RFC: x86: Fall back to lazy binding for unrelocated IFUNC symbol [BZ #23240]
Commit Message
On Mon, May 28, 2018 at 6:45 AM, Carlos O'Donell <carlos@redhat.com> wrote:
> On 05/28/2018 05:27 AM, Florian Weimer wrote:
>> On 05/27/2018 04:36 PM, H.J. Lu wrote:
>>
>>> I'd like to avoid these messages.
>>
>> We could restrict the warning to LD_DEBUG=bindings.
>
> I agree that this would be a better solution.
>
> Disabling BIND_NOW is the wrong thing to do in this case.
>
Here is the updated patch. It fails to back lazy binding only
if GOT is writable.
Comments
On 05/28/2018 09:51 PM, H.J. Lu wrote:
> Here is the updated patch. It fails to back lazy binding only
> if GOT is writable.
I think this still sets a wrong incentive, as in “you need to have a
writable GOT for full forwards compatibility with future library evolution”.
I think if we detect an IFUNC-based relocation against a
not-yet-fully-relocated object, we need to delay this relocation
processing and perform a second pass of all delayed relocations. I have
a patch which needs a little bit of work, but looks quite promising (to
me at least).
Thanks,
Florian
On 06/05/2018 09:24 PM, Florian Weimer wrote:
> On 05/28/2018 09:51 PM, H.J. Lu wrote:
>> Here is the updated patch. It fails to back lazy binding only
>> if GOT is writable.
>
> I think this still sets a wrong incentive, as in “you need to have a
> writable GOT for full forwards compatibility with future library
> evolution”.
>
> I think if we detect an IFUNC-based relocation against a
> not-yet-fully-relocated object, we need to delay this relocation
> processing and perform a second pass of all delayed relocations. I have
> a patch which needs a little bit of work, but looks quite promising (to
> me at least).
Here are my patches:
https://sourceware.org/ml/libc-alpha/2018-06/msg00076.html
https://sourceware.org/ml/libc-alpha/2018-06/msg00077.html
I hope they address your needs.
Thanks,
Florian
On Wed, Jun 6, 2018 at 7:04 AM, Florian Weimer <fweimer@redhat.com> wrote:
> On 06/05/2018 09:24 PM, Florian Weimer wrote:
>>
>> On 05/28/2018 09:51 PM, H.J. Lu wrote:
>>>
>>> Here is the updated patch. It fails to back lazy binding only
>>> if GOT is writable.
>>
>>
>> I think this still sets a wrong incentive, as in “you need to have a
>> writable GOT for full forwards compatibility with future library evolution”.
>>
>> I think if we detect an IFUNC-based relocation against a
>> not-yet-fully-relocated object, we need to delay this relocation processing
>> and perform a second pass of all delayed relocations. I have a patch which
>> needs a little bit of work, but looks quite promising (to me at least).
>
>
> Here are my patches:
>
> https://sourceware.org/ml/libc-alpha/2018-06/msg00076.html
> https://sourceware.org/ml/libc-alpha/2018-06/msg00077.html
>
> I hope they address your needs.
Do you have a git branch I can use?
On 06/06/2018 04:50 PM, H.J. Lu wrote:
> On Wed, Jun 6, 2018 at 7:04 AM, Florian Weimer <fweimer@redhat.com> wrote:
>> On 06/05/2018 09:24 PM, Florian Weimer wrote:
>>>
>>> On 05/28/2018 09:51 PM, H.J. Lu wrote:
>>>>
>>>> Here is the updated patch. It fails to back lazy binding only
>>>> if GOT is writable.
>>>
>>>
>>> I think this still sets a wrong incentive, as in “you need to have a
>>> writable GOT for full forwards compatibility with future library evolution”.
>>>
>>> I think if we detect an IFUNC-based relocation against a
>>> not-yet-fully-relocated object, we need to delay this relocation processing
>>> and perform a second pass of all delayed relocations. I have a patch which
>>> needs a little bit of work, but looks quite promising (to me at least).
>>
>>
>> Here are my patches:
>>
>> https://sourceware.org/ml/libc-alpha/2018-06/msg00076.html
>> https://sourceware.org/ml/libc-alpha/2018-06/msg00077.html
>>
>> I hope they address your needs.
>
> Do you have a git branch I can use?
I've pushed it to fw/bug21242.
Florian
On Wed, Jun 6, 2018 at 7:54 AM, Florian Weimer <fweimer@redhat.com> wrote:
> On 06/06/2018 04:50 PM, H.J. Lu wrote:
>>
>> On Wed, Jun 6, 2018 at 7:04 AM, Florian Weimer <fweimer@redhat.com> wrote:
>>>
>>> On 06/05/2018 09:24 PM, Florian Weimer wrote:
>>>>
>>>>
>>>> On 05/28/2018 09:51 PM, H.J. Lu wrote:
>>>>>
>>>>>
>>>>> Here is the updated patch. It fails to back lazy binding only
>>>>> if GOT is writable.
>>>>
>>>>
>>>>
>>>> I think this still sets a wrong incentive, as in “you need to have a
>>>> writable GOT for full forwards compatibility with future library
>>>> evolution”.
>>>>
>>>> I think if we detect an IFUNC-based relocation against a
>>>> not-yet-fully-relocated object, we need to delay this relocation
>>>> processing
>>>> and perform a second pass of all delayed relocations. I have a patch
>>>> which
>>>> needs a little bit of work, but looks quite promising (to me at least).
>>>
>>>
>>>
>>> Here are my patches:
>>>
>>> https://sourceware.org/ml/libc-alpha/2018-06/msg00076.html
>>> https://sourceware.org/ml/libc-alpha/2018-06/msg00077.html
>>>
>>> I hope they address your needs.
>>
>>
>> Do you have a git branch I can use?
>
>
> I've pushed it to fw/bug21242.
>
I pushed hjl/pr23240/fw, which is fw/bug21242 + tests for [BZ #23176] and
[BZ #23240].
1. On x86-64:
FAIL: elf/reldep6a
2. On i686:
FAIL: elf/ifuncpreload1
FAIL: elf/reldep6a
So your patches didn't fix:
1,. IFUNC on i686.
2. Weak reference.
On 06/06/2018 05:38 PM, H.J. Lu wrote:
> I pushed hjl/pr23240/fw, which is fw/bug21242 + tests for [BZ #23176] and
> [BZ #23240].
>
> 1. On x86-64:
>
> FAIL: elf/reldep6a
The crash happens because “weak” is undefined in reldep6mod1.so.
The “weak” function is defined in reldep6mod4.so. But reldep6mod1.so is
loaded before reldep6mod4.so:
14289: calling init:
/home/fweimer/src/gnu/glibc/build/elf/reldep6mod1.so
…
14289: opening
file=/home/fweimer/src/gnu/glibc/build/elf/reldep6mod4.so [0];
direct_opencount=1
14289:
14289: binding file
/home/fweimer/src/gnu/glibc/build/elf/reldep6mod4.so [0] to
/home/fweimer/src/gnu/glibc/build/elf/reldep6mod1.so [0]: normal symbol
`baz'
I think this is simply invalid. I think that if you had a “weak ==
NULL” check in reldep6mod1.so, it would be false even after loading
reldep6mod4.so because the address is not lazily bound.
Is this derived from the X server use case, where they load additional
DSOs which provide definitions of functions which are lazily bound?
That's just not compatible with BIND_NOW, and I don't see a compelling
reason why it should be. If it's just about fixing X, then we need to
fix X, and not enhance the dynamic linker.
> 2. On i686:
>
> FAIL: elf/ifuncpreload1
> FAIL: elf/reldep6a
There's no delayed processing for i686 yet. I wanted to get consensus
on the overall approach first.
Thanks,
Florian
On Wed, Jun 6, 2018 at 8:53 AM, Florian Weimer <fweimer@redhat.com> wrote:
> On 06/06/2018 05:38 PM, H.J. Lu wrote:
>>
>> I pushed hjl/pr23240/fw, which is fw/bug21242 + tests for [BZ #23176] and
>> [BZ #23240].
>>
>> 1. On x86-64:
>>
>> FAIL: elf/reldep6a
>
>
> The crash happens because “weak” is undefined in reldep6mod1.so.
>
> The “weak” function is defined in reldep6mod4.so. But reldep6mod1.so is
> loaded before reldep6mod4.so:
>
> 14289: calling init:
> /home/fweimer/src/gnu/glibc/build/elf/reldep6mod1.so
> …
> 14289: opening
> file=/home/fweimer/src/gnu/glibc/build/elf/reldep6mod4.so [0];
> direct_opencount=1
> 14289:
> 14289: binding file
> /home/fweimer/src/gnu/glibc/build/elf/reldep6mod4.so [0] to
> /home/fweimer/src/gnu/glibc/build/elf/reldep6mod1.so [0]: normal symbol
> `baz'
>
> I think this is simply invalid. I think that if you had a “weak == NULL”
> check in reldep6mod1.so, it would be false even after loading reldep6mod4.so
> because the address is not lazily bound.
>
> Is this derived from the X server use case, where they load additional DSOs
No idea. It is in elf.
> which provide definitions of functions which are lazily bound? That's just
> not compatible with BIND_NOW, and I don't see a compelling reason why it
> should be. If it's just about fixing X, then we need to fix X, and not
> enhance the dynamic linker.
>
>> 2. On i686:
>>
>> FAIL: elf/ifuncpreload1
>> FAIL: elf/reldep6a
>
>
> There's no delayed processing for i686 yet. I wanted to get consensus on
> the overall approach first.
>
> Thanks,
> Florian
On 06/06/2018 06:10 PM, H.J. Lu wrote:
>> Is this derived from the X server use case, where they load additional DSOs
>
> No idea. It is in elf.
So you have just run the elf testsuite with LD_BIND_NOW=1 and filed a
bug because the test failed?
I'm not sure if such purely synthetic test cases make sense. If you
want to get rid of lazy binding, you should concentrate on actual issues
in the GNU system (such as the X issue I mentioned).
Thanks,
Florian
From 8c7faa86f7c920c530a56100d7e2bbbe9c5b62e9 Mon Sep 17 00:00:00 2001
From: "H.J. Lu" <hjl.tools@gmail.com>
Date: Sun, 27 May 2018 07:15:35 -0700
Subject: [PATCH] x86: Fall back to lazy binding if possible
Since we can't call unrelocated IFUNC function to get the real function
addresss, this patch falls back to lazy binding for unrelocated IFUNC
symbol if lazy binding is possible. Also in case of unresolved weak
reference in shared object which may be resolved later with lazy binding,
this patch sets up lazy binding if lazy binding is possible.
Tested on x86-64 and i686 with and without --enable-bind-now.
[BZ #23176]
[BZ #23240]
* elf/Makefile (tests): Add reldep6a.
(tests-internal): Add ifuncpreload1
(modules-names): Add ifuncpreloadmod1a and ifuncpreloadmod1b.
($(objpfx)reldep6a): New.
($(objpfx)reldep6a.out): Likewise.
(reldep6a-ENV): Likewise.
($(objpfx)ifuncpreload1): Likewise.
($(objpfx)ifuncpreload1.out): Likewise.
(ifuncpreload1-ENV): Likewise.
* elf/ifuncpreload1.c: New file.
* elf/ifuncpreloadmod1a.c: Likewise.
* elf/ifuncpreloadmod1b.c: Likewise.
* elf/reldep6a.c: Likewise.
* sysdeps/i386/dl-machine.h (elf_machine_runtime_setup): Always
set up GOT.
(elf_machine_rel): Fall back to lazy binding for unrelocated
IFUNC symbol and unresolved weak reference if possible. Issue
IFUNC warning only for LD_DEBUG=bindings.
* sysdeps/x86_64/dl-machine.h (elf_machine_runtime_setup): Always
set up GOT.
(elf_machine_rela): Fall back to lazy binding for unrelocated
IFUNC symbol and unresolved weak reference in shared object if
possible. Issue IFUNC warning only for LD_DEBUG=bindings.
---
elf/Makefile | 18 ++++++++++---
elf/ifuncpreload1.c | 39 ++++++++++++++++++++++++++++
elf/ifuncpreloadmod1a.c | 23 ++++++++++++++++
elf/ifuncpreloadmod1b.c | 49 ++++++++++++++++++++++++++++++++++
elf/reldep6a.c | 1 +
sysdeps/i386/dl-machine.h | 52 +++++++++++++++++++++++++++++++------
sysdeps/x86_64/dl-machine.h | 52 +++++++++++++++++++++++++++++++------
7 files changed, 214 insertions(+), 20 deletions(-)
create mode 100644 elf/ifuncpreload1.c
create mode 100644 elf/ifuncpreloadmod1a.c
create mode 100644 elf/ifuncpreloadmod1b.c
create mode 100644 elf/reldep6a.c
@@ -170,8 +170,8 @@ tests += restest1 preloadtest loadfail multiload origtest resolvfail \
constload1 order noload filter \
reldep reldep2 reldep3 reldep4 nodelete nodelete2 \
nodlopen nodlopen2 lateglobal initfirst global \
- restest2 next dblload dblunload reldep5 reldep6 reldep7 reldep8 \
- tst-tls4 tst-tls5 \
+ restest2 next dblload dblunload reldep5 reldep6 reldep6a \
+ reldep7 reldep8 tst-tls4 tst-tls5 \
tst-tls10 tst-tls11 tst-tls12 tst-tls13 tst-tls14 tst-tls15 \
tst-tls16 tst-tls17 tst-tls18 tst-tls19 tst-tls-dlinfo \
tst-align tst-align2 \
@@ -328,7 +328,7 @@ tests-internal += \
ifuncmain1staticpic \
ifuncmain2 ifuncmain2pic ifuncmain3 ifuncmain4 \
ifuncmain5 ifuncmain5pic ifuncmain5staticpic \
- ifuncmain7 ifuncmain7pic
+ ifuncmain7 ifuncmain7pic ifuncpreload1
ifunc-test-modules = ifuncdep1 ifuncdep1pic ifuncdep2 ifuncdep2pic \
ifuncdep5 ifuncdep5pic
extra-test-objs += $(ifunc-test-modules:=.o)
@@ -339,7 +339,8 @@ ifunc-pie-tests = ifuncmain1pie ifuncmain1vispie ifuncmain1staticpie \
tests-internal += $(ifunc-pie-tests)
tests-pie += $(ifunc-pie-tests)
endif
-modules-names += ifuncmod1 ifuncmod3 ifuncmod5 ifuncmod6
+modules-names += ifuncmod1 ifuncmod3 ifuncmod5 ifuncmod6 \
+ ifuncpreloadmod1a ifuncpreloadmod1b
endif
endif
@@ -885,6 +886,10 @@ $(objpfx)reldep5.out: $(objpfx)reldepmod5.so $(objpfx)reldepmod6.so
$(objpfx)reldep6: $(libdl)
$(objpfx)reldep6.out: $(objpfx)reldep6mod3.so $(objpfx)reldep6mod4.so
+$(objpfx)reldep6a: $(libdl)
+$(objpfx)reldep6a.out: $(objpfx)reldep6mod3.so $(objpfx)reldep6mod4.so
+reldep6a-ENV = LD_BIND_NOW=1
+
$(objpfx)reldep7: $(libdl)
$(objpfx)reldep7.out: $(objpfx)reldep7mod1.so $(objpfx)reldep7mod2.so
@@ -1454,3 +1459,8 @@ tst-libc_dlvsym-static-ENV = \
$(objpfx)tst-libc_dlvsym-static.out: $(objpfx)tst-libc_dlvsym-dso.so
$(objpfx)tst-big-note: $(objpfx)tst-big-note-lib.so
+
+$(objpfx)ifuncpreload1: $(objpfx)ifuncpreloadmod1a.so
+$(objpfx)ifuncpreload1.out: $(objpfx)ifuncpreloadmod1b.so
+ifuncpreload1-ENV = \
+ LD_PRELOAD=$(objpfx)ifuncpreloadmod1b.so LD_BIND_NOW=1
new file mode 100644
@@ -0,0 +1,39 @@
+/* Test for relocation over with IFUNC symbols.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <support/check.h>
+#include <support/support.h>
+#include <support/test-driver.h>
+
+extern void bar (char *, const char *, unsigned int);
+
+static int
+do_test (void)
+{
+ char dst[50];
+ const char src[] =
+ {
+ "This is a test"
+ };
+ bar (dst, src, sizeof (src));
+ if (__builtin_memcmp (dst, src, sizeof (src)) != 0)
+ __builtin_abort ();
+ return 0;
+}
+
+#include <support/test-driver.c>
new file mode 100644
@@ -0,0 +1,23 @@
+/* Shared module to test for relocation over with IFUNC symbols.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+void
+bar (char *dst, const char *src, unsigned int size)
+{
+ __builtin_memmove (dst, src, size);
+}
new file mode 100644
@@ -0,0 +1,49 @@
+/* Shared module to test for relocation over with IFUNC symbols.
+ Copyright (C) 2018 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+#include <stddef.h>
+
+void *
+my_memmove(void *dst_p, const void *src_p, size_t n)
+{
+ const char *src = src_p;
+ char *dst = dst_p;
+ char *ret = dst;
+ if (src < dst)
+ {
+ dst += n;
+ src += n;
+ while (n--)
+ *--dst = *--src;
+ }
+ else
+ while (n--)
+ *dst++ = *src++;
+ return ret;
+}
+
+void *memmove (void *, const void *, size_t)
+ __attribute__ ((ifunc ("resolve_memmove")));
+
+typedef void *(*memmove_t) (void *, const void *, size_t);
+
+static memmove_t
+resolve_memmove (void)
+{
+ return my_memmove;
+}
new file mode 100644
@@ -0,0 +1 @@
+#include "reldep6.c"
@@ -68,7 +68,12 @@ elf_machine_runtime_setup (struct link_map *l, int lazy, int profile)
extern void _dl_runtime_resolve (Elf32_Word) attribute_hidden;
extern void _dl_runtime_profile (Elf32_Word) attribute_hidden;
- if (l->l_info[DT_JMPREL] && lazy)
+ /* Set up GOT if lazy binding is allowed since we may have to fall
+ back to lazy binding even with non-lazy binding is requested. */
+ if (l->l_info[DT_JMPREL]
+ && (lazy
+ || l->l_relro_size == 0
+ || l->l_info[DT_BIND_NOW] == NULL))
{
/* The GOT entries for functions in the PLT have not yet been filled
in. Their initial contents will arrange when called to push an
@@ -332,16 +337,31 @@ elf_machine_rel (struct link_map *map, const Elf32_Rel *reloc,
&& sym_map->l_type != lt_executable
&& !sym_map->l_relocated)
{
- const char *strtab
- = (const char *) D_PTR (map, l_info[DT_STRTAB]);
- _dl_error_printf ("\
+ /* NB: The symbol reference is resolved to IFUNC symbol
+ from an unrelocated shared object. Relocate the GOT
+ entry to enable lazy binding. Issue a warning if lazy
+ binding isn't possible or GOT will be made read-only
+ after shared objects have been relocated. */
+ if (r_type == R_386_JMP_SLOT
+ && ((Elf32_Addr *) D_PTR (map,
+ l_info[DT_PLTGOT]))[2] != 0)
+ value = map->l_addr + *reloc_addr;
+ else if (__glibc_unlikely (GLRO(dl_debug_mask)
+ & (DL_DEBUG_BINDINGS|DL_DEBUG_PRELINK)))
+ {
+ const char *strtab
+ = (const char *) D_PTR (map, l_info[DT_STRTAB]);
+ _dl_error_printf ("\
%s: Relink `%s' with `%s' for IFUNC symbol `%s'\n",
- RTLD_PROGNAME, map->l_name,
- sym_map->l_name,
- strtab + refsym->st_name);
+ RTLD_PROGNAME, map->l_name,
+ sym_map->l_name,
+ strtab + refsym->st_name);
+ value = ((Elf32_Addr (*) (void)) value) ();
+ }
}
+ else
# endif
- value = ((Elf32_Addr (*) (void)) value) ();
+ value = ((Elf32_Addr (*) (void)) value) ();
}
switch (r_type)
@@ -353,7 +373,23 @@ elf_machine_rel (struct link_map *map, const Elf32_Rel *reloc,
break;
# endif
case R_386_GLOB_DAT:
+# ifdef RTLD_BOOTSTRAP
+ case R_386_JMP_SLOT:
+# endif
+ *reloc_addr = value;
+ break;
+# ifndef RTLD_BOOTSTRAP
case R_386_JMP_SLOT:
+ /* NB: If a weak symbol reference in shared object isn't
+ resolved, relocate the GOT entry to enable lazy binding
+ if possible. */
+ if (sym_map == NULL
+ && map->l_type != lt_executable
+ && ELF32_ST_BIND (refsym->st_info) == STB_WEAK
+ && ((Elf32_Addr *) D_PTR (map,
+ l_info[DT_PLTGOT]))[2] != 0)
+ value = map->l_addr + *reloc_addr;
+# endif
*reloc_addr = value;
break;
@@ -73,7 +73,12 @@ elf_machine_runtime_setup (struct link_map *l, int lazy, int profile)
extern void _dl_runtime_profile_avx (ElfW(Word)) attribute_hidden;
extern void _dl_runtime_profile_avx512 (ElfW(Word)) attribute_hidden;
- if (l->l_info[DT_JMPREL] && lazy)
+ /* Set up GOT if lazy binding is allowed since we may have to fall
+ back to lazy binding even with non-lazy binding is requested. */
+ if (l->l_info[DT_JMPREL]
+ && (lazy
+ || l->l_relro_size == 0
+ || l->l_info[DT_BIND_NOW] == NULL))
{
/* The GOT entries for functions in the PLT have not yet been filled
in. Their initial contents will arrange when called to push an
@@ -318,16 +323,31 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
&& sym_map->l_type != lt_executable
&& !sym_map->l_relocated)
{
- const char *strtab
- = (const char *) D_PTR (map, l_info[DT_STRTAB]);
- _dl_error_printf ("\
+ /* NB: The symbol reference is resolved to IFUNC symbol
+ from an unrelocated shared object. Relocate the GOT
+ entry to enable lazy binding. Issue a warning if lazy
+ binding isn't possible or GOT will be made read-only
+ after shared objects have been relocated. */
+ if (r_type == R_X86_64_JUMP_SLOT
+ && ((ElfW(Addr) *) D_PTR (map,
+ l_info[DT_PLTGOT]))[2] != 0)
+ value = map->l_addr + *reloc_addr;
+ else if (__glibc_unlikely (GLRO(dl_debug_mask)
+ & (DL_DEBUG_BINDINGS|DL_DEBUG_PRELINK)))
+ {
+ const char *strtab
+ = (const char *) D_PTR (map, l_info[DT_STRTAB]);
+ _dl_error_printf ("\
%s: Relink `%s' with `%s' for IFUNC symbol `%s'\n",
- RTLD_PROGNAME, map->l_name,
- sym_map->l_name,
- strtab + refsym->st_name);
+ RTLD_PROGNAME, map->l_name,
+ sym_map->l_name,
+ strtab + refsym->st_name);
+ value = ((ElfW(Addr) (*) (void)) value) ();
+ }
}
+ else
# endif
- value = ((ElfW(Addr) (*) (void)) value) ();
+ value = ((ElfW(Addr) (*) (void)) value) ();
}
switch (r_type)
@@ -348,7 +368,23 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
value = sym->st_size;
# endif
case R_X86_64_GLOB_DAT:
+# ifdef RTLD_BOOTSTRAP
+ case R_X86_64_JUMP_SLOT:
+# endif
+ *reloc_addr = value + reloc->r_addend;
+ break;
+# ifndef RTLD_BOOTSTRAP
case R_X86_64_JUMP_SLOT:
+ /* NB: If a weak symbol reference in shared object isn't
+ resolved, relocate the GOT entry to enable lazy binding
+ if possible. */
+ if (sym_map == NULL
+ && map->l_type != lt_executable
+ && ELFW(ST_BIND) (refsym->st_info) == STB_WEAK
+ && ((ElfW(Addr) *) D_PTR (map,
+ l_info[DT_PLTGOT]))[2] != 0)
+ value = map->l_addr + *reloc_addr;
+# endif
*reloc_addr = value + reloc->r_addend;
break;
--
2.17.0