidden, but this solves bug
> > > > > >     27072 without causing problems on i686)
> > > > > >
> > > > > > (4) make pie always use copy relocations on i686.
> > > > > >     (and then no hidden annotation is needed, object
> > > > > >     access is always local in pie).
> > > > ...
> > > > > > > --- a/elf/dl-reloc-static-pie.c
> > > > > > > +++ b/elf/dl-reloc-static-pie.c
> > > > > > > @@ -17,6 +17,7 @@
> > > > > > >     <https://www.gnu.org/licenses/>.  */
> > > > > > >
> > > > > > >  #if ENABLE_STATIC_PIE
> > > > > > > +#pragma GCC visibility push(hidden)
> > > > > >
> > > > > > yes, this is option (3). you will also need it in _dl_aux_init
> > > > > > and __libc_init_secure and __tunables_init.
> > > > >
> > > > > I will try it.
> > > >
> > > > the naive way does not seem to work:
> > > >
> > > > _dl_support.c has _dl_aux_init as well as _dl_non_dynamic_init,
> > > > the former needs hidden the latter does not and calls rawmemchr
> > > > which is ifunc on i686.
> > > >
> > > > i think the easiest fix is to move those two functions into
> > > > separate files. (ideally we would have a small set of files
> > > > that are involved in the start code before self relocation)
> > > >
> > > > now i realized that there is another option:
> > > >
> > > > (5) remove all ifuncs from i686 libc.
> > > >
> > > > i assume there are not many users who care about i686 performance.
> > >
> > > I don't know if this will work on i686.  Since i386 doesn't have IP relative
> > > addressing, we can't remove all RELATIVE relocations.  We need to
> > > call _dl_aux_init again after relocating PIE.  I don't know what other symbols
> > > are affected.  My current patches are on users/hjl/pr27072/master branch:
> > >
> > > https://gitlab.com/x86-glibc/glibc/-/commits/users/hjl/pr27072/master
> > >
> >
> > The problem is
> >
> > #ifdef NEED_DL_SYSINFO
> > /* Needed for improved syscall handling on at least x86/Linux.  */
> > uintptr_t _dl_sysinfo = DL_SYSINFO_DEFAULT;
> > #endif
> >
> > We can initialize it in _dl_aux_init instead.
> >
> > I am testing this set of patches on top of yours on i686 and x86-64.
> >
>
> They worked and they passed build-many-glibcs.py.
>

I combined my patches, including 4 testcases, with yours in the right
order here:

https://gitlab.com/x86-glibc/glibc/-/commits/users/hjl/pr27072/master


-- 
H.J.