elf: Avoid nested functions in the loader (all ports) [BZ #27220]

Message ID 20210823043648.2648608-1-maskray@google.com
State Superseded
Headers
Series elf: Avoid nested functions in the loader (all ports) [BZ #27220] |

Checks

Context Check Description
dj/TryBot-apply_patch success Patch applied to master at the time it was sent
dj/TryBot-32bit success Build for i686

Commit Message

Fangrui Song Aug. 23, 2021, 4:36 a.m. UTC
  [Alternative to https://sourceware.org/pipermail/libc-alpha/2021-August/130340.html
This version fixes all ports and doesn't add NESTING dispatches.]

dynamic-link.h is included more than once in some elf/ files (rtld.c,
dl-conflict.c, dl-reloc.c, dl-reloc-static-pie.c) and uses GCC nested
functions. This harms readability and the nested functions usage
is the biggest obstacle prevents CC=clang (which doesn't support the
feature).

To un-nest elf_machine_rela, the key idea is to pass the variable in
the containing scope as an extra argument.
Stan Shebs implemented ppc64/x86-64 parts in the google/grte/v5-2.27/master branch.
This patch squashes and cleans up the commits, and fixes all ports.

Tested on aarch64-linux-gnu and x86_64-linux-gnu.
On x86_64, ld.so size is slightly larger (0.25% .text size).
The performance is unrelated because the very few extra instructions in
RESOLVE_MAP are negligible compared with expensive symbol lookup costs.
On aarch64, ld.so size is slightly smaller. The nested functions appear
to need more .rodata and .eh_frame space.

Tested build-many-glibcs.py with
{alpha,arc,csky,hppa,ia64,microblaze,s390,s390x,sh,sparc64,sparcv9}-linux-gnu,
arm-linux-gnueabi{,hf}, mips64{,el}-linux-gnu{,-n32,-n64}, and
riscv64-linux-gnu-rv64imafdc-lp64d.
---
 elf/dl-conflict.c                      | 35 +++++++-------
 elf/dl-reloc-static-pie.c              | 12 ++---
 elf/dl-reloc.c                         | 66 +++++++++++++++-----------
 elf/do-rel.h                           | 12 ++---
 elf/dynamic-link.h                     | 40 ++++++++--------
 elf/get-dynamic-info.h                 |  5 ++
 elf/rtld.c                             | 10 ++--
 sysdeps/aarch64/dl-machine.h           | 16 +++++--
 sysdeps/alpha/dl-machine.h             | 13 +++--
 sysdeps/arc/dl-machine.h               |  8 +++-
 sysdeps/arm/dl-machine.h               | 22 +++++----
 sysdeps/csky/dl-machine.h              | 12 +++--
 sysdeps/hppa/dl-machine.h              | 10 ++--
 sysdeps/i386/dl-machine.h              | 22 ++++-----
 sysdeps/ia64/dl-machine.h              | 16 ++++---
 sysdeps/m68k/dl-machine.h              | 12 +++--
 sysdeps/microblaze/dl-machine.h        | 12 +++--
 sysdeps/mips/dl-machine.h              | 45 +++++++++++-------
 sysdeps/nios2/dl-machine.h             | 14 ++++--
 sysdeps/powerpc/powerpc32/dl-machine.h | 12 +++--
 sysdeps/powerpc/powerpc64/dl-machine.h | 16 ++++---
 sysdeps/riscv/dl-machine.h             | 10 ++--
 sysdeps/s390/s390-32/dl-machine.h      |  8 ++--
 sysdeps/s390/s390-64/dl-machine.h      |  8 ++--
 sysdeps/sh/dl-machine.h                | 12 +++--
 sysdeps/sparc/sparc32/dl-machine.h     | 12 +++--
 sysdeps/sparc/sparc64/dl-machine.h     | 12 +++--
 sysdeps/x86_64/dl-machine.h            | 16 +++++--
 28 files changed, 291 insertions(+), 197 deletions(-)
  

Comments

Fangrui Song Sept. 3, 2021, 5:41 a.m. UTC | #1
Ping https://sourceware.org/pipermail/libc-alpha/2021-August/130391.html

On Sun, Aug 22, 2021 at 9:36 PM Fangrui Song <maskray@google.com> wrote:
>
> [Alternative to https://sourceware.org/pipermail/libc-alpha/2021-August/130340.html
> This version fixes all ports and doesn't add NESTING dispatches.]
>
> dynamic-link.h is included more than once in some elf/ files (rtld.c,
> dl-conflict.c, dl-reloc.c, dl-reloc-static-pie.c) and uses GCC nested
> functions. This harms readability and the nested functions usage
> is the biggest obstacle prevents CC=clang (which doesn't support the
> feature).
>
> To un-nest elf_machine_rela, the key idea is to pass the variable in
> the containing scope as an extra argument.
> Stan Shebs implemented ppc64/x86-64 parts in the google/grte/v5-2.27/master branch.
> This patch squashes and cleans up the commits, and fixes all ports.
>
> Tested on aarch64-linux-gnu and x86_64-linux-gnu.
> On x86_64, ld.so size is slightly larger (0.25% .text size).
> The performance is unrelated because the very few extra instructions in
> RESOLVE_MAP are negligible compared with expensive symbol lookup costs.
> On aarch64, ld.so size is slightly smaller. The nested functions appear
> to need more .rodata and .eh_frame space.
>
> Tested build-many-glibcs.py with
> {alpha,arc,csky,hppa,ia64,microblaze,s390,s390x,sh,sparc64,sparcv9}-linux-gnu,
> arm-linux-gnueabi{,hf}, mips64{,el}-linux-gnu{,-n32,-n64}, and
> riscv64-linux-gnu-rv64imafdc-lp64d.
> ---
>  elf/dl-conflict.c                      | 35 +++++++-------
>  elf/dl-reloc-static-pie.c              | 12 ++---
>  elf/dl-reloc.c                         | 66 +++++++++++++++-----------
>  elf/do-rel.h                           | 12 ++---
>  elf/dynamic-link.h                     | 40 ++++++++--------
>  elf/get-dynamic-info.h                 |  5 ++
>  elf/rtld.c                             | 10 ++--
>  sysdeps/aarch64/dl-machine.h           | 16 +++++--
>  sysdeps/alpha/dl-machine.h             | 13 +++--
>  sysdeps/arc/dl-machine.h               |  8 +++-
>  sysdeps/arm/dl-machine.h               | 22 +++++----
>  sysdeps/csky/dl-machine.h              | 12 +++--
>  sysdeps/hppa/dl-machine.h              | 10 ++--
>  sysdeps/i386/dl-machine.h              | 22 ++++-----
>  sysdeps/ia64/dl-machine.h              | 16 ++++---
>  sysdeps/m68k/dl-machine.h              | 12 +++--
>  sysdeps/microblaze/dl-machine.h        | 12 +++--
>  sysdeps/mips/dl-machine.h              | 45 +++++++++++-------
>  sysdeps/nios2/dl-machine.h             | 14 ++++--
>  sysdeps/powerpc/powerpc32/dl-machine.h | 12 +++--
>  sysdeps/powerpc/powerpc64/dl-machine.h | 16 ++++---
>  sysdeps/riscv/dl-machine.h             | 10 ++--
>  sysdeps/s390/s390-32/dl-machine.h      |  8 ++--
>  sysdeps/s390/s390-64/dl-machine.h      |  8 ++--
>  sysdeps/sh/dl-machine.h                | 12 +++--
>  sysdeps/sparc/sparc32/dl-machine.h     | 12 +++--
>  sysdeps/sparc/sparc64/dl-machine.h     | 12 +++--
>  sysdeps/x86_64/dl-machine.h            | 16 +++++--
>  28 files changed, 291 insertions(+), 197 deletions(-)
>
> diff --git a/elf/dl-conflict.c b/elf/dl-conflict.c
> index 31a2f90770..d7219b726a 100644
> --- a/elf/dl-conflict.c
> +++ b/elf/dl-conflict.c
> @@ -27,17 +27,9 @@
>  #include <sys/types.h>
>  #include "dynamic-link.h"
>
> -void
> -_dl_resolve_conflicts (struct link_map *l, ElfW(Rela) *conflict,
> -                      ElfW(Rela) *conflictend)
> -{
> -#if ! ELF_MACHINE_NO_RELA
> -  if (__glibc_unlikely (GLRO(dl_debug_mask) & DL_DEBUG_RELOC))
> -    _dl_debug_printf ("\nconflict processing: %s\n", DSO_FILENAME (l->l_name));
> -
> -  {
> -    /* Do the conflict relocation of the object and library GOT and other
> -       data.  */
> +/* Used by RESOLVE_CONFLICT_FIND_MAP at init time. Cannot be accessed
> + * concurrently.  */
> +static struct link_map *resolve_conflict_map __attribute__ ((__unused__));
>
>      /* This macro is used as a callback from the ELF_DYNAMIC_RELOCATE code.  */
>  #define RESOLVE_MAP(ref, version, flags) (*ref = NULL, NULL)
> @@ -51,12 +43,23 @@ _dl_resolve_conflicts (struct link_map *l, ElfW(Rela) *conflict,
>      (map) = resolve_conflict_map;                                            \
>    } while (0)
>
> +#include "dynamic-link.h"
> +
> +void
> +_dl_resolve_conflicts (struct link_map *l, ElfW(Rela) *conflict,
> +                      ElfW(Rela) *conflictend)
> +{
> +#if ! ELF_MACHINE_NO_RELA
> +  if (__glibc_unlikely (GLRO(dl_debug_mask) & DL_DEBUG_RELOC))
> +    _dl_debug_printf ("\nconflict processing: %s\n", DSO_FILENAME (l->l_name));
> +
> +  {
> +    /* Do the conflict relocation of the object and library GOT and other
> +       data.  */
> +
>      /* Prelinking makes no sense for anything but the main namespace.  */
>      assert (l->l_ns == LM_ID_BASE);
> -    struct link_map *resolve_conflict_map __attribute__ ((__unused__))
> -      = GL(dl_ns)[LM_ID_BASE]._ns_loaded;
> -
> -#include "dynamic-link.h"
> +    resolve_conflict_map = GL(dl_ns)[LM_ID_BASE]._ns_loaded;
>
>      /* Override these, defined in dynamic-link.h.  */
>  #undef CHECK_STATIC_TLS
> @@ -68,7 +71,7 @@ _dl_resolve_conflicts (struct link_map *l, ElfW(Rela) *conflict,
>
>      for (; conflict < conflictend; ++conflict)
>        elf_machine_rela (l, conflict, NULL, NULL, (void *) conflict->r_offset,
> -                       0);
> +                       0, NULL);
>    }
>  #endif
>  }
> diff --git a/elf/dl-reloc-static-pie.c b/elf/dl-reloc-static-pie.c
> index d5bd2f31e9..25d1c53aa2 100644
> --- a/elf/dl-reloc-static-pie.c
> +++ b/elf/dl-reloc-static-pie.c
> @@ -23,6 +23,11 @@
>  #include <ldsodefs.h>
>  #include "dynamic-link.h"
>
> +#define STATIC_PIE_BOOTSTRAP
> +#define BOOTSTRAP_MAP (main_map)
> +#define RESOLVE_MAP(sym, version, flags) BOOTSTRAP_MAP
> +#include "dynamic-link.h"
> +
>  /* Relocate static executable with PIE.  */
>
>  void
> @@ -30,11 +35,6 @@ _dl_relocate_static_pie (void)
>  {
>    struct link_map *main_map = _dl_get_dl_main_map ();
>
> -# define STATIC_PIE_BOOTSTRAP
> -# define BOOTSTRAP_MAP (main_map)
> -# define RESOLVE_MAP(sym, version, flags) BOOTSTRAP_MAP
> -# include "dynamic-link.h"
> -
>    /* Figure out the run-time load address of static PIE.  */
>    main_map->l_addr = elf_machine_load_address ();
>
> @@ -48,7 +48,7 @@ _dl_relocate_static_pie (void)
>
>    /* Relocate ourselves so we can do normal function calls and
>       data access using the global offset table.  */
> -  ELF_DYNAMIC_RELOCATE (main_map, 0, 0, 0);
> +  ELF_DYNAMIC_RELOCATE (main_map, 0, 0, 0, main_map);
>    main_map->l_relocated = 1;
>
>    /* Initialize _r_debug.  */
> diff --git a/elf/dl-reloc.c b/elf/dl-reloc.c
> index e13a672ade..35cab08683 100644
> --- a/elf/dl-reloc.c
> +++ b/elf/dl-reloc.c
> @@ -162,6 +162,39 @@ _dl_nothread_init_static_tls (struct link_map *map)
>  }
>  #endif /* !THREAD_GSCOPE_IN_TCB */
>
> +/* Used by RESOLVE_MAP. _dl_relocate_object is either called at init time or
> + * by dlopen with a global lock, so the variables cannot be accessed
> + * concurrently.  */
> +static struct link_map *cur_l;
> +static struct r_scope_elem **cur_scope;
> +static const char *cur_strtab;
> +
> +/* This macro is used as a callback from the ELF_DYNAMIC_RELOCATE code.  */
> +#define RESOLVE_MAP(ref, version, r_type) \
> +    ({ struct link_map *l = cur_l;                                           \
> +      (ELFW(ST_BIND) ((*ref)->st_info) != STB_LOCAL                          \
> +      && __glibc_likely (!dl_symbol_visibility_binds_local_p (*ref)))        \
> +     ? ((__builtin_expect ((*ref) == l->l_lookup_cache.sym, 0)               \
> +        && elf_machine_type_class (r_type) == l->l_lookup_cache.type_class)  \
> +       ? (bump_num_cache_relocations (),                                     \
> +          (*ref) = l->l_lookup_cache.ret,                                    \
> +          l->l_lookup_cache.value)                                           \
> +       : ({ lookup_t _lr;                                                    \
> +            int _tc = elf_machine_type_class (r_type);                       \
> +            l->l_lookup_cache.type_class = _tc;                              \
> +            l->l_lookup_cache.sym = (*ref);                                  \
> +            const struct r_found_version *v = NULL;                          \
> +            if ((version) != NULL && (version)->hash != 0)                   \
> +              v = (version);                                                 \
> +            _lr = _dl_lookup_symbol_x (cur_strtab + (*ref)->st_name, l,      \
> +                                       (ref), cur_scope, v, _tc,             \
> +                                       DL_LOOKUP_ADD_DEPENDENCY, NULL);      \
> +            l->l_lookup_cache.ret = (*ref);                                  \
> +            l->l_lookup_cache.value = _lr; }))                               \
> +     : l; })
> +
> +#include "dynamic-link.h"
> +
>  void
>  _dl_relocate_object (struct link_map *l, struct r_scope_elem *scope[],
>                      int reloc_mode, int consider_profiling)
> @@ -243,36 +276,11 @@ _dl_relocate_object (struct link_map *l, struct r_scope_elem *scope[],
>    {
>      /* Do the actual relocation of the object's GOT and other data.  */
>
> -    /* String table object symbols.  */
> -    const char *strtab = (const void *) D_PTR (l, l_info[DT_STRTAB]);
> +    cur_l = l;
> +    cur_scope = scope;
> +    cur_strtab = (const void *) D_PTR (cur_l, l_info[DT_STRTAB]);
>
> -    /* This macro is used as a callback from the ELF_DYNAMIC_RELOCATE code.  */
> -#define RESOLVE_MAP(ref, version, r_type) \
> -    ((ELFW(ST_BIND) ((*ref)->st_info) != STB_LOCAL                           \
> -      && __glibc_likely (!dl_symbol_visibility_binds_local_p (*ref)))        \
> -     ? ((__builtin_expect ((*ref) == l->l_lookup_cache.sym, 0)               \
> -        && elf_machine_type_class (r_type) == l->l_lookup_cache.type_class)  \
> -       ? (bump_num_cache_relocations (),                                     \
> -          (*ref) = l->l_lookup_cache.ret,                                    \
> -          l->l_lookup_cache.value)                                           \
> -       : ({ lookup_t _lr;                                                    \
> -            int _tc = elf_machine_type_class (r_type);                       \
> -            l->l_lookup_cache.type_class = _tc;                              \
> -            l->l_lookup_cache.sym = (*ref);                                  \
> -            const struct r_found_version *v = NULL;                          \
> -            if ((version) != NULL && (version)->hash != 0)                   \
> -              v = (version);                                                 \
> -            _lr = _dl_lookup_symbol_x (strtab + (*ref)->st_name, l, (ref),   \
> -                                       scope, v, _tc,                        \
> -                                       DL_LOOKUP_ADD_DEPENDENCY              \
> -                                       | DL_LOOKUP_FOR_RELOCATE, NULL);      \
> -            l->l_lookup_cache.ret = (*ref);                                  \
> -            l->l_lookup_cache.value = _lr; }))                               \
> -     : l)
> -
> -#include "dynamic-link.h"
> -
> -    ELF_DYNAMIC_RELOCATE (l, lazy, consider_profiling, skip_ifunc);
> +    ELF_DYNAMIC_RELOCATE (l, lazy, consider_profiling, skip_ifunc, NULL);
>
>  #ifndef PROF
>      if (__glibc_unlikely (consider_profiling)
> diff --git a/elf/do-rel.h b/elf/do-rel.h
> index 321ac2b359..19fc6c085e 100644
> --- a/elf/do-rel.h
> +++ b/elf/do-rel.h
> @@ -37,11 +37,11 @@
>     relocations; they should be set up to call _dl_runtime_resolve, rather
>     than fully resolved now.  */
>
> -auto inline void __attribute__ ((always_inline))
> +static inline void __attribute__ ((always_inline))
>  elf_dynamic_do_Rel (struct link_map *map,
>                     ElfW(Addr) reladdr, ElfW(Addr) relsize,
>                     __typeof (((ElfW(Dyn) *) 0)->d_un.d_val) nrelative,
> -                   int lazy, int skip_ifunc)
> +                   int lazy, int skip_ifunc, struct link_map *boot_map)
>  {
>    const ElfW(Rel) *r = (const void *) reladdr;
>    const ElfW(Rel) *end = (const void *) (reladdr + relsize);
> @@ -136,7 +136,7 @@ elf_dynamic_do_Rel (struct link_map *map,
>               ElfW(Half) ndx = version[ELFW(R_SYM) (r->r_info)] & 0x7fff;
>               elf_machine_rel (map, r, &symtab[ELFW(R_SYM) (r->r_info)],
>                                &map->l_versions[ndx],
> -                              (void *) (l_addr + r->r_offset), skip_ifunc);
> +                              (void *) (l_addr + r->r_offset), skip_ifunc, boot_map);
>             }
>
>  #if defined ELF_MACHINE_IRELATIVE && !defined RTLD_BOOTSTRAP
> @@ -150,7 +150,7 @@ elf_dynamic_do_Rel (struct link_map *map,
>                                    &symtab[ELFW(R_SYM) (r2->r_info)],
>                                    &map->l_versions[ndx],
>                                    (void *) (l_addr + r2->r_offset),
> -                                  skip_ifunc);
> +                                  skip_ifunc, boot_map);
>                 }
>  #endif
>         }
> @@ -168,7 +168,7 @@ elf_dynamic_do_Rel (struct link_map *map,
>             else
>  # endif
>               elf_machine_rel (map, r, &symtab[ELFW(R_SYM) (r->r_info)], NULL,
> -                              (void *) (l_addr + r->r_offset), skip_ifunc);
> +                              (void *) (l_addr + r->r_offset), skip_ifunc, boot_map);
>
>  # ifdef ELF_MACHINE_IRELATIVE
>           if (r2 != NULL)
> @@ -176,7 +176,7 @@ elf_dynamic_do_Rel (struct link_map *map,
>               if (ELFW(R_TYPE) (r2->r_info) == ELF_MACHINE_IRELATIVE)
>                 elf_machine_rel (map, r2, &symtab[ELFW(R_SYM) (r2->r_info)],
>                                  NULL, (void *) (l_addr + r2->r_offset),
> -                                skip_ifunc);
> +                                skip_ifunc, boot_map);
>  # endif
>         }
>  #endif
> diff --git a/elf/dynamic-link.h b/elf/dynamic-link.h
> index 3eb24ba3a6..6d71427675 100644
> --- a/elf/dynamic-link.h
> +++ b/elf/dynamic-link.h
> @@ -59,30 +59,30 @@ int _dl_try_allocate_static_tls (struct link_map *map, bool optional)
>     copying memory, breaking the very code written to handle the
>     unaligned cases.  */
>  # if ! ELF_MACHINE_NO_REL
> -auto inline void __attribute__((always_inline))
> +static inline void __attribute__((always_inline))
>  elf_machine_rel (struct link_map *map, const ElfW(Rel) *reloc,
>                  const ElfW(Sym) *sym, const struct r_found_version *version,
> -                void *const reloc_addr, int skip_ifunc);
> -auto inline void __attribute__((always_inline))
> +                void *const reloc_addr, int skip_ifunc, struct link_map *boot_map);
> +static inline void __attribute__((always_inline))
>  elf_machine_rel_relative (ElfW(Addr) l_addr, const ElfW(Rel) *reloc,
>                           void *const reloc_addr);
>  # endif
>  # if ! ELF_MACHINE_NO_RELA
> -auto inline void __attribute__((always_inline))
> +static inline void __attribute__((always_inline))
>  elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
>                   const ElfW(Sym) *sym, const struct r_found_version *version,
> -                 void *const reloc_addr, int skip_ifunc);
> -auto inline void __attribute__((always_inline))
> +                 void *const reloc_addr, int skip_ifunc, struct link_map *boot_map);
> +static inline void __attribute__((always_inline))
>  elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
>                            void *const reloc_addr);
>  # endif
>  # if ELF_MACHINE_NO_RELA || defined ELF_MACHINE_PLT_REL
> -auto inline void __attribute__((always_inline))
> +static inline void __attribute__((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       ElfW(Addr) l_addr, const ElfW(Rel) *reloc,
>                       int skip_ifunc);
>  # else
> -auto inline void __attribute__((always_inline))
> +static inline void __attribute__((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
>                       int skip_ifunc);
> @@ -114,7 +114,7 @@ elf_machine_lazy_rel (struct link_map *map,
>     consumes precisely the very end of the DT_REL*, or DT_JMPREL and DT_REL*
>     are completely separate and there is a gap between them.  */
>
> -# define _ELF_DYNAMIC_DO_RELOC(RELOC, reloc, map, do_lazy, skip_ifunc, test_rel) \
> +# define _ELF_DYNAMIC_DO_RELOC(RELOC, reloc, map, do_lazy, skip_ifunc, test_rel, boot_map) \
>    do {                                                                       \
>      struct { ElfW(Addr) start, size;                                         \
>              __typeof (((ElfW(Dyn) *) 0)->d_un.d_val) nrelative; int lazy; }  \
> @@ -153,7 +153,7 @@ elf_machine_lazy_rel (struct link_map *map,
>                                                                               \
>      if (ELF_DURING_STARTUP)                                                  \
>        elf_dynamic_do_##reloc ((map), ranges[0].start, ranges[0].size,        \
> -                             ranges[0].nrelative, 0, skip_ifunc);            \
> +                             ranges[0].nrelative, 0, skip_ifunc, boot_map);  \
>      else                                                                     \
>        {                                                                              \
>         int ranges_index;                                                     \
> @@ -163,7 +163,7 @@ elf_machine_lazy_rel (struct link_map *map,
>                                   ranges[ranges_index].size,                  \
>                                   ranges[ranges_index].nrelative,             \
>                                   ranges[ranges_index].lazy,                  \
> -                                 skip_ifunc);                                \
> +                                 skip_ifunc, boot_map);                      \
>        }                                                                              \
>    } while (0)
>
> @@ -175,29 +175,29 @@ elf_machine_lazy_rel (struct link_map *map,
>
>  # if ! ELF_MACHINE_NO_REL
>  #  include "do-rel.h"
> -#  define ELF_DYNAMIC_DO_REL(map, lazy, skip_ifunc) \
> -  _ELF_DYNAMIC_DO_RELOC (REL, Rel, map, lazy, skip_ifunc, _ELF_CHECK_REL)
> +#  define ELF_DYNAMIC_DO_REL(map, lazy, skip_ifunc, boot_map)                  \
> +  _ELF_DYNAMIC_DO_RELOC (REL, Rel, map, lazy, skip_ifunc, _ELF_CHECK_REL, boot_map)
>  # else
> -#  define ELF_DYNAMIC_DO_REL(map, lazy, skip_ifunc) /* Nothing to do.  */
> +#  define ELF_DYNAMIC_DO_REL(map, lazy, skip_ifunc, boot_map) /* Nothing to do.  */
>  # endif
>
>  # if ! ELF_MACHINE_NO_RELA
>  #  define DO_RELA
>  #  include "do-rel.h"
> -#  define ELF_DYNAMIC_DO_RELA(map, lazy, skip_ifunc) \
> -  _ELF_DYNAMIC_DO_RELOC (RELA, Rela, map, lazy, skip_ifunc, _ELF_CHECK_REL)
> +#  define ELF_DYNAMIC_DO_RELA(map, lazy, skip_ifunc, boot_map)                 \
> +  _ELF_DYNAMIC_DO_RELOC (RELA, Rela, map, lazy, skip_ifunc, _ELF_CHECK_REL, boot_map)
>  # else
> -#  define ELF_DYNAMIC_DO_RELA(map, lazy, skip_ifunc) /* Nothing to do.  */
> +#  define ELF_DYNAMIC_DO_RELA(map, lazy, skip_ifunc, boot_map) /* Nothing to do.  */
>  # endif
>
>  /* This can't just be an inline function because GCC is too dumb
>     to inline functions containing inlines themselves.  */
> -# define ELF_DYNAMIC_RELOCATE(map, lazy, consider_profile, skip_ifunc) \
> +# define ELF_DYNAMIC_RELOCATE(map, lazy, consider_profile, skip_ifunc, boot_map) \
>    do {                                                                       \
>      int edr_lazy = elf_machine_runtime_setup ((map), (lazy),                 \
>                                               (consider_profile));            \
> -    ELF_DYNAMIC_DO_REL ((map), edr_lazy, skip_ifunc);                        \
> -    ELF_DYNAMIC_DO_RELA ((map), edr_lazy, skip_ifunc);                       \
> +    ELF_DYNAMIC_DO_REL ((map), edr_lazy, skip_ifunc, boot_map);                      \
> +    ELF_DYNAMIC_DO_RELA ((map), edr_lazy, skip_ifunc, boot_map);             \
>    } while (0)
>
>  #endif
> diff --git a/elf/get-dynamic-info.h b/elf/get-dynamic-info.h
> index d8ec32377d..5827ae6e24 100644
> --- a/elf/get-dynamic-info.h
> +++ b/elf/get-dynamic-info.h
> @@ -19,6 +19,8 @@
>  /* This file is included multiple times and therefore lacks a header
>     file inclusion guard.  */
>
> +#ifndef _GET_DYNAMIC_INFO_H
> +
>  #include <assert.h>
>  #include <libc-diag.h>
>
> @@ -180,3 +182,6 @@ elf_get_dynamic_info (struct link_map *l, ElfW(Dyn) *temp)
>      info[DT_RPATH] = NULL;
>  #endif
>  }
> +
> +#define _GET_DYNAMIC_INFO_H
> +#endif
> diff --git a/elf/rtld.c b/elf/rtld.c
> index 878e6480f4..c62efa7308 100644
> --- a/elf/rtld.c
> +++ b/elf/rtld.c
> @@ -499,13 +499,9 @@ _dl_start_final (void *arg, struct dl_start_final_info *info)
>    return start_addr;
>  }
>
> -static ElfW(Addr) __attribute_used__
> -_dl_start (void *arg)
> -{
>  #ifdef DONT_USE_BOOTSTRAP_MAP
>  # define bootstrap_map GL(dl_rtld_map)
>  #else
> -  struct dl_start_final_info info;
>  # define bootstrap_map info.l
>  #endif
>
> @@ -518,9 +514,13 @@ _dl_start (void *arg)
>  #define RESOLVE_MAP(sym, version, flags) BOOTSTRAP_MAP
>  #include "dynamic-link.h"
>
> +static ElfW(Addr) __attribute_used__
> +_dl_start (void *arg)
> +{
>  #ifdef DONT_USE_BOOTSTRAP_MAP
>    rtld_timer_start (&start_time);
>  #else
> +  struct dl_start_final_info info;
>    rtld_timer_start (&info.start_time);
>  #endif
>
> @@ -561,7 +561,7 @@ _dl_start (void *arg)
>        /* Relocate ourselves so we can do normal function calls and
>          data access using the global offset table.  */
>
> -      ELF_DYNAMIC_RELOCATE (&bootstrap_map, 0, 0, 0);
> +      ELF_DYNAMIC_RELOCATE (&bootstrap_map, 0, 0, 0, &bootstrap_map);
>      }
>    bootstrap_map.l_relocated = 1;
>
> diff --git a/sysdeps/aarch64/dl-machine.h b/sysdeps/aarch64/dl-machine.h
> index 3e10cb462f..ab9e5c4d4b 100644
> --- a/sysdeps/aarch64/dl-machine.h
> +++ b/sysdeps/aarch64/dl-machine.h
> @@ -237,11 +237,12 @@ elf_machine_plt_value (struct link_map *map,
>
>  #ifdef RESOLVE_MAP
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
>                   const ElfW(Sym) *sym, const struct r_found_version *version,
> -                 void *const reloc_addr_arg, int skip_ifunc)
> +                 void *const reloc_addr_arg, int skip_ifunc,
> +                 struct link_map *boot_map)
>  {
>    ElfW(Addr) *const reloc_addr = reloc_addr_arg;
>    const unsigned int r_type = ELFW (R_TYPE) (reloc->r_info);
> @@ -253,7 +254,11 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
>    else
>      {
>        const ElfW(Sym) *const refsym = sym;
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +      struct link_map *sym_map = boot_map;
> +#else
>        struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>        ElfW(Addr) value = SYMBOL_ADDRESS (sym_map, sym, true);
>
>        if (sym != NULL
> @@ -377,7 +382,7 @@ elf_machine_rela_relative (ElfW(Addr) l_addr,
>    *reloc_addr = l_addr + reloc->r_addend;
>  }
>
> -inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       ElfW(Addr) l_addr,
> @@ -407,7 +412,7 @@ elf_machine_lazy_rel (struct link_map *map,
>                   version = &map->l_versions[vernum[symndx] & 0x7fff];
>                 }
>               elf_machine_rela (map, reloc, sym, version, reloc_addr,
> -                               skip_ifunc);
> +                               skip_ifunc, NULL);
>               return;
>             }
>         }
> @@ -433,7 +438,8 @@ elf_machine_lazy_rel (struct link_map *map,
>
>        /* Always initialize TLS descriptors completely, because lazy
>          initialization requires synchronization at every TLS access.  */
> -      elf_machine_rela (map, reloc, sym, version, reloc_addr, skip_ifunc);
> +      elf_machine_rela (map, reloc, sym, version, reloc_addr, skip_ifunc,
> +                        NULL);
>      }
>    else if (__glibc_unlikely (r_type == AARCH64_R(IRELATIVE)))
>      {
> diff --git a/sysdeps/alpha/dl-machine.h b/sysdeps/alpha/dl-machine.h
> index 2cd2213d9a..dc22baa68a 100644
> --- a/sysdeps/alpha/dl-machine.h
> +++ b/sysdeps/alpha/dl-machine.h
> @@ -361,14 +361,15 @@ elf_machine_plt_value (struct link_map *map, const Elf64_Rela *reloc,
>
>  /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
>     MAP is the object containing the reloc.  */
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela (struct link_map *map,
>                   const Elf64_Rela *reloc,
>                   const Elf64_Sym *sym,
>                   const struct r_found_version *version,
>                   void *const reloc_addr_arg,
> -                 int skip_ifunc)
> +                 int skip_ifunc,
> +                 struct link_map *boot_map)
>  {
>    Elf64_Addr *const reloc_addr = reloc_addr_arg;
>    unsigned long int const r_type = ELF64_R_TYPE (reloc->r_info);
> @@ -411,7 +412,11 @@ elf_machine_rela (struct link_map *map,
>        return;
>    else
>      {
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +      struct link_map *sym_map = boot_map;
> +#else
>        struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>        Elf64_Addr sym_value;
>        Elf64_Addr sym_raw_value;
>
> @@ -489,7 +494,7 @@ elf_machine_rela (struct link_map *map,
>     can be skipped.  */
>  #define ELF_MACHINE_REL_RELATIVE 1
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
>                            void *const reloc_addr_arg)
> @@ -506,7 +511,7 @@ elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
>    memcpy (reloc_addr_arg, &reloc_addr_val, 8);
>  }
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       Elf64_Addr l_addr, const Elf64_Rela *reloc,
> diff --git a/sysdeps/arc/dl-machine.h b/sysdeps/arc/dl-machine.h
> index e6ce7f0ff6..1054530321 100644
> --- a/sysdeps/arc/dl-machine.h
> +++ b/sysdeps/arc/dl-machine.h
> @@ -228,11 +228,11 @@ elf_machine_fixup_plt (struct link_map *map, lookup_t t,
>
>  #ifdef RESOLVE_MAP
>
> -inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
>                    const ElfW(Sym) *sym, const struct r_found_version *version,
> -                  void *const reloc_addr_arg, int skip_ifunc)
> +                  void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
>  {
>    ElfW(Addr) r_info = reloc->r_info;
>    const unsigned long int r_type = ELFW (R_TYPE) (r_info);
> @@ -245,7 +245,11 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
>    else
>      {
>        const ElfW(Sym) *const refsym = sym;
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +      struct link_map *sym_map = boot_map;
> +#else
>        struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>        ElfW(Addr) value = SYMBOL_ADDRESS (sym_map, sym, true);
>
>        switch (r_type)
> diff --git a/sysdeps/arm/dl-machine.h b/sysdeps/arm/dl-machine.h
> index eb13cb8b57..84f84a1965 100644
> --- a/sysdeps/arm/dl-machine.h
> +++ b/sysdeps/arm/dl-machine.h
> @@ -276,7 +276,7 @@ elf_machine_plt_value (struct link_map *map, const Elf32_Rel *reloc,
>
>  #ifdef RESOLVE_MAP
>  /* Handle a PC24 reloc, including the out-of-range case.  */
> -auto void
> +static void
>  relocate_pc24 (struct link_map *map, Elf32_Addr value,
>                 Elf32_Addr *const reloc_addr, Elf32_Sword addend)
>  {
> @@ -330,11 +330,11 @@ relocate_pc24 (struct link_map *map, Elf32_Addr value,
>  /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
>     MAP is the object containing the reloc.  */
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rel (struct link_map *map, const Elf32_Rel *reloc,
>                  const Elf32_Sym *sym, const struct r_found_version *version,
> -                void *const reloc_addr_arg, int skip_ifunc)
> +                void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf32_Addr *const reloc_addr = reloc_addr_arg;
>    const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
> @@ -364,7 +364,11 @@ elf_machine_rel (struct link_map *map, const Elf32_Rel *reloc,
>  #endif
>      {
>        const Elf32_Sym *const refsym = sym;
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +      struct link_map *sym_map = boot_map;
> +#else
>        struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>        Elf32_Addr value = SYMBOL_ADDRESS (sym_map, sym, true);
>
>        if (sym != NULL
> @@ -508,11 +512,11 @@ elf_machine_rel (struct link_map *map, const Elf32_Rel *reloc,
>  }
>
>  # ifndef RTLD_BOOTSTRAP
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>                   const Elf32_Sym *sym, const struct r_found_version *version,
> -                 void *const reloc_addr_arg, int skip_ifunc)
> +                 void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf32_Addr *const reloc_addr = reloc_addr_arg;
>    const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
> @@ -601,7 +605,7 @@ elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>  }
>  # endif
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rel_relative (Elf32_Addr l_addr, const Elf32_Rel *reloc,
>                           void *const reloc_addr_arg)
> @@ -611,7 +615,7 @@ elf_machine_rel_relative (Elf32_Addr l_addr, const Elf32_Rel *reloc,
>  }
>
>  # ifndef RTLD_BOOTSTRAP
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>                            void *const reloc_addr_arg)
> @@ -621,7 +625,7 @@ elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>  }
>  # endif
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       Elf32_Addr l_addr, const Elf32_Rel *reloc,
> @@ -653,7 +657,7 @@ elf_machine_lazy_rel (struct link_map *map,
>
>        /* Always initialize TLS descriptors completely, because lazy
>          initialization requires synchronization at every TLS access.  */
> -      elf_machine_rel (map, reloc, sym, version, reloc_addr, skip_ifunc);
> +      elf_machine_rel (map, reloc, sym, version, reloc_addr, skip_ifunc, NULL);
>      }
>    else
>      _dl_reloc_bad_type (map, r_type, 1);
> diff --git a/sysdeps/csky/dl-machine.h b/sysdeps/csky/dl-machine.h
> index b08f06d74c..8702fe9917 100644
> --- a/sysdeps/csky/dl-machine.h
> +++ b/sysdeps/csky/dl-machine.h
> @@ -215,10 +215,10 @@ elf_machine_plt_value (struct link_map *map, const Elf32_Rela *reloc,
>  /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
>     MAP is the object containing the reloc.  */
>
> -auto inline void __attribute__ ((unused, always_inline))
> +static inline void __attribute__ ((unused, always_inline))
>  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>                   const Elf32_Sym *sym, const struct r_found_version *version,
> -                 void *const reloc_addr_arg, int skip_ifunc)
> +                 void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf32_Addr *const reloc_addr = reloc_addr_arg;
>    const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
> @@ -230,7 +230,11 @@ elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>    else
>      {
>        const Elf32_Sym *const refsym = sym;
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +      struct link_map *sym_map = boot_map;
> +#else
>        struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>        ElfW(Addr) value = SYMBOL_ADDRESS (sym_map, sym, true);
>        opcode16_addr = (unsigned short *)reloc_addr;
>
> @@ -331,7 +335,7 @@ elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>      }
>  }
>
> -auto inline void __attribute__ ((unused, always_inline))
> +static inline void __attribute__ ((unused, always_inline))
>  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>                            void *const reloc_addr_arg)
>  {
> @@ -339,7 +343,7 @@ elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>    *reloc_addr = l_addr + reloc->r_addend;
>  }
>
> -auto inline void __attribute__ ((unused, always_inline))
> +static inline void __attribute__ ((unused, always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       Elf32_Addr l_addr, const Elf32_Rela *reloc,
>                       int skip_ifunc)
> diff --git a/sysdeps/hppa/dl-machine.h b/sysdeps/hppa/dl-machine.h
> index f048fd2072..4aa159cfbd 100644
> --- a/sysdeps/hppa/dl-machine.h
> +++ b/sysdeps/hppa/dl-machine.h
> @@ -549,13 +549,13 @@ dl_platform_init (void)
>    (  (((as14) & 0x1fff) << 1) \
>     | (((as14) & 0x2000) >> 13))
>
> -auto void __attribute__((always_inline))
> +static void __attribute__((always_inline))
>  elf_machine_rela (struct link_map *map,
>                   const Elf32_Rela *reloc,
>                   const Elf32_Sym *sym,
>                   const struct r_found_version *version,
>                   void *const reloc_addr_arg,
> -                 int skip_ifunc)
> +                 int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf32_Addr *const reloc_addr = reloc_addr_arg;
>    const Elf32_Sym *const refsym = sym;
> @@ -581,7 +581,7 @@ elf_machine_rela (struct link_map *map,
>  # ifdef RTLD_BOOTSTRAP
>    /* RESOLVE_MAP in rtld.c doesn't have the local sym test.  */
>    sym_map = (ELF32_ST_BIND (sym->st_info) != STB_LOCAL
> -            ? RESOLVE_MAP (&sym, version, r_type) : map);
> +            ? boot_map : map);
>  # else
>    sym_map = RESOLVE_MAP (&sym, version, r_type);
>  # endif
> @@ -741,7 +741,7 @@ elf_machine_rela (struct link_map *map,
>
>  /* hppa doesn't have an R_PARISC_RELATIVE reloc, but uses relocs with
>     ELF32_R_SYM (info) == 0 for a similar purpose.  */
> -auto void __attribute__((always_inline))
> +static void __attribute__((always_inline))
>  elf_machine_rela_relative (Elf32_Addr l_addr,
>                            const Elf32_Rela *reloc,
>                            void *const reloc_addr_arg)
> @@ -794,7 +794,7 @@ elf_machine_rela_relative (Elf32_Addr l_addr,
>    *reloc_addr = value;
>  }
>
> -auto void __attribute__((always_inline))
> +static void __attribute__((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       Elf32_Addr l_addr, const Elf32_Rela *reloc,
>                       int skip_ifunc)
> diff --git a/sysdeps/i386/dl-machine.h b/sysdeps/i386/dl-machine.h
> index 590b41d8d7..9028e53ff1 100644
> --- a/sysdeps/i386/dl-machine.h
> +++ b/sysdeps/i386/dl-machine.h
> @@ -291,11 +291,11 @@ elf_machine_plt_value (struct link_map *map, const Elf32_Rel *reloc,
>  /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
>     MAP is the object containing the reloc.  */
>
> -auto inline void
> +static inline void
>  __attribute ((always_inline))
>  elf_machine_rel (struct link_map *map, const Elf32_Rel *reloc,
>                  const Elf32_Sym *sym, const struct r_found_version *version,
> -                void *const reloc_addr_arg, int skip_ifunc)
> +                void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf32_Addr *const reloc_addr = reloc_addr_arg;
>    const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
> @@ -498,11 +498,11 @@ and creates an unsatisfiable circular dependency.\n",
>  }
>
>  # ifndef RTLD_BOOTSTRAP
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>                   const Elf32_Sym *sym, const struct r_found_version *version,
> -                 void *const reloc_addr_arg, int skip_ifunc)
> +                 void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf32_Addr *const reloc_addr = reloc_addr_arg;
>    const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
> @@ -647,7 +647,7 @@ elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>  }
>  # endif        /* !RTLD_BOOTSTRAP */
>
> -auto inline void
> +static inline void
>  __attribute ((always_inline))
>  elf_machine_rel_relative (Elf32_Addr l_addr, const Elf32_Rel *reloc,
>                           void *const reloc_addr_arg)
> @@ -658,7 +658,7 @@ elf_machine_rel_relative (Elf32_Addr l_addr, const Elf32_Rel *reloc,
>  }
>
>  # ifndef RTLD_BOOTSTRAP
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>                            void *const reloc_addr_arg)
> @@ -668,7 +668,7 @@ elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>  }
>  # endif        /* !RTLD_BOOTSTRAP */
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       Elf32_Addr l_addr, const Elf32_Rel *reloc,
> @@ -707,12 +707,12 @@ elf_machine_lazy_rel (struct link_map *map,
>           ElfW(Half) ndx = version[ELFW(R_SYM) (r->r_info)] & 0x7fff;
>           elf_machine_rel (map, r, &symtab[ELFW(R_SYM) (r->r_info)],
>                            &map->l_versions[ndx],
> -                          (void *) (l_addr + r->r_offset), skip_ifunc);
> +                          (void *) (l_addr + r->r_offset), skip_ifunc, NULL);
>         }
>  # ifndef RTLD_BOOTSTRAP
>        else
>         elf_machine_rel (map, r, &symtab[ELFW(R_SYM) (r->r_info)], NULL,
> -                        (void *) (l_addr + r->r_offset), skip_ifunc);
> +                        (void *) (l_addr + r->r_offset), skip_ifunc, NULL);
>  # endif
>      }
>    else if (__glibc_unlikely (r_type == R_386_IRELATIVE))
> @@ -728,7 +728,7 @@ elf_machine_lazy_rel (struct link_map *map,
>
>  # ifndef RTLD_BOOTSTRAP
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_lazy_rela (struct link_map *map,
>                        Elf32_Addr l_addr, const Elf32_Rela *reloc,
> @@ -754,7 +754,7 @@ elf_machine_lazy_rela (struct link_map *map,
>
>        /* Always initialize TLS descriptors completely at load time, in
>          case static TLS is allocated for it that requires locking.  */
> -      elf_machine_rela (map, reloc, sym, version, reloc_addr, skip_ifunc);
> +      elf_machine_rela (map, reloc, sym, version, reloc_addr, skip_ifunc, NULL);
>      }
>    else if (__glibc_unlikely (r_type == R_386_IRELATIVE))
>      {
> diff --git a/sysdeps/ia64/dl-machine.h b/sysdeps/ia64/dl-machine.h
> index 4403e7767a..8390148134 100644
> --- a/sysdeps/ia64/dl-machine.h
> +++ b/sysdeps/ia64/dl-machine.h
> @@ -371,14 +371,14 @@ elf_machine_plt_value (struct link_map *map, const Elf64_Rela *reloc,
>
>  /* Perform the relocation specified by RELOC and SYM (which is fully
>     resolved).  MAP is the object containing the reloc.  */
> -auto inline void
> +static inline void
>  __attribute ((always_inline))
>  elf_machine_rela (struct link_map *map,
>                   const Elf64_Rela *reloc,
>                   const Elf64_Sym *sym,
>                   const struct r_found_version *version,
>                   void *const reloc_addr_arg,
> -                 int skip_ifunc)
> +                 int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf64_Addr *const reloc_addr = reloc_addr_arg;
>    const unsigned long int r_type = ELF64_R_TYPE (reloc->r_info);
> @@ -414,10 +414,14 @@ elf_machine_rela (struct link_map *map,
>        return;
>    else
>      {
> -      struct link_map *sym_map;
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +      struct link_map *sym_map = boot_map;
> +#else
> +      struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>
>        /* RESOLVE_MAP() will return NULL if it fail to locate the symbol.  */
> -      if ((sym_map = RESOLVE_MAP (&sym, version, r_type)))
> +      if (sym_map != NULL)
>         {
>           value = SYMBOL_ADDRESS (sym_map, sym, true) + reloc->r_addend;
>
> @@ -476,7 +480,7 @@ elf_machine_rela (struct link_map *map,
>     can be skipped.  */
>  #define ELF_MACHINE_REL_RELATIVE 1
>
> -auto inline void
> +static inline void
>  __attribute ((always_inline))
>  elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
>                            void *const reloc_addr_arg)
> @@ -489,7 +493,7 @@ elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
>  }
>
>  /* Perform a RELATIVE reloc on the .got entry that transfers to the .plt.  */
> -auto inline void
> +static inline void
>  __attribute ((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       Elf64_Addr l_addr, const Elf64_Rela *reloc,
> diff --git a/sysdeps/m68k/dl-machine.h b/sysdeps/m68k/dl-machine.h
> index 86a8c67e2a..9c7657c636 100644
> --- a/sysdeps/m68k/dl-machine.h
> +++ b/sysdeps/m68k/dl-machine.h
> @@ -215,10 +215,10 @@ elf_machine_plt_value (struct link_map *map, const Elf32_Rela *reloc,
>  /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
>     MAP is the object containing the reloc.  */
>
> -auto inline void __attribute__ ((unused, always_inline))
> +static inline void __attribute__ ((unused, always_inline))
>  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>                   const Elf32_Sym *sym, const struct r_found_version *version,
> -                 void *const reloc_addr_arg, int skip_ifunc)
> +                 void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf32_Addr *const reloc_addr = reloc_addr_arg;
>    const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
> @@ -228,7 +228,11 @@ elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>    else
>      {
>        const Elf32_Sym *const refsym = sym;
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +      struct link_map *sym_map = boot_map;
> +#else
>        struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>        Elf32_Addr value = SYMBOL_ADDRESS (sym_map, sym, true);
>
>        switch (r_type)
> @@ -303,7 +307,7 @@ elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>      }
>  }
>
> -auto inline void __attribute__ ((unused, always_inline))
> +static inline void __attribute__ ((unused, always_inline))
>  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>                            void *const reloc_addr_arg)
>  {
> @@ -311,7 +315,7 @@ elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>    *reloc_addr = l_addr + reloc->r_addend;
>  }
>
> -auto inline void __attribute__ ((unused, always_inline))
> +static inline void __attribute__ ((unused, always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       Elf32_Addr l_addr, const Elf32_Rela *reloc,
>                       int skip_ifunc)
> diff --git a/sysdeps/microblaze/dl-machine.h b/sysdeps/microblaze/dl-machine.h
> index e460f6f195..3e54676afe 100644
> --- a/sysdeps/microblaze/dl-machine.h
> +++ b/sysdeps/microblaze/dl-machine.h
> @@ -207,10 +207,10 @@ elf_machine_plt_value (struct link_map *map, const Elf32_Rela *reloc,
>      ((unsigned short *)(rel_addr))[3] = (val) & 0xffff; \
>    } while (0)
>
> -auto inline void __attribute__ ((always_inline))
> +static inline void __attribute__ ((always_inline))
>  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>                   const Elf32_Sym *sym, const struct r_found_version *version,
> -                 void *const reloc_addr_arg, int skip_ifunc)
> +                 void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf32_Addr *const reloc_addr = reloc_addr_arg;
>    const int r_type = ELF32_R_TYPE (reloc->r_info);
> @@ -222,7 +222,11 @@ elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>    else
>      {
>        const Elf32_Sym *const refsym = sym;
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +      struct link_map *sym_map = boot_map;
> +#else
>        struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>        Elf32_Addr value = SYMBOL_ADDRESS (sym_map, sym, true);
>
>        value += reloc->r_addend;
> @@ -277,7 +281,7 @@ elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>      }
>  }
>
> -auto inline void
> +static inline void
>  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>                            void *const reloc_addr_arg)
>  {
> @@ -285,7 +289,7 @@ elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>    PUT_REL_64 (reloc_addr, l_addr + reloc->r_addend);
>  }
>
> -auto inline void
> +static inline void
>  elf_machine_lazy_rel (struct link_map *map,
>                       Elf32_Addr l_addr, const Elf32_Rela *reloc,
>                       int skip_ifunc)
> diff --git a/sysdeps/mips/dl-machine.h b/sysdeps/mips/dl-machine.h
> index d9c6d33d0c..2293ea0b3e 100644
> --- a/sysdeps/mips/dl-machine.h
> +++ b/sysdeps/mips/dl-machine.h
> @@ -475,11 +475,12 @@ elf_machine_plt_value (struct link_map *map, const ElfW(Rel) *reloc,
>     by RELOC_ADDR.  SYM is the relocation symbol specified by R_INFO and
>     MAP is the object containing the reloc.  */
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_reloc (struct link_map *map, ElfW(Addr) r_info,
>                    const ElfW(Sym) *sym, const struct r_found_version *version,
> -                  void *reloc_addr, ElfW(Addr) r_addend, int inplace_p)
> +                  void *reloc_addr, ElfW(Addr) r_addend, int inplace_p,
> +                  struct link_map *boot_map)
>  {
>    const unsigned long int r_type = ELFW(R_TYPE) (r_info);
>    ElfW(Addr) *addr_field = (ElfW(Addr) *) reloc_addr;
> @@ -507,7 +508,11 @@ elf_machine_reloc (struct link_map *map, ElfW(Addr) r_info,
>      case R_MIPS_TLS_TPREL32:
>  # endif
>        {
> -       struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +      struct link_map *sym_map = boot_map;
> +#else
> +      struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>
>         switch (r_type)
>           {
> @@ -647,7 +652,11 @@ elf_machine_reloc (struct link_map *map, ElfW(Addr) r_info,
>           _dl_signal_error (0, map->l_name, NULL,
>                             "found jump slot relocation with non-zero addend");
>
> -       sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +        sym_map = boot_map;
> +#else
> +        sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>         value = SYMBOL_ADDRESS (sym_map, sym, true);
>         *addr_field = value;
>
> @@ -661,7 +670,11 @@ elf_machine_reloc (struct link_map *map, ElfW(Addr) r_info,
>         ElfW(Addr) value;
>
>         /* Calculate the address of the symbol.  */
> -       sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +        sym_map = boot_map;
> +#else
> +        sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>         value = SYMBOL_ADDRESS (sym_map, sym, true);
>
>         if (__builtin_expect (sym == NULL, 0))
> @@ -708,16 +721,16 @@ elf_machine_reloc (struct link_map *map, ElfW(Addr) r_info,
>  /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
>     MAP is the object containing the reloc.  */
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rel (struct link_map *map, const ElfW(Rel) *reloc,
>                  const ElfW(Sym) *sym, const struct r_found_version *version,
> -                void *const reloc_addr, int skip_ifunc)
> +                void *const reloc_addr, int skip_ifunc, struct link_map *boot_map)
>  {
> -  elf_machine_reloc (map, reloc->r_info, sym, version, reloc_addr, 0, 1);
> +  elf_machine_reloc (map, reloc->r_info, sym, version, reloc_addr, 0, 1, boot_map);
>  }
>
> -auto inline void
> +static inline void
>  __attribute__((always_inline))
>  elf_machine_rel_relative (ElfW(Addr) l_addr, const ElfW(Rel) *reloc,
>                           void *const reloc_addr)
> @@ -725,7 +738,7 @@ elf_machine_rel_relative (ElfW(Addr) l_addr, const ElfW(Rel) *reloc,
>    /* XXX Nothing to do.  There is no relative relocation, right?  */
>  }
>
> -auto inline void
> +static inline void
>  __attribute__((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       ElfW(Addr) l_addr, const ElfW(Rel) *reloc,
> @@ -748,17 +761,17 @@ elf_machine_lazy_rel (struct link_map *map,
>      _dl_reloc_bad_type (map, r_type, 1);
>  }
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
>                   const ElfW(Sym) *sym, const struct r_found_version *version,
> -                 void *const reloc_addr, int skip_ifunc)
> +                 void *const reloc_addr, int skip_ifunc, struct link_map *boot_map)
>  {
>    elf_machine_reloc (map, reloc->r_info, sym, version, reloc_addr,
> -                    reloc->r_addend, 0);
> +                    reloc->r_addend, 0, boot_map);
>  }
>
> -auto inline void
> +static inline void
>  __attribute__((always_inline))
>  elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
>                            void *const reloc_addr)
> @@ -767,7 +780,7 @@ elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
>
>  #ifndef RTLD_BOOTSTRAP
>  /* Relocate GOT. */
> -auto inline void
> +static inline void
>  __attribute__((always_inline))
>  elf_machine_got_rel (struct link_map *map, int lazy)
>  {
> @@ -868,7 +881,7 @@ elf_machine_got_rel (struct link_map *map, int lazy)
>  /* Set up the loaded object described by L so its stub function
>     will jump to the on-demand fixup code __dl_runtime_resolve.  */
>
> -auto inline int
> +static inline int
>  __attribute__((always_inline))
>  elf_machine_runtime_setup (struct link_map *l, int lazy, int profile)
>  {
> diff --git a/sysdeps/nios2/dl-machine.h b/sysdeps/nios2/dl-machine.h
> index e000cd081f..cd03ea93bd 100644
> --- a/sysdeps/nios2/dl-machine.h
> +++ b/sysdeps/nios2/dl-machine.h
> @@ -234,10 +234,10 @@ elf_machine_plt_value (struct link_map *map, const Elf32_Rela *reloc,
>     LOADADDR is the load address of the object; INFO is an array indexed
>     by DT_* of the .dynamic section info.  */
>
> -auto inline void __attribute__ ((always_inline))
> +static inline void __attribute__ ((always_inline))
>  elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
>                    const ElfW(Sym) *sym, const struct r_found_version *version,
> -                  void *const reloc_addr_arg, int skip_ifunc)
> +                  void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf32_Addr *const reloc_addr = reloc_addr_arg;
>    const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
> @@ -249,7 +249,11 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
>    else
>      {
>        const Elf32_Sym *const refsym = sym;
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +      struct link_map *sym_map = boot_map;
> +#else
>        struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>        Elf32_Addr value = SYMBOL_ADDRESS (sym_map, sym, true);
>
>        switch (r_type)
> @@ -314,7 +318,7 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
>      }
>  }
>
> -auto inline void __attribute__((always_inline))
> +static inline void __attribute__((always_inline))
>  elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
>                            void *const reloc_addr_arg)
>  {
> @@ -322,10 +326,10 @@ elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
>    *reloc_addr = l_addr + reloc->r_addend;
>  }
>
> -auto inline void __attribute__((always_inline))
> +static inline void __attribute__((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
> -                     int skip_ifunc)
> +                     int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf32_Addr *const reloc_addr = (void *) (l_addr + reloc->r_offset);
>    if (ELF32_R_TYPE (reloc->r_info) == R_NIOS2_JUMP_SLOT)
> diff --git a/sysdeps/powerpc/powerpc32/dl-machine.h b/sysdeps/powerpc/powerpc32/dl-machine.h
> index ced3a7b659..eaf3a1202f 100644
> --- a/sysdeps/powerpc/powerpc32/dl-machine.h
> +++ b/sysdeps/powerpc/powerpc32/dl-machine.h
> @@ -286,10 +286,10 @@ extern void _dl_reloc_overflow (struct link_map *map,
>     LOADADDR is the load address of the object; INFO is an array indexed
>     by DT_* of the .dynamic section info.  */
>
> -auto inline void __attribute__ ((always_inline))
> +static inline void __attribute__ ((always_inline))
>  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>                   const Elf32_Sym *sym, const struct r_found_version *version,
> -                 void *const reloc_addr_arg, int skip_ifunc)
> +                 void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf32_Addr *const reloc_addr = reloc_addr_arg;
>    const Elf32_Sym *const refsym = sym;
> @@ -317,7 +317,11 @@ elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>      }
>    else
>      {
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +      sym_map = boot_map;
> +#else
>        sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>        value = SYMBOL_ADDRESS (sym_map, sym, true);
>      }
>    value += reloc->r_addend;
> @@ -441,7 +445,7 @@ elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>      }
>  }
>
> -auto inline void __attribute__ ((always_inline))
> +static inline void __attribute__ ((always_inline))
>  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>                            void *const reloc_addr_arg)
>  {
> @@ -449,7 +453,7 @@ elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>    *reloc_addr = l_addr + reloc->r_addend;
>  }
>
> -auto inline void __attribute__ ((always_inline))
> +static inline void __attribute__ ((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       Elf32_Addr l_addr, const Elf32_Rela *reloc,
>                       int skip_ifunc)
> diff --git a/sysdeps/powerpc/powerpc64/dl-machine.h b/sysdeps/powerpc/powerpc64/dl-machine.h
> index b90f407119..b88d9d7912 100644
> --- a/sysdeps/powerpc/powerpc64/dl-machine.h
> +++ b/sysdeps/powerpc/powerpc64/dl-machine.h
> @@ -620,7 +620,7 @@ extern void attribute_hidden _dl_reloc_overflow (struct link_map *map,
>                                                  Elf64_Addr *const reloc_addr,
>                                                  const Elf64_Sym *refsym);
>
> -auto inline void __attribute__ ((always_inline))
> +static inline void __attribute__ ((always_inline))
>  elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
>                            void *const reloc_addr_arg)
>  {
> @@ -629,7 +629,7 @@ elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
>  }
>
>  /* This computes the value used by TPREL* relocs.  */
> -auto inline Elf64_Addr __attribute__ ((always_inline, const))
> +static inline Elf64_Addr __attribute__ ((always_inline, const))
>  elf_machine_tprel (struct link_map *map,
>                    struct link_map *sym_map,
>                    const Elf64_Sym *sym,
> @@ -648,7 +648,7 @@ elf_machine_tprel (struct link_map *map,
>  }
>
>  /* Call function at address VALUE (an OPD entry) to resolve ifunc relocs.  */
> -auto inline Elf64_Addr __attribute__ ((always_inline))
> +static inline Elf64_Addr __attribute__ ((always_inline))
>  resolve_ifunc (Elf64_Addr value,
>                const struct link_map *map, const struct link_map *sym_map)
>  {
> @@ -678,13 +678,13 @@ resolve_ifunc (Elf64_Addr value,
>
>  /* Perform the relocation specified by RELOC and SYM (which is fully
>     resolved).  MAP is the object containing the reloc.  */
> -auto inline void __attribute__ ((always_inline))
> +static inline void __attribute__ ((always_inline))
>  elf_machine_rela (struct link_map *map,
>                   const Elf64_Rela *reloc,
>                   const Elf64_Sym *sym,
>                   const struct r_found_version *version,
>                   void *const reloc_addr_arg,
> -                 int skip_ifunc)
> +                 int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf64_Addr *const reloc_addr = reloc_addr_arg;
>    const int r_type = ELF64_R_TYPE (reloc->r_info);
> @@ -707,7 +707,11 @@ elf_machine_rela (struct link_map *map,
>
>    /* We need SYM_MAP even in the absence of TLS, for elf_machine_fixup_plt
>       and STT_GNU_IFUNC.  */
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +  struct link_map *sym_map = boot_map;
> +#else
>    struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>    Elf64_Addr value = SYMBOL_ADDRESS (sym_map, sym, true) + reloc->r_addend;
>
>    if (sym != NULL
> @@ -1037,7 +1041,7 @@ elf_machine_rela (struct link_map *map,
>    MODIFIED_CODE_NOQUEUE (reloc_addr);
>  }
>
> -auto inline void __attribute__ ((always_inline))
> +static inline void __attribute__ ((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       Elf64_Addr l_addr, const Elf64_Rela *reloc,
>                       int skip_ifunc)
> diff --git a/sysdeps/riscv/dl-machine.h b/sysdeps/riscv/dl-machine.h
> index 5b0746175c..8e759ac8d3 100644
> --- a/sysdeps/riscv/dl-machine.h
> +++ b/sysdeps/riscv/dl-machine.h
> @@ -161,11 +161,11 @@ elf_machine_fixup_plt (struct link_map *map, lookup_t t,
>     by RELOC_ADDR.  SYM is the relocation symbol specified by R_INFO and
>     MAP is the object containing the reloc.  */
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
>                   const ElfW(Sym) *sym, const struct r_found_version *version,
> -                 void *const reloc_addr, int skip_ifunc)
> +                 void *const reloc_addr, int skip_ifunc, struct link_map *boot_map)
>  {
>    ElfW(Addr) r_info = reloc->r_info;
>    const unsigned long int r_type = ELFW (R_TYPE) (r_info);
> @@ -279,7 +279,7 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
>      }
>  }
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
>                           void *const reloc_addr)
> @@ -287,7 +287,7 @@ elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
>    *(ElfW(Addr) *) reloc_addr = l_addr + reloc->r_addend;
>  }
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_lazy_rel (struct link_map *map, ElfW(Addr) l_addr,
>                       const ElfW(Rela) *reloc, int skip_ifunc)
> @@ -320,7 +320,7 @@ elf_machine_lazy_rel (struct link_map *map, ElfW(Addr) l_addr,
>  /* Set up the loaded object described by L so its stub function
>     will jump to the on-demand fixup code __dl_runtime_resolve.  */
>
> -auto inline int
> +static inline int
>  __attribute__ ((always_inline))
>  elf_machine_runtime_setup (struct link_map *l, int lazy, int profile)
>  {
> diff --git a/sysdeps/s390/s390-32/dl-machine.h b/sysdeps/s390/s390-32/dl-machine.h
> index d0ccd69261..712367fa16 100644
> --- a/sysdeps/s390/s390-32/dl-machine.h
> +++ b/sysdeps/s390/s390-32/dl-machine.h
> @@ -321,11 +321,11 @@ elf_machine_plt_value (struct link_map *map, const Elf32_Rela *reloc,
>  /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
>     MAP is the object containing the reloc.  */
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>                   const Elf32_Sym *sym, const struct r_found_version *version,
> -                 void *const reloc_addr_arg, int skip_ifunc)
> +                 void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf32_Addr *const reloc_addr = reloc_addr_arg;
>    const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
> @@ -484,7 +484,7 @@ elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>      }
>  }
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>                            void *const reloc_addr_arg)
> @@ -493,7 +493,7 @@ elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>    *reloc_addr = l_addr + reloc->r_addend;
>  }
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       Elf32_Addr l_addr, const Elf32_Rela *reloc,
> diff --git a/sysdeps/s390/s390-64/dl-machine.h b/sysdeps/s390/s390-64/dl-machine.h
> index 543361c836..77a658e17f 100644
> --- a/sysdeps/s390/s390-64/dl-machine.h
> +++ b/sysdeps/s390/s390-64/dl-machine.h
> @@ -268,11 +268,11 @@ elf_machine_plt_value (struct link_map *map, const Elf64_Rela *reloc,
>  /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
>     MAP is the object containing the reloc.  */
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela (struct link_map *map, const Elf64_Rela *reloc,
>                   const Elf64_Sym *sym, const struct r_found_version *version,
> -                 void *const reloc_addr_arg, int skip_ifunc)
> +                 void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf64_Addr *const reloc_addr = reloc_addr_arg;
>    const unsigned int r_type = ELF64_R_TYPE (reloc->r_info);
> @@ -438,7 +438,7 @@ elf_machine_rela (struct link_map *map, const Elf64_Rela *reloc,
>      }
>  }
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
>                            void *const reloc_addr_arg)
> @@ -447,7 +447,7 @@ elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
>    *reloc_addr = l_addr + reloc->r_addend;
>  }
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       Elf64_Addr l_addr, const Elf64_Rela *reloc,
> diff --git a/sysdeps/sh/dl-machine.h b/sysdeps/sh/dl-machine.h
> index 122b417a17..9c3c0ba099 100644
> --- a/sysdeps/sh/dl-machine.h
> +++ b/sysdeps/sh/dl-machine.h
> @@ -259,11 +259,11 @@ elf_machine_plt_value (struct link_map *map, const Elf32_Rela *reloc,
>  /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
>     MAP is the object containing the reloc.  */
>
> -auto inline void
> +static inline void
>  __attribute ((always_inline))
>  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>                   const Elf32_Sym *sym, const struct r_found_version *version,
> -                 void *const reloc_addr_arg, int skip_ifunc)
> +                 void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf32_Addr *const reloc_addr = reloc_addr_arg;
>    const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
> @@ -318,7 +318,11 @@ elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>    else
>      {
>        const Elf32_Sym *const refsym = sym;
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +      struct link_map *sym_map = boot_map;
> +#else
>        struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>
>        value = SYMBOL_ADDRESS (sym_map, sym, true);
>        value += reloc->r_addend;
> @@ -424,7 +428,7 @@ elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>      }
>  }
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>                            void *const reloc_addr_arg)
> @@ -443,7 +447,7 @@ elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>  #undef COPY_UNALIGNED_WORD
>  }
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       Elf32_Addr l_addr, const Elf32_Rela *reloc,
> diff --git a/sysdeps/sparc/sparc32/dl-machine.h b/sysdeps/sparc/sparc32/dl-machine.h
> index 0269e458ea..20f60b2a3b 100644
> --- a/sysdeps/sparc/sparc32/dl-machine.h
> +++ b/sysdeps/sparc/sparc32/dl-machine.h
> @@ -327,11 +327,11 @@ elf_machine_plt_value (struct link_map *map, const Elf32_Rela *reloc,
>  /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
>     MAP is the object containing the reloc.  */
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>                   const Elf32_Sym *sym, const struct r_found_version *version,
> -                 void *const reloc_addr_arg, int skip_ifunc)
> +                 void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf32_Addr *const reloc_addr = reloc_addr_arg;
>  #if !defined RTLD_BOOTSTRAP && !defined RESOLVE_CONFLICT_FIND_MAP
> @@ -381,7 +381,11 @@ elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>      }
>    else
>      {
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +      sym_map = boot_map;
> +#else
>        sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>        value = SYMBOL_ADDRESS (sym_map, sym, true);
>      }
>  #else
> @@ -536,7 +540,7 @@ elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
>      }
>  }
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>                            void *const reloc_addr_arg)
> @@ -545,7 +549,7 @@ elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
>    *reloc_addr += l_addr + reloc->r_addend;
>  }
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       Elf32_Addr l_addr, const Elf32_Rela *reloc,
> diff --git a/sysdeps/sparc/sparc64/dl-machine.h b/sysdeps/sparc/sparc64/dl-machine.h
> index bbd4566d8a..4d47b3a150 100644
> --- a/sysdeps/sparc/sparc64/dl-machine.h
> +++ b/sysdeps/sparc/sparc64/dl-machine.h
> @@ -354,11 +354,11 @@ elf_machine_runtime_setup (struct link_map *l, int lazy, int profile)
>  /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
>     MAP is the object containing the reloc.  */
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela (struct link_map *map, const Elf64_Rela *reloc,
>                   const Elf64_Sym *sym, const struct r_found_version *version,
> -                 void *const reloc_addr_arg, int skip_ifunc)
> +                 void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
>  {
>    Elf64_Addr *const reloc_addr = reloc_addr_arg;
>  #if !defined RTLD_BOOTSTRAP && !defined RESOLVE_CONFLICT_FIND_MAP
> @@ -408,7 +408,11 @@ elf_machine_rela (struct link_map *map, const Elf64_Rela *reloc,
>      }
>    else
>      {
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +      sym_map = boot_map;
> +#else
>        sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>        value = SYMBOL_ADDRESS (sym_map, sym, true);
>      }
>  #else
> @@ -646,7 +650,7 @@ elf_machine_rela (struct link_map *map, const Elf64_Rela *reloc,
>      }
>  }
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
>                            void *const reloc_addr_arg)
> @@ -655,7 +659,7 @@ elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
>    *reloc_addr = l_addr + reloc->r_addend;
>  }
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       Elf64_Addr l_addr, const Elf64_Rela *reloc,
> diff --git a/sysdeps/x86_64/dl-machine.h b/sysdeps/x86_64/dl-machine.h
> index ceee50734e..738663ec5d 100644
> --- a/sysdeps/x86_64/dl-machine.h
> +++ b/sysdeps/x86_64/dl-machine.h
> @@ -251,11 +251,12 @@ elf_machine_plt_value (struct link_map *map, const ElfW(Rela) *reloc,
>  /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
>     MAP is the object containing the reloc.  */
>
> -auto inline void
> +static inline void
>  __attribute__ ((always_inline))
>  elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
>                   const ElfW(Sym) *sym, const struct r_found_version *version,
> -                 void *const reloc_addr_arg, int skip_ifunc)
> +                 void *const reloc_addr_arg, int skip_ifunc,
> +                 struct link_map *boot_map)
>  {
>    ElfW(Addr) *const reloc_addr = reloc_addr_arg;
>    const unsigned long int r_type = ELFW(R_TYPE) (reloc->r_info);
> @@ -293,7 +294,11 @@ elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
>  # ifndef RTLD_BOOTSTRAP
>        const ElfW(Sym) *const refsym = sym;
>  # endif
> +#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
> +      struct link_map *sym_map = boot_map;
> +#else
>        struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
> +#endif
>        ElfW(Addr) value = SYMBOL_ADDRESS (sym_map, sym, true);
>
>        if (sym != NULL
> @@ -518,7 +523,7 @@ and creates an unsatisfiable circular dependency.\n",
>      }
>  }
>
> -auto inline void
> +static inline void
>  __attribute ((always_inline))
>  elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
>                            void *const reloc_addr_arg)
> @@ -537,7 +542,7 @@ elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
>      }
>  }
>
> -auto inline void
> +static inline void
>  __attribute ((always_inline))
>  elf_machine_lazy_rel (struct link_map *map,
>                       ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
> @@ -573,7 +578,8 @@ elf_machine_lazy_rel (struct link_map *map,
>
>        /* Always initialize TLS descriptors completely at load time, in
>          case static TLS is allocated for it that requires locking.  */
> -      elf_machine_rela (map, reloc, sym, version, reloc_addr, skip_ifunc);
> +      elf_machine_rela (map, reloc, sym, version, reloc_addr, skip_ifunc,
> +                        NULL);
>      }
>    else if (__glibc_unlikely (r_type == R_X86_64_IRELATIVE))
>      {
> --
> 2.33.0.rc2.250.ged5fa647cd-goog
>
  
Florian Weimer Sept. 3, 2021, 10:09 a.m. UTC | #2
* Fangrui Song via Libc-alpha:

> [Alternative to https://sourceware.org/pipermail/libc-alpha/2021-August/130340.html
> This version fixes all ports and doesn't add NESTING dispatches.]
>
> dynamic-link.h is included more than once in some elf/ files (rtld.c,
> dl-conflict.c, dl-reloc.c, dl-reloc-static-pie.c) and uses GCC nested
> functions. This harms readability and the nested functions usage
> is the biggest obstacle prevents CC=clang (which doesn't support the
> feature).

This patch causes elf/tst-nodelete to fail on x86_64, powerpc64,
powerpc64le at least:

| Unique symbols test failed

I assume something is wrong with the initialization of the new global
variables.

Thanks,
Florian
  
Fangrui Song Sept. 4, 2021, 3:52 a.m. UTC | #3
On 2021-09-03, Florian Weimer wrote:
>* Fangrui Song via Libc-alpha:
>
>> [Alternative to https://sourceware.org/pipermail/libc-alpha/2021-August/130340.html
>> This version fixes all ports and doesn't add NESTING dispatches.]
>>
>> dynamic-link.h is included more than once in some elf/ files (rtld.c,
>> dl-conflict.c, dl-reloc.c, dl-reloc-static-pie.c) and uses GCC nested
>> functions. This harms readability and the nested functions usage
>> is the biggest obstacle prevents CC=clang (which doesn't support the
>> feature).
>
>This patch causes elf/tst-nodelete to fail on x86_64, powerpc64,
>powerpc64le at least:
>
>| Unique symbols test failed
>
>I assume something is wrong with the initialization of the new global
>variables.

I unintentionally missed DL_LOOKUP_FOR_RELOCATE
(fcb04b9aed26a737159ef7be9c5a6ad0994437dc ("Introduce
DL_LOOKUP_FOR_RELOCATE flag for _dl_lookup_symbol_x")) when rebasing the
multi-line macro RESOLVE_MAP.

Fixed in
https://sourceware.org/git/?p=glibc.git;a=shortlog;h=refs/heads/maskray/unnest

(Sorry about my negligence. On my Debian, master has 14 FAIL, and I
probably missed that tst-nodelete additionally failed.)
  
Fangrui Song Sept. 13, 2021, 4:36 p.m. UTC | #4
ping

On Fri, Sep 3, 2021 at 8:52 PM Fangrui Song <maskray@google.com> wrote:
>
> On 2021-09-03, Florian Weimer wrote:
> >* Fangrui Song via Libc-alpha:
> >
> >> [Alternative to https://sourceware.org/pipermail/libc-alpha/2021-August/130340.html
> >> This version fixes all ports and doesn't add NESTING dispatches.]
> >>
> >> dynamic-link.h is included more than once in some elf/ files (rtld.c,
> >> dl-conflict.c, dl-reloc.c, dl-reloc-static-pie.c) and uses GCC nested
> >> functions. This harms readability and the nested functions usage
> >> is the biggest obstacle prevents CC=clang (which doesn't support the
> >> feature).
> >
> >This patch causes elf/tst-nodelete to fail on x86_64, powerpc64,
> >powerpc64le at least:
> >
> >| Unique symbols test failed
> >
> >I assume something is wrong with the initialization of the new global
> >variables.
>
> I unintentionally missed DL_LOOKUP_FOR_RELOCATE
> (fcb04b9aed26a737159ef7be9c5a6ad0994437dc ("Introduce
> DL_LOOKUP_FOR_RELOCATE flag for _dl_lookup_symbol_x")) when rebasing the
> multi-line macro RESOLVE_MAP.
>
> Fixed in
> https://sourceware.org/git/?p=glibc.git;a=shortlog;h=refs/heads/maskray/unnest
>
> (Sorry about my negligence. On my Debian, master has 14 FAIL, and I
> probably missed that tst-nodelete additionally failed.)
  
Fangrui Song Sept. 20, 2021, 5:55 p.m. UTC | #5
ping

On Mon, Sep 13, 2021 at 9:36 AM Fāng-ruì Sòng <maskray@google.com> wrote:
>
> ping
>
> On Fri, Sep 3, 2021 at 8:52 PM Fangrui Song <maskray@google.com> wrote:
> >
> > On 2021-09-03, Florian Weimer wrote:
> > >* Fangrui Song via Libc-alpha:
> > >
> > >> [Alternative to https://sourceware.org/pipermail/libc-alpha/2021-August/130340.html
> > >> This version fixes all ports and doesn't add NESTING dispatches.]
> > >>
> > >> dynamic-link.h is included more than once in some elf/ files (rtld.c,
> > >> dl-conflict.c, dl-reloc.c, dl-reloc-static-pie.c) and uses GCC nested
> > >> functions. This harms readability and the nested functions usage
> > >> is the biggest obstacle prevents CC=clang (which doesn't support the
> > >> feature).
> > >
> > >This patch causes elf/tst-nodelete to fail on x86_64, powerpc64,
> > >powerpc64le at least:
> > >
> > >| Unique symbols test failed
> > >
> > >I assume something is wrong with the initialization of the new global
> > >variables.
> >
> > I unintentionally missed DL_LOOKUP_FOR_RELOCATE
> > (fcb04b9aed26a737159ef7be9c5a6ad0994437dc ("Introduce
> > DL_LOOKUP_FOR_RELOCATE flag for _dl_lookup_symbol_x")) when rebasing the
> > multi-line macro RESOLVE_MAP.
> >
> > Fixed in
> > https://sourceware.org/git/?p=glibc.git;a=shortlog;h=refs/heads/maskray/unnest
> >
> > (Sorry about my negligence. On my Debian, master has 14 FAIL, and I
> > probably missed that tst-nodelete additionally failed.)
  
Carlos O'Donell Sept. 22, 2021, 5:53 p.m. UTC | #6
On 9/20/21 13:55, Fāng-ruì Sòng wrote:
> ping

Thanks for pinging and continuing to ping this issue.

Thanks also for raising this issue at the glibc bof in the GNU Tools track at LPC.

The consensus in the room was that two things need to be done next:

- Split this patch up into more logical changes.
- Serious review of the changes in bootstrap relocation requirements due to the
  removal of nested functions. That in some cases switching to static may
  create bootstrap relocation requirements that aren't currently met. This may
  mean we need to refactor how the data is passed around to avoid those relocations.
  
Joseph Myers Sept. 22, 2021, 6:09 p.m. UTC | #7
On Wed, 22 Sep 2021, Carlos O'Donell via Libc-alpha wrote:

> - Split this patch up into more logical changes.

I'm not convinced there's much scope for splitting up this patch more; it 
is a logical change (versions that changed only some architectures were a 
lot uglier).  My comment about logical changes was a more generic one, not 
suggesting that this particular patch should be split up further.
  
Carlos O'Donell Sept. 22, 2021, 6:20 p.m. UTC | #8
On 9/22/21 14:09, Joseph Myers wrote:
> On Wed, 22 Sep 2021, Carlos O'Donell via Libc-alpha wrote:
> 
>> - Split this patch up into more logical changes.
> 
> I'm not convinced there's much scope for splitting up this patch more; it 
> is a logical change (versions that changed only some architectures were a 
> lot uglier).  My comment about logical changes was a more generic one, not 
> suggesting that this particular patch should be split up further.

Thanks for the clarification. In which case the more detailed review of the boostrap
relocations is the remaining point.
  
Fangrui Song Sept. 22, 2021, 6:40 p.m. UTC | #9
On 2021-09-22, Carlos O'Donell wrote:
>On 9/22/21 14:09, Joseph Myers wrote:
>> On Wed, 22 Sep 2021, Carlos O'Donell via Libc-alpha wrote:
>>
>>> - Split this patch up into more logical changes.
>>
>> I'm not convinced there's much scope for splitting up this patch more; it
>> is a logical change (versions that changed only some architectures were a
>> lot uglier).  My comment about logical changes was a more generic one, not
>> suggesting that this particular patch should be split up further.
>
>Thanks for the clarification. In which case the more detailed review of the boostrap
>relocations is the remaining point.

I really appreciate that Clang buildability is well-received :)
I went ahead and filed https://sourceware.org/bugzilla/show_bug.cgi?id=28376 "[meta] Build glibc with Clang"
to collect miscellaneous small fixes.

---

I agree that this patch is difficult to split.
Modifying the elf/ macro parameters requires fixing all use sites.

The latest version is at https://github.com/MaskRay/glibc/tree/unnest
(I use a different repo instead of the canonical sourceware repo in the fear of pushing bad commits to sourceware :)

About the global variables (cur_*): they are in elf/dl-reloc.c

-    /* String table object symbols.  */
-    const char *strtab = (const void *) D_PTR (l, l_info[DT_STRTAB]);
+    cur_l = l;
+    cur_scope = scope;
+    cur_strtab = (const void *) D_PTR (cur_l, l_info[DT_STRTAB]);

removing them requires more parameters to ELF_DYNAMIC_RELOCATE and RESOLVE_MAP.
Is that the direction you are imagining?

BTW: I am not familiar with the concept of bootstrap relocations.
Does it refer to relocations in ld.so itself   and   all relocations in a static pie?

---

FWIW I initially sent
https://sourceware.org/pipermail/libc-alpha/2021-August/130340.html
(https://github.com/MaskRay/glibc/tree/nesting)
which is a more literal take from
https://sourceware.org/git/?p=glibc.git;a=shortlog;h=refs/heads/google/grte/v5-2.27/master

But I slightly dislike the version myself because it duplicates lots of macros.
(This branch can give some insight for needed Clang fixes but many commits need rewrite for master.)
  
Joseph Myers Sept. 22, 2021, 7:21 p.m. UTC | #10
On Wed, 22 Sep 2021, Fāng-ruì Sòng via Libc-alpha wrote:

> I went ahead and filed https://sourceware.org/bugzilla/show_bug.cgi?id=28376
> "[meta] Build glibc with Clang"
> to collect miscellaneous small fixes.

Note also that issues regarding use of glibc installed headers should 
probably be handled separately (with a separate meta-bug to depend on 
them; we already have a few such bugs open, e.g. 26287).  The key 
difference there is that installed headers should support a wide range of 
compilers, including old versions.  Whereas for issues building glibc, the 
conclusion in some cases may well be that a change should be made to 
Clang, and only new versions with that fix supported, rather than working 
around a limitation in existing Clang versions.

> About the global variables (cur_*): they are in elf/dl-reloc.c
> 
> -    /* String table object symbols.  */
> -    const char *strtab = (const void *) D_PTR (l, l_info[DT_STRTAB]);
> +    cur_l = l;
> +    cur_scope = scope;
> +    cur_strtab = (const void *) D_PTR (cur_l, l_info[DT_STRTAB]);
> 
> removing them requires more parameters to ELF_DYNAMIC_RELOCATE and
> RESOLVE_MAP.
> Is that the direction you are imagining?

Yes.  Explicit parameters (whether individual parameters or in a struct) 
seems cleaner than using global state to pass parameters.
  
Fangrui Song Sept. 22, 2021, 7:44 p.m. UTC | #11
On 2021-09-22, Joseph Myers wrote:
>On Wed, 22 Sep 2021, Fāng-ruì Sòng via Libc-alpha wrote:
>
>> I went ahead and filed https://sourceware.org/bugzilla/show_bug.cgi?id=28376
>> "[meta] Build glibc with Clang"
>> to collect miscellaneous small fixes.
>
>Note also that issues regarding use of glibc installed headers should
>probably be handled separately (with a separate meta-bug to depend on
>them; we already have a few such bugs open, e.g. 26287).  The key
>difference there is that installed headers should support a wide range of
>compilers, including old versions.  Whereas for issues building glibc, the
>conclusion in some cases may well be that a change should be made to
>Clang, and only new versions with that fix supported, rather than working
>around a limitation in existing Clang versions.

Sure! Changes need to be assessed where they belong to, Clang or glibc.

For folks who don't know me, I have pushed many Clang fixes to make
glibc builds better, e.g.
(Some were from my observation that some
https://sourceware.org/git/?p=glibc.git;a=shortlog;h=refs/heads/google/grte/v5-2.27/master
commits could be avoided by fixing Clang instead)

* https://reviews.llvm.org/rGc841b9abf039ec0457752cd96f7e4716c1c7a323 "[MC][ELF] Don't create relocations with section symbols for STB_LOCAL ifunc"
* https://reviews.llvm.org/D64283 PowerPC64 -mabi=ieeelongdouble (the heavy lift work is on IBM's side)
* https://reviews.llvm.org/rGf931290308abd0eebecae385cd32ca3a25ddd9be [PowerPC] Parse and ignore .machine
* https://reviews.llvm.org/D88625 better support for asm("memcpy = __GI_memcpy");
* https://reviews.llvm.org/D88712 respect asm label for built-in functions

However, for

void bar();
void foo() { bar(); }
void bar() asm("bar1");

GCC happily redirects the bar call to bar1.
Clang rejects this 
x.c:5:6: error: cannot apply asm label to function after its first use
void bar() asm("bar1");
      ^         ~~~~~~

The issue is really to fix on Clang's side.
(I spent many hours in a weekend for investigation and made the
conclusion. Basically Clang does "parse decl A; codegen decl A; parse
decl B; codegen decl B; ..." When it sees the asm label, it is too late
to change the previous codegen.)

This just needs some declaration reordering (the number of lines doesn't
even need to increase) which is probably not a big burden on glibc's side.

I hope glibc can be lenient in such a situation where Clang fixes may
not be feasible.

>> About the global variables (cur_*): they are in elf/dl-reloc.c
>>
>> -    /* String table object symbols.  */
>> -    const char *strtab = (const void *) D_PTR (l, l_info[DT_STRTAB]);
>> +    cur_l = l;
>> +    cur_scope = scope;
>> +    cur_strtab = (const void *) D_PTR (cur_l, l_info[DT_STRTAB]);
>>
>> removing them requires more parameters to ELF_DYNAMIC_RELOCATE and
>> RESOLVE_MAP.
>> Is that the direction you are imagining?
>
>Yes.  Explicit parameters (whether individual parameters or in a struct)
>seems cleaner than using global state to pass parameters.

Thanks for the confirmation. I will take a stab.
  
Florian Weimer Sept. 23, 2021, 10 p.m. UTC | #12
* Fāng-ruì Sòng:

> For folks who don't know me, I have pushed many Clang fixes to make
> glibc builds better, e.g.
> (Some were from my observation that some
> https://sourceware.org/git/?p=glibc.git;a=shortlog;h=refs/heads/google/grte/v5-2.27/master
> commits could be avoided by fixing Clang instead)
>
> * https://reviews.llvm.org/rGc841b9abf039ec0457752cd96f7e4716c1c7a323 "[MC][ELF] Don't create relocations with section symbols for STB_LOCAL ifunc"
> * https://reviews.llvm.org/D64283 PowerPC64 -mabi=ieeelongdouble (the heavy lift work is on IBM's side)
> * https://reviews.llvm.org/rGf931290308abd0eebecae385cd32ca3a25ddd9be [PowerPC] Parse and ignore .machine
> * https://reviews.llvm.org/D88625 better support for asm("memcpy = __GI_memcpy");
> * https://reviews.llvm.org/D88712 respect asm label for built-in functions

I very much appreciate that you have fixed those minor Clang
incompatibility problems that have been in a WONTFIX state for so long
(especially the last change).

> However, for
>
> void bar();
> void foo() { bar(); }
> void bar() asm("bar1");
>
> GCC happily redirects the bar call to bar1.
> Clang rejects this x.c:5:6: error: cannot apply asm label to function
> after its first use
> void bar() asm("bar1");
>      ^         ~~~~~~
>
> The issue is really to fix on Clang's side.
> (I spent many hours in a weekend for investigation and made the
> conclusion. Basically Clang does "parse decl A; codegen decl A; parse
> decl B; codegen decl B; ..." When it sees the asm label, it is too late
> to change the previous codegen.)

I think GCC used to call this function-at-a-time mode.

> This just needs some declaration reordering (the number of lines doesn't
> even need to increase) which is probably not a big burden on glibc's
> side.

This is a bit unfortunate.  I want to rework how we deal with hidden
prototypes (for PLT/relocation avoidance), so that what glibc does
becomes much more contributor-friendly.  In part I want to to this to
offset some of the maintenance cost I added with the preprocessor goo
for the libpthread merge.  What I want to do is to parse the installed
headers and automatically generate the aliases in a consistent fashion.
With GCC, I could add the aliases in a separate (wrapper) header.  To
support Clang, it looks like I would have to rewrite the header instead
of merely parsing it, which hopefully won't be too bad.

Does Clang support #pragma redefine_extname?  And can that be applied
*before* the declaration?  (We can't do that with __asm__ because we
don't know the appropriate prototype at this point.)  If there's nothing
that speaks against using #pragma redefine_extname, it might provide a
way to avoid ordering issues.

Thanks,
Florian
  
Fangrui Song Sept. 23, 2021, 10:14 p.m. UTC | #13
On Thu, Sep 23, 2021 at 3:01 PM Florian Weimer <fweimer@redhat.com> wrote:
>
> * Fāng-ruì Sòng:
>
> > For folks who don't know me, I have pushed many Clang fixes to make
> > glibc builds better, e.g.
> > (Some were from my observation that some
> > https://sourceware.org/git/?p=glibc.git;a=shortlog;h=refs/heads/google/grte/v5-2.27/master
> > commits could be avoided by fixing Clang instead)
> >
> > * https://reviews.llvm.org/rGc841b9abf039ec0457752cd96f7e4716c1c7a323 "[MC][ELF] Don't create relocations with section symbols for STB_LOCAL ifunc"
> > * https://reviews.llvm.org/D64283 PowerPC64 -mabi=ieeelongdouble (the heavy lift work is on IBM's side)
> > * https://reviews.llvm.org/rGf931290308abd0eebecae385cd32ca3a25ddd9be [PowerPC] Parse and ignore .machine
> > * https://reviews.llvm.org/D88625 better support for asm("memcpy = __GI_memcpy");
> > * https://reviews.llvm.org/D88712 respect asm label for built-in functions
>
> I very much appreciate that you have fixed those minor Clang
> incompatibility problems that have been in a WONTFIX state for so long
> (especially the last change).

Thanks

> > However, for
> >
> > void bar();
> > void foo() { bar(); }
> > void bar() asm("bar1");
> >
> > GCC happily redirects the bar call to bar1.
> > Clang rejects this x.c:5:6: error: cannot apply asm label to function
> > after its first use
> > void bar() asm("bar1");
> >      ^         ~~~~~~
> >
> > The issue is really to fix on Clang's side.
> > (I spent many hours in a weekend for investigation and made the
> > conclusion. Basically Clang does "parse decl A; codegen decl A; parse
> > decl B; codegen decl B; ..." When it sees the asm label, it is too late
> > to change the previous codegen.)
>
> I think GCC used to call this function-at-a-time mode.
>
> > This just needs some declaration reordering (the number of lines doesn't
> > even need to increase) which is probably not a big burden on glibc's
> > side.
>
> This is a bit unfortunate.  I want to rework how we deal with hidden
> prototypes (for PLT/relocation avoidance), so that what glibc does
> becomes much more contributor-friendly.  In part I want to to this to
> offset some of the maintenance cost I added with the preprocessor goo
> for the libpthread merge.  What I want to do is to parse the installed
> headers and automatically generate the aliases in a consistent fashion.
> With GCC, I could add the aliases in a separate (wrapper) header.  To
> support Clang, it looks like I would have to rewrite the header instead
> of merely parsing it, which hopefully won't be too bad.
>
> Does Clang support #pragma redefine_extname?  And can that be applied
> *before* the declaration?  (We can't do that with __asm__ because we
> don't know the appropriate prototype at this point.)  If there's nothing
> that speaks against using #pragma redefine_extname, it might provide a
> way to avoid ordering issues.

`#pragma redefine_extname` works. Internally Clang translates it into
an asm label.
I'll give annotated example below to demonstrate where an asm label
can be placed.
A function can be redeclared multiple times. The requirement is that
an asm label is added before the first use.

typedef unsigned long size_t;

// #pragma redefine_extname memcpy __GI_memcpy // before the first use, works
extern void *memcpy(void *, const void *, size_t);

#pragma redefine_extname memcpy __GI_memcpy // before the first use, works

void *test_memcpy(void *dst, const void *src, size_t n) { return
memcpy(dst, src, n); }

// #pragma redefine_extname memcpy __GI_memcpy // after the first use,
does not work


As https://reviews.llvm.org/D88712 mentions, the asm label does not
apply to Clang generated memcpy calls.
Certain optimization passes can synthesize built-in function calls.
It's really difficult to fix, also related to the function-at-a-time
mode.


> Thanks,
> Florian
>
  

Patch

diff --git a/elf/dl-conflict.c b/elf/dl-conflict.c
index 31a2f90770..d7219b726a 100644
--- a/elf/dl-conflict.c
+++ b/elf/dl-conflict.c
@@ -27,17 +27,9 @@ 
 #include <sys/types.h>
 #include "dynamic-link.h"
 
-void
-_dl_resolve_conflicts (struct link_map *l, ElfW(Rela) *conflict,
-		       ElfW(Rela) *conflictend)
-{
-#if ! ELF_MACHINE_NO_RELA
-  if (__glibc_unlikely (GLRO(dl_debug_mask) & DL_DEBUG_RELOC))
-    _dl_debug_printf ("\nconflict processing: %s\n", DSO_FILENAME (l->l_name));
-
-  {
-    /* Do the conflict relocation of the object and library GOT and other
-       data.  */
+/* Used by RESOLVE_CONFLICT_FIND_MAP at init time. Cannot be accessed
+ * concurrently.  */
+static struct link_map *resolve_conflict_map __attribute__ ((__unused__));
 
     /* This macro is used as a callback from the ELF_DYNAMIC_RELOCATE code.  */
 #define RESOLVE_MAP(ref, version, flags) (*ref = NULL, NULL)
@@ -51,12 +43,23 @@  _dl_resolve_conflicts (struct link_map *l, ElfW(Rela) *conflict,
     (map) = resolve_conflict_map;					      \
   } while (0)
 
+#include "dynamic-link.h"
+
+void
+_dl_resolve_conflicts (struct link_map *l, ElfW(Rela) *conflict,
+		       ElfW(Rela) *conflictend)
+{
+#if ! ELF_MACHINE_NO_RELA
+  if (__glibc_unlikely (GLRO(dl_debug_mask) & DL_DEBUG_RELOC))
+    _dl_debug_printf ("\nconflict processing: %s\n", DSO_FILENAME (l->l_name));
+
+  {
+    /* Do the conflict relocation of the object and library GOT and other
+       data.  */
+
     /* Prelinking makes no sense for anything but the main namespace.  */
     assert (l->l_ns == LM_ID_BASE);
-    struct link_map *resolve_conflict_map __attribute__ ((__unused__))
-      = GL(dl_ns)[LM_ID_BASE]._ns_loaded;
-
-#include "dynamic-link.h"
+    resolve_conflict_map = GL(dl_ns)[LM_ID_BASE]._ns_loaded;
 
     /* Override these, defined in dynamic-link.h.  */
 #undef CHECK_STATIC_TLS
@@ -68,7 +71,7 @@  _dl_resolve_conflicts (struct link_map *l, ElfW(Rela) *conflict,
 
     for (; conflict < conflictend; ++conflict)
       elf_machine_rela (l, conflict, NULL, NULL, (void *) conflict->r_offset,
-			0);
+			0, NULL);
   }
 #endif
 }
diff --git a/elf/dl-reloc-static-pie.c b/elf/dl-reloc-static-pie.c
index d5bd2f31e9..25d1c53aa2 100644
--- a/elf/dl-reloc-static-pie.c
+++ b/elf/dl-reloc-static-pie.c
@@ -23,6 +23,11 @@ 
 #include <ldsodefs.h>
 #include "dynamic-link.h"
 
+#define STATIC_PIE_BOOTSTRAP
+#define BOOTSTRAP_MAP (main_map)
+#define RESOLVE_MAP(sym, version, flags) BOOTSTRAP_MAP
+#include "dynamic-link.h"
+
 /* Relocate static executable with PIE.  */
 
 void
@@ -30,11 +35,6 @@  _dl_relocate_static_pie (void)
 {
   struct link_map *main_map = _dl_get_dl_main_map ();
 
-# define STATIC_PIE_BOOTSTRAP
-# define BOOTSTRAP_MAP (main_map)
-# define RESOLVE_MAP(sym, version, flags) BOOTSTRAP_MAP
-# include "dynamic-link.h"
-
   /* Figure out the run-time load address of static PIE.  */
   main_map->l_addr = elf_machine_load_address ();
 
@@ -48,7 +48,7 @@  _dl_relocate_static_pie (void)
 
   /* Relocate ourselves so we can do normal function calls and
      data access using the global offset table.  */
-  ELF_DYNAMIC_RELOCATE (main_map, 0, 0, 0);
+  ELF_DYNAMIC_RELOCATE (main_map, 0, 0, 0, main_map);
   main_map->l_relocated = 1;
 
   /* Initialize _r_debug.  */
diff --git a/elf/dl-reloc.c b/elf/dl-reloc.c
index e13a672ade..35cab08683 100644
--- a/elf/dl-reloc.c
+++ b/elf/dl-reloc.c
@@ -162,6 +162,39 @@  _dl_nothread_init_static_tls (struct link_map *map)
 }
 #endif /* !THREAD_GSCOPE_IN_TCB */
 
+/* Used by RESOLVE_MAP. _dl_relocate_object is either called at init time or
+ * by dlopen with a global lock, so the variables cannot be accessed
+ * concurrently.  */
+static struct link_map *cur_l;
+static struct r_scope_elem **cur_scope;
+static const char *cur_strtab;
+
+/* This macro is used as a callback from the ELF_DYNAMIC_RELOCATE code.  */
+#define RESOLVE_MAP(ref, version, r_type) \
+    ({ struct link_map *l = cur_l;					      \
+      (ELFW(ST_BIND) ((*ref)->st_info) != STB_LOCAL			      \
+      && __glibc_likely (!dl_symbol_visibility_binds_local_p (*ref)))	      \
+     ? ((__builtin_expect ((*ref) == l->l_lookup_cache.sym, 0)		      \
+	 && elf_machine_type_class (r_type) == l->l_lookup_cache.type_class)  \
+	? (bump_num_cache_relocations (),				      \
+	   (*ref) = l->l_lookup_cache.ret,				      \
+	   l->l_lookup_cache.value)					      \
+	: ({ lookup_t _lr;						      \
+	     int _tc = elf_machine_type_class (r_type);			      \
+	     l->l_lookup_cache.type_class = _tc;			      \
+	     l->l_lookup_cache.sym = (*ref);				      \
+	     const struct r_found_version *v = NULL;			      \
+	     if ((version) != NULL && (version)->hash != 0)		      \
+	       v = (version);						      \
+	     _lr = _dl_lookup_symbol_x (cur_strtab + (*ref)->st_name, l,      \
+					(ref), cur_scope, v, _tc,	      \
+					DL_LOOKUP_ADD_DEPENDENCY, NULL);      \
+	     l->l_lookup_cache.ret = (*ref);				      \
+	     l->l_lookup_cache.value = _lr; }))				      \
+     : l; })
+
+#include "dynamic-link.h"
+
 void
 _dl_relocate_object (struct link_map *l, struct r_scope_elem *scope[],
 		     int reloc_mode, int consider_profiling)
@@ -243,36 +276,11 @@  _dl_relocate_object (struct link_map *l, struct r_scope_elem *scope[],
   {
     /* Do the actual relocation of the object's GOT and other data.  */
 
-    /* String table object symbols.  */
-    const char *strtab = (const void *) D_PTR (l, l_info[DT_STRTAB]);
+    cur_l = l;
+    cur_scope = scope;
+    cur_strtab = (const void *) D_PTR (cur_l, l_info[DT_STRTAB]);
 
-    /* This macro is used as a callback from the ELF_DYNAMIC_RELOCATE code.  */
-#define RESOLVE_MAP(ref, version, r_type) \
-    ((ELFW(ST_BIND) ((*ref)->st_info) != STB_LOCAL			      \
-      && __glibc_likely (!dl_symbol_visibility_binds_local_p (*ref)))	      \
-     ? ((__builtin_expect ((*ref) == l->l_lookup_cache.sym, 0)		      \
-	 && elf_machine_type_class (r_type) == l->l_lookup_cache.type_class)  \
-	? (bump_num_cache_relocations (),				      \
-	   (*ref) = l->l_lookup_cache.ret,				      \
-	   l->l_lookup_cache.value)					      \
-	: ({ lookup_t _lr;						      \
-	     int _tc = elf_machine_type_class (r_type);			      \
-	     l->l_lookup_cache.type_class = _tc;			      \
-	     l->l_lookup_cache.sym = (*ref);				      \
-	     const struct r_found_version *v = NULL;			      \
-	     if ((version) != NULL && (version)->hash != 0)		      \
-	       v = (version);						      \
-	     _lr = _dl_lookup_symbol_x (strtab + (*ref)->st_name, l, (ref),   \
-					scope, v, _tc,			      \
-					DL_LOOKUP_ADD_DEPENDENCY	      \
-					| DL_LOOKUP_FOR_RELOCATE, NULL);      \
-	     l->l_lookup_cache.ret = (*ref);				      \
-	     l->l_lookup_cache.value = _lr; }))				      \
-     : l)
-
-#include "dynamic-link.h"
-
-    ELF_DYNAMIC_RELOCATE (l, lazy, consider_profiling, skip_ifunc);
+    ELF_DYNAMIC_RELOCATE (l, lazy, consider_profiling, skip_ifunc, NULL);
 
 #ifndef PROF
     if (__glibc_unlikely (consider_profiling)
diff --git a/elf/do-rel.h b/elf/do-rel.h
index 321ac2b359..19fc6c085e 100644
--- a/elf/do-rel.h
+++ b/elf/do-rel.h
@@ -37,11 +37,11 @@ 
    relocations; they should be set up to call _dl_runtime_resolve, rather
    than fully resolved now.  */
 
-auto inline void __attribute__ ((always_inline))
+static inline void __attribute__ ((always_inline))
 elf_dynamic_do_Rel (struct link_map *map,
 		    ElfW(Addr) reladdr, ElfW(Addr) relsize,
 		    __typeof (((ElfW(Dyn) *) 0)->d_un.d_val) nrelative,
-		    int lazy, int skip_ifunc)
+		    int lazy, int skip_ifunc, struct link_map *boot_map)
 {
   const ElfW(Rel) *r = (const void *) reladdr;
   const ElfW(Rel) *end = (const void *) (reladdr + relsize);
@@ -136,7 +136,7 @@  elf_dynamic_do_Rel (struct link_map *map,
 	      ElfW(Half) ndx = version[ELFW(R_SYM) (r->r_info)] & 0x7fff;
 	      elf_machine_rel (map, r, &symtab[ELFW(R_SYM) (r->r_info)],
 			       &map->l_versions[ndx],
-			       (void *) (l_addr + r->r_offset), skip_ifunc);
+			       (void *) (l_addr + r->r_offset), skip_ifunc, boot_map);
 	    }
 
 #if defined ELF_MACHINE_IRELATIVE && !defined RTLD_BOOTSTRAP
@@ -150,7 +150,7 @@  elf_dynamic_do_Rel (struct link_map *map,
 				   &symtab[ELFW(R_SYM) (r2->r_info)],
 				   &map->l_versions[ndx],
 				   (void *) (l_addr + r2->r_offset),
-				   skip_ifunc);
+				   skip_ifunc, boot_map);
 		}
 #endif
 	}
@@ -168,7 +168,7 @@  elf_dynamic_do_Rel (struct link_map *map,
 	    else
 # endif
 	      elf_machine_rel (map, r, &symtab[ELFW(R_SYM) (r->r_info)], NULL,
-			       (void *) (l_addr + r->r_offset), skip_ifunc);
+			       (void *) (l_addr + r->r_offset), skip_ifunc, boot_map);
 
 # ifdef ELF_MACHINE_IRELATIVE
 	  if (r2 != NULL)
@@ -176,7 +176,7 @@  elf_dynamic_do_Rel (struct link_map *map,
 	      if (ELFW(R_TYPE) (r2->r_info) == ELF_MACHINE_IRELATIVE)
 		elf_machine_rel (map, r2, &symtab[ELFW(R_SYM) (r2->r_info)],
 				 NULL, (void *) (l_addr + r2->r_offset),
-				 skip_ifunc);
+				 skip_ifunc, boot_map);
 # endif
 	}
 #endif
diff --git a/elf/dynamic-link.h b/elf/dynamic-link.h
index 3eb24ba3a6..6d71427675 100644
--- a/elf/dynamic-link.h
+++ b/elf/dynamic-link.h
@@ -59,30 +59,30 @@  int _dl_try_allocate_static_tls (struct link_map *map, bool optional)
    copying memory, breaking the very code written to handle the
    unaligned cases.  */
 # if ! ELF_MACHINE_NO_REL
-auto inline void __attribute__((always_inline))
+static inline void __attribute__((always_inline))
 elf_machine_rel (struct link_map *map, const ElfW(Rel) *reloc,
 		 const ElfW(Sym) *sym, const struct r_found_version *version,
-		 void *const reloc_addr, int skip_ifunc);
-auto inline void __attribute__((always_inline))
+		 void *const reloc_addr, int skip_ifunc, struct link_map *boot_map);
+static inline void __attribute__((always_inline))
 elf_machine_rel_relative (ElfW(Addr) l_addr, const ElfW(Rel) *reloc,
 			  void *const reloc_addr);
 # endif
 # if ! ELF_MACHINE_NO_RELA
-auto inline void __attribute__((always_inline))
+static inline void __attribute__((always_inline))
 elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
 		  const ElfW(Sym) *sym, const struct r_found_version *version,
-		  void *const reloc_addr, int skip_ifunc);
-auto inline void __attribute__((always_inline))
+		  void *const reloc_addr, int skip_ifunc, struct link_map *boot_map);
+static inline void __attribute__((always_inline))
 elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
 			   void *const reloc_addr);
 # endif
 # if ELF_MACHINE_NO_RELA || defined ELF_MACHINE_PLT_REL
-auto inline void __attribute__((always_inline))
+static inline void __attribute__((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      ElfW(Addr) l_addr, const ElfW(Rel) *reloc,
 		      int skip_ifunc);
 # else
-auto inline void __attribute__((always_inline))
+static inline void __attribute__((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
 		      int skip_ifunc);
@@ -114,7 +114,7 @@  elf_machine_lazy_rel (struct link_map *map,
    consumes precisely the very end of the DT_REL*, or DT_JMPREL and DT_REL*
    are completely separate and there is a gap between them.  */
 
-# define _ELF_DYNAMIC_DO_RELOC(RELOC, reloc, map, do_lazy, skip_ifunc, test_rel) \
+# define _ELF_DYNAMIC_DO_RELOC(RELOC, reloc, map, do_lazy, skip_ifunc, test_rel, boot_map) \
   do {									      \
     struct { ElfW(Addr) start, size;					      \
 	     __typeof (((ElfW(Dyn) *) 0)->d_un.d_val) nrelative; int lazy; }  \
@@ -153,7 +153,7 @@  elf_machine_lazy_rel (struct link_map *map,
 									      \
     if (ELF_DURING_STARTUP)						      \
       elf_dynamic_do_##reloc ((map), ranges[0].start, ranges[0].size,	      \
-			      ranges[0].nrelative, 0, skip_ifunc);	      \
+			      ranges[0].nrelative, 0, skip_ifunc, boot_map);  \
     else								      \
       {									      \
 	int ranges_index;						      \
@@ -163,7 +163,7 @@  elf_machine_lazy_rel (struct link_map *map,
 				  ranges[ranges_index].size,		      \
 				  ranges[ranges_index].nrelative,	      \
 				  ranges[ranges_index].lazy,		      \
-				  skip_ifunc);				      \
+				  skip_ifunc, boot_map);		      \
       }									      \
   } while (0)
 
@@ -175,29 +175,29 @@  elf_machine_lazy_rel (struct link_map *map,
 
 # if ! ELF_MACHINE_NO_REL
 #  include "do-rel.h"
-#  define ELF_DYNAMIC_DO_REL(map, lazy, skip_ifunc) \
-  _ELF_DYNAMIC_DO_RELOC (REL, Rel, map, lazy, skip_ifunc, _ELF_CHECK_REL)
+#  define ELF_DYNAMIC_DO_REL(map, lazy, skip_ifunc, boot_map)			\
+  _ELF_DYNAMIC_DO_RELOC (REL, Rel, map, lazy, skip_ifunc, _ELF_CHECK_REL, boot_map)
 # else
-#  define ELF_DYNAMIC_DO_REL(map, lazy, skip_ifunc) /* Nothing to do.  */
+#  define ELF_DYNAMIC_DO_REL(map, lazy, skip_ifunc, boot_map) /* Nothing to do.  */
 # endif
 
 # if ! ELF_MACHINE_NO_RELA
 #  define DO_RELA
 #  include "do-rel.h"
-#  define ELF_DYNAMIC_DO_RELA(map, lazy, skip_ifunc) \
-  _ELF_DYNAMIC_DO_RELOC (RELA, Rela, map, lazy, skip_ifunc, _ELF_CHECK_REL)
+#  define ELF_DYNAMIC_DO_RELA(map, lazy, skip_ifunc, boot_map)			\
+  _ELF_DYNAMIC_DO_RELOC (RELA, Rela, map, lazy, skip_ifunc, _ELF_CHECK_REL, boot_map)
 # else
-#  define ELF_DYNAMIC_DO_RELA(map, lazy, skip_ifunc) /* Nothing to do.  */
+#  define ELF_DYNAMIC_DO_RELA(map, lazy, skip_ifunc, boot_map) /* Nothing to do.  */
 # endif
 
 /* This can't just be an inline function because GCC is too dumb
    to inline functions containing inlines themselves.  */
-# define ELF_DYNAMIC_RELOCATE(map, lazy, consider_profile, skip_ifunc) \
+# define ELF_DYNAMIC_RELOCATE(map, lazy, consider_profile, skip_ifunc, boot_map) \
   do {									      \
     int edr_lazy = elf_machine_runtime_setup ((map), (lazy),		      \
 					      (consider_profile));	      \
-    ELF_DYNAMIC_DO_REL ((map), edr_lazy, skip_ifunc);			      \
-    ELF_DYNAMIC_DO_RELA ((map), edr_lazy, skip_ifunc);			      \
+    ELF_DYNAMIC_DO_REL ((map), edr_lazy, skip_ifunc, boot_map);		      \
+    ELF_DYNAMIC_DO_RELA ((map), edr_lazy, skip_ifunc, boot_map);	      \
   } while (0)
 
 #endif
diff --git a/elf/get-dynamic-info.h b/elf/get-dynamic-info.h
index d8ec32377d..5827ae6e24 100644
--- a/elf/get-dynamic-info.h
+++ b/elf/get-dynamic-info.h
@@ -19,6 +19,8 @@ 
 /* This file is included multiple times and therefore lacks a header
    file inclusion guard.  */
 
+#ifndef _GET_DYNAMIC_INFO_H
+
 #include <assert.h>
 #include <libc-diag.h>
 
@@ -180,3 +182,6 @@  elf_get_dynamic_info (struct link_map *l, ElfW(Dyn) *temp)
     info[DT_RPATH] = NULL;
 #endif
 }
+
+#define _GET_DYNAMIC_INFO_H
+#endif
diff --git a/elf/rtld.c b/elf/rtld.c
index 878e6480f4..c62efa7308 100644
--- a/elf/rtld.c
+++ b/elf/rtld.c
@@ -499,13 +499,9 @@  _dl_start_final (void *arg, struct dl_start_final_info *info)
   return start_addr;
 }
 
-static ElfW(Addr) __attribute_used__
-_dl_start (void *arg)
-{
 #ifdef DONT_USE_BOOTSTRAP_MAP
 # define bootstrap_map GL(dl_rtld_map)
 #else
-  struct dl_start_final_info info;
 # define bootstrap_map info.l
 #endif
 
@@ -518,9 +514,13 @@  _dl_start (void *arg)
 #define RESOLVE_MAP(sym, version, flags) BOOTSTRAP_MAP
 #include "dynamic-link.h"
 
+static ElfW(Addr) __attribute_used__
+_dl_start (void *arg)
+{
 #ifdef DONT_USE_BOOTSTRAP_MAP
   rtld_timer_start (&start_time);
 #else
+  struct dl_start_final_info info;
   rtld_timer_start (&info.start_time);
 #endif
 
@@ -561,7 +561,7 @@  _dl_start (void *arg)
       /* Relocate ourselves so we can do normal function calls and
 	 data access using the global offset table.  */
 
-      ELF_DYNAMIC_RELOCATE (&bootstrap_map, 0, 0, 0);
+      ELF_DYNAMIC_RELOCATE (&bootstrap_map, 0, 0, 0, &bootstrap_map);
     }
   bootstrap_map.l_relocated = 1;
 
diff --git a/sysdeps/aarch64/dl-machine.h b/sysdeps/aarch64/dl-machine.h
index 3e10cb462f..ab9e5c4d4b 100644
--- a/sysdeps/aarch64/dl-machine.h
+++ b/sysdeps/aarch64/dl-machine.h
@@ -237,11 +237,12 @@  elf_machine_plt_value (struct link_map *map,
 
 #ifdef RESOLVE_MAP
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
 		  const ElfW(Sym) *sym, const struct r_found_version *version,
-		  void *const reloc_addr_arg, int skip_ifunc)
+		  void *const reloc_addr_arg, int skip_ifunc,
+		  struct link_map *boot_map)
 {
   ElfW(Addr) *const reloc_addr = reloc_addr_arg;
   const unsigned int r_type = ELFW (R_TYPE) (reloc->r_info);
@@ -253,7 +254,11 @@  elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
   else
     {
       const ElfW(Sym) *const refsym = sym;
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+      struct link_map *sym_map = boot_map;
+#else
       struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
       ElfW(Addr) value = SYMBOL_ADDRESS (sym_map, sym, true);
 
       if (sym != NULL
@@ -377,7 +382,7 @@  elf_machine_rela_relative (ElfW(Addr) l_addr,
   *reloc_addr = l_addr + reloc->r_addend;
 }
 
-inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      ElfW(Addr) l_addr,
@@ -407,7 +412,7 @@  elf_machine_lazy_rel (struct link_map *map,
 		  version = &map->l_versions[vernum[symndx] & 0x7fff];
 		}
 	      elf_machine_rela (map, reloc, sym, version, reloc_addr,
-				skip_ifunc);
+				skip_ifunc, NULL);
 	      return;
 	    }
 	}
@@ -433,7 +438,8 @@  elf_machine_lazy_rel (struct link_map *map,
 
       /* Always initialize TLS descriptors completely, because lazy
 	 initialization requires synchronization at every TLS access.  */
-      elf_machine_rela (map, reloc, sym, version, reloc_addr, skip_ifunc);
+      elf_machine_rela (map, reloc, sym, version, reloc_addr, skip_ifunc,
+                        NULL);
     }
   else if (__glibc_unlikely (r_type == AARCH64_R(IRELATIVE)))
     {
diff --git a/sysdeps/alpha/dl-machine.h b/sysdeps/alpha/dl-machine.h
index 2cd2213d9a..dc22baa68a 100644
--- a/sysdeps/alpha/dl-machine.h
+++ b/sysdeps/alpha/dl-machine.h
@@ -361,14 +361,15 @@  elf_machine_plt_value (struct link_map *map, const Elf64_Rela *reloc,
 
 /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
    MAP is the object containing the reloc.  */
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela (struct link_map *map,
 		  const Elf64_Rela *reloc,
 		  const Elf64_Sym *sym,
 		  const struct r_found_version *version,
 		  void *const reloc_addr_arg,
-		  int skip_ifunc)
+		  int skip_ifunc,
+		  struct link_map *boot_map)
 {
   Elf64_Addr *const reloc_addr = reloc_addr_arg;
   unsigned long int const r_type = ELF64_R_TYPE (reloc->r_info);
@@ -411,7 +412,11 @@  elf_machine_rela (struct link_map *map,
       return;
   else
     {
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+      struct link_map *sym_map = boot_map;
+#else
       struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
       Elf64_Addr sym_value;
       Elf64_Addr sym_raw_value;
 
@@ -489,7 +494,7 @@  elf_machine_rela (struct link_map *map,
    can be skipped.  */
 #define ELF_MACHINE_REL_RELATIVE 1
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
 			   void *const reloc_addr_arg)
@@ -506,7 +511,7 @@  elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
   memcpy (reloc_addr_arg, &reloc_addr_val, 8);
 }
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      Elf64_Addr l_addr, const Elf64_Rela *reloc,
diff --git a/sysdeps/arc/dl-machine.h b/sysdeps/arc/dl-machine.h
index e6ce7f0ff6..1054530321 100644
--- a/sysdeps/arc/dl-machine.h
+++ b/sysdeps/arc/dl-machine.h
@@ -228,11 +228,11 @@  elf_machine_fixup_plt (struct link_map *map, lookup_t t,
 
 #ifdef RESOLVE_MAP
 
-inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
                   const ElfW(Sym) *sym, const struct r_found_version *version,
-                  void *const reloc_addr_arg, int skip_ifunc)
+                  void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
 {
   ElfW(Addr) r_info = reloc->r_info;
   const unsigned long int r_type = ELFW (R_TYPE) (r_info);
@@ -245,7 +245,11 @@  elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
   else
     {
       const ElfW(Sym) *const refsym = sym;
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+      struct link_map *sym_map = boot_map;
+#else
       struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
       ElfW(Addr) value = SYMBOL_ADDRESS (sym_map, sym, true);
 
       switch (r_type)
diff --git a/sysdeps/arm/dl-machine.h b/sysdeps/arm/dl-machine.h
index eb13cb8b57..84f84a1965 100644
--- a/sysdeps/arm/dl-machine.h
+++ b/sysdeps/arm/dl-machine.h
@@ -276,7 +276,7 @@  elf_machine_plt_value (struct link_map *map, const Elf32_Rel *reloc,
 
 #ifdef RESOLVE_MAP
 /* Handle a PC24 reloc, including the out-of-range case.  */
-auto void
+static void
 relocate_pc24 (struct link_map *map, Elf32_Addr value,
                Elf32_Addr *const reloc_addr, Elf32_Sword addend)
 {
@@ -330,11 +330,11 @@  relocate_pc24 (struct link_map *map, Elf32_Addr value,
 /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
    MAP is the object containing the reloc.  */
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rel (struct link_map *map, const Elf32_Rel *reloc,
 		 const Elf32_Sym *sym, const struct r_found_version *version,
-		 void *const reloc_addr_arg, int skip_ifunc)
+		 void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
 {
   Elf32_Addr *const reloc_addr = reloc_addr_arg;
   const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
@@ -364,7 +364,11 @@  elf_machine_rel (struct link_map *map, const Elf32_Rel *reloc,
 #endif
     {
       const Elf32_Sym *const refsym = sym;
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+      struct link_map *sym_map = boot_map;
+#else
       struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
       Elf32_Addr value = SYMBOL_ADDRESS (sym_map, sym, true);
 
       if (sym != NULL
@@ -508,11 +512,11 @@  elf_machine_rel (struct link_map *map, const Elf32_Rel *reloc,
 }
 
 # ifndef RTLD_BOOTSTRAP
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
 		  const Elf32_Sym *sym, const struct r_found_version *version,
-		  void *const reloc_addr_arg, int skip_ifunc)
+		  void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
 {
   Elf32_Addr *const reloc_addr = reloc_addr_arg;
   const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
@@ -601,7 +605,7 @@  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
 }
 # endif
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rel_relative (Elf32_Addr l_addr, const Elf32_Rel *reloc,
 			  void *const reloc_addr_arg)
@@ -611,7 +615,7 @@  elf_machine_rel_relative (Elf32_Addr l_addr, const Elf32_Rel *reloc,
 }
 
 # ifndef RTLD_BOOTSTRAP
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
 			   void *const reloc_addr_arg)
@@ -621,7 +625,7 @@  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
 }
 # endif
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      Elf32_Addr l_addr, const Elf32_Rel *reloc,
@@ -653,7 +657,7 @@  elf_machine_lazy_rel (struct link_map *map,
 
       /* Always initialize TLS descriptors completely, because lazy
 	 initialization requires synchronization at every TLS access.  */
-      elf_machine_rel (map, reloc, sym, version, reloc_addr, skip_ifunc);
+      elf_machine_rel (map, reloc, sym, version, reloc_addr, skip_ifunc, NULL);
     }
   else
     _dl_reloc_bad_type (map, r_type, 1);
diff --git a/sysdeps/csky/dl-machine.h b/sysdeps/csky/dl-machine.h
index b08f06d74c..8702fe9917 100644
--- a/sysdeps/csky/dl-machine.h
+++ b/sysdeps/csky/dl-machine.h
@@ -215,10 +215,10 @@  elf_machine_plt_value (struct link_map *map, const Elf32_Rela *reloc,
 /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
    MAP is the object containing the reloc.  */
 
-auto inline void __attribute__ ((unused, always_inline))
+static inline void __attribute__ ((unused, always_inline))
 elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
 		  const Elf32_Sym *sym, const struct r_found_version *version,
-		  void *const reloc_addr_arg, int skip_ifunc)
+		  void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
 {
   Elf32_Addr *const reloc_addr = reloc_addr_arg;
   const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
@@ -230,7 +230,11 @@  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
   else
     {
       const Elf32_Sym *const refsym = sym;
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+      struct link_map *sym_map = boot_map;
+#else
       struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
       ElfW(Addr) value = SYMBOL_ADDRESS (sym_map, sym, true);
       opcode16_addr = (unsigned short *)reloc_addr;
 
@@ -331,7 +335,7 @@  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
     }
 }
 
-auto inline void __attribute__ ((unused, always_inline))
+static inline void __attribute__ ((unused, always_inline))
 elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
 			   void *const reloc_addr_arg)
 {
@@ -339,7 +343,7 @@  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
   *reloc_addr = l_addr + reloc->r_addend;
 }
 
-auto inline void __attribute__ ((unused, always_inline))
+static inline void __attribute__ ((unused, always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      Elf32_Addr l_addr, const Elf32_Rela *reloc,
 		      int skip_ifunc)
diff --git a/sysdeps/hppa/dl-machine.h b/sysdeps/hppa/dl-machine.h
index f048fd2072..4aa159cfbd 100644
--- a/sysdeps/hppa/dl-machine.h
+++ b/sysdeps/hppa/dl-machine.h
@@ -549,13 +549,13 @@  dl_platform_init (void)
   (  (((as14) & 0x1fff) << 1) \
    | (((as14) & 0x2000) >> 13))
 
-auto void __attribute__((always_inline))
+static void __attribute__((always_inline))
 elf_machine_rela (struct link_map *map,
 		  const Elf32_Rela *reloc,
 		  const Elf32_Sym *sym,
 		  const struct r_found_version *version,
 		  void *const reloc_addr_arg,
-		  int skip_ifunc)
+		  int skip_ifunc, struct link_map *boot_map)
 {
   Elf32_Addr *const reloc_addr = reloc_addr_arg;
   const Elf32_Sym *const refsym = sym;
@@ -581,7 +581,7 @@  elf_machine_rela (struct link_map *map,
 # ifdef RTLD_BOOTSTRAP
   /* RESOLVE_MAP in rtld.c doesn't have the local sym test.  */
   sym_map = (ELF32_ST_BIND (sym->st_info) != STB_LOCAL
-	     ? RESOLVE_MAP (&sym, version, r_type) : map);
+	     ? boot_map : map);
 # else
   sym_map = RESOLVE_MAP (&sym, version, r_type);
 # endif
@@ -741,7 +741,7 @@  elf_machine_rela (struct link_map *map,
 
 /* hppa doesn't have an R_PARISC_RELATIVE reloc, but uses relocs with
    ELF32_R_SYM (info) == 0 for a similar purpose.  */
-auto void __attribute__((always_inline))
+static void __attribute__((always_inline))
 elf_machine_rela_relative (Elf32_Addr l_addr,
 			   const Elf32_Rela *reloc,
 			   void *const reloc_addr_arg)
@@ -794,7 +794,7 @@  elf_machine_rela_relative (Elf32_Addr l_addr,
   *reloc_addr = value;
 }
 
-auto void __attribute__((always_inline))
+static void __attribute__((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      Elf32_Addr l_addr, const Elf32_Rela *reloc,
 		      int skip_ifunc)
diff --git a/sysdeps/i386/dl-machine.h b/sysdeps/i386/dl-machine.h
index 590b41d8d7..9028e53ff1 100644
--- a/sysdeps/i386/dl-machine.h
+++ b/sysdeps/i386/dl-machine.h
@@ -291,11 +291,11 @@  elf_machine_plt_value (struct link_map *map, const Elf32_Rel *reloc,
 /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
    MAP is the object containing the reloc.  */
 
-auto inline void
+static inline void
 __attribute ((always_inline))
 elf_machine_rel (struct link_map *map, const Elf32_Rel *reloc,
 		 const Elf32_Sym *sym, const struct r_found_version *version,
-		 void *const reloc_addr_arg, int skip_ifunc)
+		 void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
 {
   Elf32_Addr *const reloc_addr = reloc_addr_arg;
   const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
@@ -498,11 +498,11 @@  and creates an unsatisfiable circular dependency.\n",
 }
 
 # ifndef RTLD_BOOTSTRAP
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
 		  const Elf32_Sym *sym, const struct r_found_version *version,
-		  void *const reloc_addr_arg, int skip_ifunc)
+		  void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
 {
   Elf32_Addr *const reloc_addr = reloc_addr_arg;
   const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
@@ -647,7 +647,7 @@  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
 }
 # endif	/* !RTLD_BOOTSTRAP */
 
-auto inline void
+static inline void
 __attribute ((always_inline))
 elf_machine_rel_relative (Elf32_Addr l_addr, const Elf32_Rel *reloc,
 			  void *const reloc_addr_arg)
@@ -658,7 +658,7 @@  elf_machine_rel_relative (Elf32_Addr l_addr, const Elf32_Rel *reloc,
 }
 
 # ifndef RTLD_BOOTSTRAP
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
 			   void *const reloc_addr_arg)
@@ -668,7 +668,7 @@  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
 }
 # endif	/* !RTLD_BOOTSTRAP */
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      Elf32_Addr l_addr, const Elf32_Rel *reloc,
@@ -707,12 +707,12 @@  elf_machine_lazy_rel (struct link_map *map,
 	  ElfW(Half) ndx = version[ELFW(R_SYM) (r->r_info)] & 0x7fff;
 	  elf_machine_rel (map, r, &symtab[ELFW(R_SYM) (r->r_info)],
 			   &map->l_versions[ndx],
-			   (void *) (l_addr + r->r_offset), skip_ifunc);
+			   (void *) (l_addr + r->r_offset), skip_ifunc, NULL);
 	}
 # ifndef RTLD_BOOTSTRAP
       else
 	elf_machine_rel (map, r, &symtab[ELFW(R_SYM) (r->r_info)], NULL,
-			 (void *) (l_addr + r->r_offset), skip_ifunc);
+			 (void *) (l_addr + r->r_offset), skip_ifunc, NULL);
 # endif
     }
   else if (__glibc_unlikely (r_type == R_386_IRELATIVE))
@@ -728,7 +728,7 @@  elf_machine_lazy_rel (struct link_map *map,
 
 # ifndef RTLD_BOOTSTRAP
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_lazy_rela (struct link_map *map,
 		       Elf32_Addr l_addr, const Elf32_Rela *reloc,
@@ -754,7 +754,7 @@  elf_machine_lazy_rela (struct link_map *map,
 
       /* Always initialize TLS descriptors completely at load time, in
 	 case static TLS is allocated for it that requires locking.  */
-      elf_machine_rela (map, reloc, sym, version, reloc_addr, skip_ifunc);
+      elf_machine_rela (map, reloc, sym, version, reloc_addr, skip_ifunc, NULL);
     }
   else if (__glibc_unlikely (r_type == R_386_IRELATIVE))
     {
diff --git a/sysdeps/ia64/dl-machine.h b/sysdeps/ia64/dl-machine.h
index 4403e7767a..8390148134 100644
--- a/sysdeps/ia64/dl-machine.h
+++ b/sysdeps/ia64/dl-machine.h
@@ -371,14 +371,14 @@  elf_machine_plt_value (struct link_map *map, const Elf64_Rela *reloc,
 
 /* Perform the relocation specified by RELOC and SYM (which is fully
    resolved).  MAP is the object containing the reloc.  */
-auto inline void
+static inline void
 __attribute ((always_inline))
 elf_machine_rela (struct link_map *map,
 		  const Elf64_Rela *reloc,
 		  const Elf64_Sym *sym,
 		  const struct r_found_version *version,
 		  void *const reloc_addr_arg,
-		  int skip_ifunc)
+		  int skip_ifunc, struct link_map *boot_map)
 {
   Elf64_Addr *const reloc_addr = reloc_addr_arg;
   const unsigned long int r_type = ELF64_R_TYPE (reloc->r_info);
@@ -414,10 +414,14 @@  elf_machine_rela (struct link_map *map,
       return;
   else
     {
-      struct link_map *sym_map;
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+      struct link_map *sym_map = boot_map;
+#else
+      struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
 
       /* RESOLVE_MAP() will return NULL if it fail to locate the symbol.  */
-      if ((sym_map = RESOLVE_MAP (&sym, version, r_type)))
+      if (sym_map != NULL)
 	{
 	  value = SYMBOL_ADDRESS (sym_map, sym, true) + reloc->r_addend;
 
@@ -476,7 +480,7 @@  elf_machine_rela (struct link_map *map,
    can be skipped.  */
 #define ELF_MACHINE_REL_RELATIVE 1
 
-auto inline void
+static inline void
 __attribute ((always_inline))
 elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
 			   void *const reloc_addr_arg)
@@ -489,7 +493,7 @@  elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
 }
 
 /* Perform a RELATIVE reloc on the .got entry that transfers to the .plt.  */
-auto inline void
+static inline void
 __attribute ((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      Elf64_Addr l_addr, const Elf64_Rela *reloc,
diff --git a/sysdeps/m68k/dl-machine.h b/sysdeps/m68k/dl-machine.h
index 86a8c67e2a..9c7657c636 100644
--- a/sysdeps/m68k/dl-machine.h
+++ b/sysdeps/m68k/dl-machine.h
@@ -215,10 +215,10 @@  elf_machine_plt_value (struct link_map *map, const Elf32_Rela *reloc,
 /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
    MAP is the object containing the reloc.  */
 
-auto inline void __attribute__ ((unused, always_inline))
+static inline void __attribute__ ((unused, always_inline))
 elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
 		  const Elf32_Sym *sym, const struct r_found_version *version,
-		  void *const reloc_addr_arg, int skip_ifunc)
+		  void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
 {
   Elf32_Addr *const reloc_addr = reloc_addr_arg;
   const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
@@ -228,7 +228,11 @@  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
   else
     {
       const Elf32_Sym *const refsym = sym;
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+      struct link_map *sym_map = boot_map;
+#else
       struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
       Elf32_Addr value = SYMBOL_ADDRESS (sym_map, sym, true);
 
       switch (r_type)
@@ -303,7 +307,7 @@  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
     }
 }
 
-auto inline void __attribute__ ((unused, always_inline))
+static inline void __attribute__ ((unused, always_inline))
 elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
 			   void *const reloc_addr_arg)
 {
@@ -311,7 +315,7 @@  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
   *reloc_addr = l_addr + reloc->r_addend;
 }
 
-auto inline void __attribute__ ((unused, always_inline))
+static inline void __attribute__ ((unused, always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      Elf32_Addr l_addr, const Elf32_Rela *reloc,
 		      int skip_ifunc)
diff --git a/sysdeps/microblaze/dl-machine.h b/sysdeps/microblaze/dl-machine.h
index e460f6f195..3e54676afe 100644
--- a/sysdeps/microblaze/dl-machine.h
+++ b/sysdeps/microblaze/dl-machine.h
@@ -207,10 +207,10 @@  elf_machine_plt_value (struct link_map *map, const Elf32_Rela *reloc,
     ((unsigned short *)(rel_addr))[3] = (val) & 0xffff; \
   } while (0)
 
-auto inline void __attribute__ ((always_inline))
+static inline void __attribute__ ((always_inline))
 elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
 		  const Elf32_Sym *sym, const struct r_found_version *version,
-		  void *const reloc_addr_arg, int skip_ifunc)
+		  void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
 {
   Elf32_Addr *const reloc_addr = reloc_addr_arg;
   const int r_type = ELF32_R_TYPE (reloc->r_info);
@@ -222,7 +222,11 @@  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
   else
     {
       const Elf32_Sym *const refsym = sym;
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+      struct link_map *sym_map = boot_map;
+#else
       struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
       Elf32_Addr value = SYMBOL_ADDRESS (sym_map, sym, true);
 
       value += reloc->r_addend;
@@ -277,7 +281,7 @@  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
     }
 }
 
-auto inline void
+static inline void
 elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
 			   void *const reloc_addr_arg)
 {
@@ -285,7 +289,7 @@  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
   PUT_REL_64 (reloc_addr, l_addr + reloc->r_addend);
 }
 
-auto inline void
+static inline void
 elf_machine_lazy_rel (struct link_map *map,
 		      Elf32_Addr l_addr, const Elf32_Rela *reloc,
 		      int skip_ifunc)
diff --git a/sysdeps/mips/dl-machine.h b/sysdeps/mips/dl-machine.h
index d9c6d33d0c..2293ea0b3e 100644
--- a/sysdeps/mips/dl-machine.h
+++ b/sysdeps/mips/dl-machine.h
@@ -475,11 +475,12 @@  elf_machine_plt_value (struct link_map *map, const ElfW(Rel) *reloc,
    by RELOC_ADDR.  SYM is the relocation symbol specified by R_INFO and
    MAP is the object containing the reloc.  */
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_reloc (struct link_map *map, ElfW(Addr) r_info,
 		   const ElfW(Sym) *sym, const struct r_found_version *version,
-		   void *reloc_addr, ElfW(Addr) r_addend, int inplace_p)
+		   void *reloc_addr, ElfW(Addr) r_addend, int inplace_p,
+		   struct link_map *boot_map)
 {
   const unsigned long int r_type = ELFW(R_TYPE) (r_info);
   ElfW(Addr) *addr_field = (ElfW(Addr) *) reloc_addr;
@@ -507,7 +508,11 @@  elf_machine_reloc (struct link_map *map, ElfW(Addr) r_info,
     case R_MIPS_TLS_TPREL32:
 # endif
       {
-	struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+      struct link_map *sym_map = boot_map;
+#else
+      struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
 
 	switch (r_type)
 	  {
@@ -647,7 +652,11 @@  elf_machine_reloc (struct link_map *map, ElfW(Addr) r_info,
 	  _dl_signal_error (0, map->l_name, NULL,
 			    "found jump slot relocation with non-zero addend");
 
-	sym_map = RESOLVE_MAP (&sym, version, r_type);
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+        sym_map = boot_map;
+#else
+        sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
 	value = SYMBOL_ADDRESS (sym_map, sym, true);
 	*addr_field = value;
 
@@ -661,7 +670,11 @@  elf_machine_reloc (struct link_map *map, ElfW(Addr) r_info,
 	ElfW(Addr) value;
 
 	/* Calculate the address of the symbol.  */
-	sym_map = RESOLVE_MAP (&sym, version, r_type);
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+        sym_map = boot_map;
+#else
+        sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
 	value = SYMBOL_ADDRESS (sym_map, sym, true);
 
 	if (__builtin_expect (sym == NULL, 0))
@@ -708,16 +721,16 @@  elf_machine_reloc (struct link_map *map, ElfW(Addr) r_info,
 /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
    MAP is the object containing the reloc.  */
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rel (struct link_map *map, const ElfW(Rel) *reloc,
 		 const ElfW(Sym) *sym, const struct r_found_version *version,
-		 void *const reloc_addr, int skip_ifunc)
+		 void *const reloc_addr, int skip_ifunc, struct link_map *boot_map)
 {
-  elf_machine_reloc (map, reloc->r_info, sym, version, reloc_addr, 0, 1);
+  elf_machine_reloc (map, reloc->r_info, sym, version, reloc_addr, 0, 1, boot_map);
 }
 
-auto inline void
+static inline void
 __attribute__((always_inline))
 elf_machine_rel_relative (ElfW(Addr) l_addr, const ElfW(Rel) *reloc,
 			  void *const reloc_addr)
@@ -725,7 +738,7 @@  elf_machine_rel_relative (ElfW(Addr) l_addr, const ElfW(Rel) *reloc,
   /* XXX Nothing to do.  There is no relative relocation, right?  */
 }
 
-auto inline void
+static inline void
 __attribute__((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      ElfW(Addr) l_addr, const ElfW(Rel) *reloc,
@@ -748,17 +761,17 @@  elf_machine_lazy_rel (struct link_map *map,
     _dl_reloc_bad_type (map, r_type, 1);
 }
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
 		  const ElfW(Sym) *sym, const struct r_found_version *version,
-		  void *const reloc_addr, int skip_ifunc)
+		  void *const reloc_addr, int skip_ifunc, struct link_map *boot_map)
 {
   elf_machine_reloc (map, reloc->r_info, sym, version, reloc_addr,
-		     reloc->r_addend, 0);
+		     reloc->r_addend, 0, boot_map);
 }
 
-auto inline void
+static inline void
 __attribute__((always_inline))
 elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
 			   void *const reloc_addr)
@@ -767,7 +780,7 @@  elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
 
 #ifndef RTLD_BOOTSTRAP
 /* Relocate GOT. */
-auto inline void
+static inline void
 __attribute__((always_inline))
 elf_machine_got_rel (struct link_map *map, int lazy)
 {
@@ -868,7 +881,7 @@  elf_machine_got_rel (struct link_map *map, int lazy)
 /* Set up the loaded object described by L so its stub function
    will jump to the on-demand fixup code __dl_runtime_resolve.  */
 
-auto inline int
+static inline int
 __attribute__((always_inline))
 elf_machine_runtime_setup (struct link_map *l, int lazy, int profile)
 {
diff --git a/sysdeps/nios2/dl-machine.h b/sysdeps/nios2/dl-machine.h
index e000cd081f..cd03ea93bd 100644
--- a/sysdeps/nios2/dl-machine.h
+++ b/sysdeps/nios2/dl-machine.h
@@ -234,10 +234,10 @@  elf_machine_plt_value (struct link_map *map, const Elf32_Rela *reloc,
    LOADADDR is the load address of the object; INFO is an array indexed
    by DT_* of the .dynamic section info.  */
 
-auto inline void __attribute__ ((always_inline))
+static inline void __attribute__ ((always_inline))
 elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
                   const ElfW(Sym) *sym, const struct r_found_version *version,
-                  void *const reloc_addr_arg, int skip_ifunc)
+                  void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
 {
   Elf32_Addr *const reloc_addr = reloc_addr_arg;
   const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
@@ -249,7 +249,11 @@  elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
   else
     {
       const Elf32_Sym *const refsym = sym;
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+      struct link_map *sym_map = boot_map;
+#else
       struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
       Elf32_Addr value = SYMBOL_ADDRESS (sym_map, sym, true);
 
       switch (r_type)
@@ -314,7 +318,7 @@  elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
     }
 }
 
-auto inline void __attribute__((always_inline))
+static inline void __attribute__((always_inline))
 elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
 			   void *const reloc_addr_arg)
 {
@@ -322,10 +326,10 @@  elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
   *reloc_addr = l_addr + reloc->r_addend;
 }
 
-auto inline void __attribute__((always_inline))
+static inline void __attribute__((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
-		      int skip_ifunc)
+		      int skip_ifunc, struct link_map *boot_map)
 {
   Elf32_Addr *const reloc_addr = (void *) (l_addr + reloc->r_offset);
   if (ELF32_R_TYPE (reloc->r_info) == R_NIOS2_JUMP_SLOT)
diff --git a/sysdeps/powerpc/powerpc32/dl-machine.h b/sysdeps/powerpc/powerpc32/dl-machine.h
index ced3a7b659..eaf3a1202f 100644
--- a/sysdeps/powerpc/powerpc32/dl-machine.h
+++ b/sysdeps/powerpc/powerpc32/dl-machine.h
@@ -286,10 +286,10 @@  extern void _dl_reloc_overflow (struct link_map *map,
    LOADADDR is the load address of the object; INFO is an array indexed
    by DT_* of the .dynamic section info.  */
 
-auto inline void __attribute__ ((always_inline))
+static inline void __attribute__ ((always_inline))
 elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
 		  const Elf32_Sym *sym, const struct r_found_version *version,
-		  void *const reloc_addr_arg, int skip_ifunc)
+		  void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
 {
   Elf32_Addr *const reloc_addr = reloc_addr_arg;
   const Elf32_Sym *const refsym = sym;
@@ -317,7 +317,11 @@  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
     }
   else
     {
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+      sym_map = boot_map;
+#else
       sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
       value = SYMBOL_ADDRESS (sym_map, sym, true);
     }
   value += reloc->r_addend;
@@ -441,7 +445,7 @@  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
     }
 }
 
-auto inline void __attribute__ ((always_inline))
+static inline void __attribute__ ((always_inline))
 elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
 			   void *const reloc_addr_arg)
 {
@@ -449,7 +453,7 @@  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
   *reloc_addr = l_addr + reloc->r_addend;
 }
 
-auto inline void __attribute__ ((always_inline))
+static inline void __attribute__ ((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      Elf32_Addr l_addr, const Elf32_Rela *reloc,
 		      int skip_ifunc)
diff --git a/sysdeps/powerpc/powerpc64/dl-machine.h b/sysdeps/powerpc/powerpc64/dl-machine.h
index b90f407119..b88d9d7912 100644
--- a/sysdeps/powerpc/powerpc64/dl-machine.h
+++ b/sysdeps/powerpc/powerpc64/dl-machine.h
@@ -620,7 +620,7 @@  extern void attribute_hidden _dl_reloc_overflow (struct link_map *map,
 						 Elf64_Addr *const reloc_addr,
 						 const Elf64_Sym *refsym);
 
-auto inline void __attribute__ ((always_inline))
+static inline void __attribute__ ((always_inline))
 elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
 			   void *const reloc_addr_arg)
 {
@@ -629,7 +629,7 @@  elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
 }
 
 /* This computes the value used by TPREL* relocs.  */
-auto inline Elf64_Addr __attribute__ ((always_inline, const))
+static inline Elf64_Addr __attribute__ ((always_inline, const))
 elf_machine_tprel (struct link_map *map,
 		   struct link_map *sym_map,
 		   const Elf64_Sym *sym,
@@ -648,7 +648,7 @@  elf_machine_tprel (struct link_map *map,
 }
 
 /* Call function at address VALUE (an OPD entry) to resolve ifunc relocs.  */
-auto inline Elf64_Addr __attribute__ ((always_inline))
+static inline Elf64_Addr __attribute__ ((always_inline))
 resolve_ifunc (Elf64_Addr value,
 	       const struct link_map *map, const struct link_map *sym_map)
 {
@@ -678,13 +678,13 @@  resolve_ifunc (Elf64_Addr value,
 
 /* Perform the relocation specified by RELOC and SYM (which is fully
    resolved).  MAP is the object containing the reloc.  */
-auto inline void __attribute__ ((always_inline))
+static inline void __attribute__ ((always_inline))
 elf_machine_rela (struct link_map *map,
 		  const Elf64_Rela *reloc,
 		  const Elf64_Sym *sym,
 		  const struct r_found_version *version,
 		  void *const reloc_addr_arg,
-		  int skip_ifunc)
+		  int skip_ifunc, struct link_map *boot_map)
 {
   Elf64_Addr *const reloc_addr = reloc_addr_arg;
   const int r_type = ELF64_R_TYPE (reloc->r_info);
@@ -707,7 +707,11 @@  elf_machine_rela (struct link_map *map,
 
   /* We need SYM_MAP even in the absence of TLS, for elf_machine_fixup_plt
      and STT_GNU_IFUNC.  */
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+  struct link_map *sym_map = boot_map;
+#else
   struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
   Elf64_Addr value = SYMBOL_ADDRESS (sym_map, sym, true) + reloc->r_addend;
 
   if (sym != NULL
@@ -1037,7 +1041,7 @@  elf_machine_rela (struct link_map *map,
   MODIFIED_CODE_NOQUEUE (reloc_addr);
 }
 
-auto inline void __attribute__ ((always_inline))
+static inline void __attribute__ ((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      Elf64_Addr l_addr, const Elf64_Rela *reloc,
 		      int skip_ifunc)
diff --git a/sysdeps/riscv/dl-machine.h b/sysdeps/riscv/dl-machine.h
index 5b0746175c..8e759ac8d3 100644
--- a/sysdeps/riscv/dl-machine.h
+++ b/sysdeps/riscv/dl-machine.h
@@ -161,11 +161,11 @@  elf_machine_fixup_plt (struct link_map *map, lookup_t t,
    by RELOC_ADDR.  SYM is the relocation symbol specified by R_INFO and
    MAP is the object containing the reloc.  */
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
 		  const ElfW(Sym) *sym, const struct r_found_version *version,
-		  void *const reloc_addr, int skip_ifunc)
+		  void *const reloc_addr, int skip_ifunc, struct link_map *boot_map)
 {
   ElfW(Addr) r_info = reloc->r_info;
   const unsigned long int r_type = ELFW (R_TYPE) (r_info);
@@ -279,7 +279,7 @@  elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
     }
 }
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
 			  void *const reloc_addr)
@@ -287,7 +287,7 @@  elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
   *(ElfW(Addr) *) reloc_addr = l_addr + reloc->r_addend;
 }
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_lazy_rel (struct link_map *map, ElfW(Addr) l_addr,
 		      const ElfW(Rela) *reloc, int skip_ifunc)
@@ -320,7 +320,7 @@  elf_machine_lazy_rel (struct link_map *map, ElfW(Addr) l_addr,
 /* Set up the loaded object described by L so its stub function
    will jump to the on-demand fixup code __dl_runtime_resolve.  */
 
-auto inline int
+static inline int
 __attribute__ ((always_inline))
 elf_machine_runtime_setup (struct link_map *l, int lazy, int profile)
 {
diff --git a/sysdeps/s390/s390-32/dl-machine.h b/sysdeps/s390/s390-32/dl-machine.h
index d0ccd69261..712367fa16 100644
--- a/sysdeps/s390/s390-32/dl-machine.h
+++ b/sysdeps/s390/s390-32/dl-machine.h
@@ -321,11 +321,11 @@  elf_machine_plt_value (struct link_map *map, const Elf32_Rela *reloc,
 /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
    MAP is the object containing the reloc.  */
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
 		  const Elf32_Sym *sym, const struct r_found_version *version,
-		  void *const reloc_addr_arg, int skip_ifunc)
+		  void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
 {
   Elf32_Addr *const reloc_addr = reloc_addr_arg;
   const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
@@ -484,7 +484,7 @@  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
     }
 }
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
 			   void *const reloc_addr_arg)
@@ -493,7 +493,7 @@  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
   *reloc_addr = l_addr + reloc->r_addend;
 }
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      Elf32_Addr l_addr, const Elf32_Rela *reloc,
diff --git a/sysdeps/s390/s390-64/dl-machine.h b/sysdeps/s390/s390-64/dl-machine.h
index 543361c836..77a658e17f 100644
--- a/sysdeps/s390/s390-64/dl-machine.h
+++ b/sysdeps/s390/s390-64/dl-machine.h
@@ -268,11 +268,11 @@  elf_machine_plt_value (struct link_map *map, const Elf64_Rela *reloc,
 /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
    MAP is the object containing the reloc.  */
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela (struct link_map *map, const Elf64_Rela *reloc,
 		  const Elf64_Sym *sym, const struct r_found_version *version,
-		  void *const reloc_addr_arg, int skip_ifunc)
+		  void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
 {
   Elf64_Addr *const reloc_addr = reloc_addr_arg;
   const unsigned int r_type = ELF64_R_TYPE (reloc->r_info);
@@ -438,7 +438,7 @@  elf_machine_rela (struct link_map *map, const Elf64_Rela *reloc,
     }
 }
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
 			   void *const reloc_addr_arg)
@@ -447,7 +447,7 @@  elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
   *reloc_addr = l_addr + reloc->r_addend;
 }
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      Elf64_Addr l_addr, const Elf64_Rela *reloc,
diff --git a/sysdeps/sh/dl-machine.h b/sysdeps/sh/dl-machine.h
index 122b417a17..9c3c0ba099 100644
--- a/sysdeps/sh/dl-machine.h
+++ b/sysdeps/sh/dl-machine.h
@@ -259,11 +259,11 @@  elf_machine_plt_value (struct link_map *map, const Elf32_Rela *reloc,
 /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
    MAP is the object containing the reloc.  */
 
-auto inline void
+static inline void
 __attribute ((always_inline))
 elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
 		  const Elf32_Sym *sym, const struct r_found_version *version,
-		  void *const reloc_addr_arg, int skip_ifunc)
+		  void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
 {
   Elf32_Addr *const reloc_addr = reloc_addr_arg;
   const unsigned int r_type = ELF32_R_TYPE (reloc->r_info);
@@ -318,7 +318,11 @@  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
   else
     {
       const Elf32_Sym *const refsym = sym;
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+      struct link_map *sym_map = boot_map;
+#else
       struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
 
       value = SYMBOL_ADDRESS (sym_map, sym, true);
       value += reloc->r_addend;
@@ -424,7 +428,7 @@  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
     }
 }
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
 			   void *const reloc_addr_arg)
@@ -443,7 +447,7 @@  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
 #undef COPY_UNALIGNED_WORD
 }
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      Elf32_Addr l_addr, const Elf32_Rela *reloc,
diff --git a/sysdeps/sparc/sparc32/dl-machine.h b/sysdeps/sparc/sparc32/dl-machine.h
index 0269e458ea..20f60b2a3b 100644
--- a/sysdeps/sparc/sparc32/dl-machine.h
+++ b/sysdeps/sparc/sparc32/dl-machine.h
@@ -327,11 +327,11 @@  elf_machine_plt_value (struct link_map *map, const Elf32_Rela *reloc,
 /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
    MAP is the object containing the reloc.  */
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
 		  const Elf32_Sym *sym, const struct r_found_version *version,
-		  void *const reloc_addr_arg, int skip_ifunc)
+		  void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
 {
   Elf32_Addr *const reloc_addr = reloc_addr_arg;
 #if !defined RTLD_BOOTSTRAP && !defined RESOLVE_CONFLICT_FIND_MAP
@@ -381,7 +381,11 @@  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
     }
   else
     {
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+      sym_map = boot_map;
+#else
       sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
       value = SYMBOL_ADDRESS (sym_map, sym, true);
     }
 #else
@@ -536,7 +540,7 @@  elf_machine_rela (struct link_map *map, const Elf32_Rela *reloc,
     }
 }
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
 			   void *const reloc_addr_arg)
@@ -545,7 +549,7 @@  elf_machine_rela_relative (Elf32_Addr l_addr, const Elf32_Rela *reloc,
   *reloc_addr += l_addr + reloc->r_addend;
 }
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      Elf32_Addr l_addr, const Elf32_Rela *reloc,
diff --git a/sysdeps/sparc/sparc64/dl-machine.h b/sysdeps/sparc/sparc64/dl-machine.h
index bbd4566d8a..4d47b3a150 100644
--- a/sysdeps/sparc/sparc64/dl-machine.h
+++ b/sysdeps/sparc/sparc64/dl-machine.h
@@ -354,11 +354,11 @@  elf_machine_runtime_setup (struct link_map *l, int lazy, int profile)
 /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
    MAP is the object containing the reloc.  */
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela (struct link_map *map, const Elf64_Rela *reloc,
 		  const Elf64_Sym *sym, const struct r_found_version *version,
-		  void *const reloc_addr_arg, int skip_ifunc)
+		  void *const reloc_addr_arg, int skip_ifunc, struct link_map *boot_map)
 {
   Elf64_Addr *const reloc_addr = reloc_addr_arg;
 #if !defined RTLD_BOOTSTRAP && !defined RESOLVE_CONFLICT_FIND_MAP
@@ -408,7 +408,11 @@  elf_machine_rela (struct link_map *map, const Elf64_Rela *reloc,
     }
   else
     {
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+      sym_map = boot_map;
+#else
       sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
       value = SYMBOL_ADDRESS (sym_map, sym, true);
     }
 #else
@@ -646,7 +650,7 @@  elf_machine_rela (struct link_map *map, const Elf64_Rela *reloc,
     }
 }
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
 			   void *const reloc_addr_arg)
@@ -655,7 +659,7 @@  elf_machine_rela_relative (Elf64_Addr l_addr, const Elf64_Rela *reloc,
   *reloc_addr = l_addr + reloc->r_addend;
 }
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      Elf64_Addr l_addr, const Elf64_Rela *reloc,
diff --git a/sysdeps/x86_64/dl-machine.h b/sysdeps/x86_64/dl-machine.h
index ceee50734e..738663ec5d 100644
--- a/sysdeps/x86_64/dl-machine.h
+++ b/sysdeps/x86_64/dl-machine.h
@@ -251,11 +251,12 @@  elf_machine_plt_value (struct link_map *map, const ElfW(Rela) *reloc,
 /* Perform the relocation specified by RELOC and SYM (which is fully resolved).
    MAP is the object containing the reloc.  */
 
-auto inline void
+static inline void
 __attribute__ ((always_inline))
 elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
 		  const ElfW(Sym) *sym, const struct r_found_version *version,
-		  void *const reloc_addr_arg, int skip_ifunc)
+		  void *const reloc_addr_arg, int skip_ifunc,
+		  struct link_map *boot_map)
 {
   ElfW(Addr) *const reloc_addr = reloc_addr_arg;
   const unsigned long int r_type = ELFW(R_TYPE) (reloc->r_info);
@@ -293,7 +294,11 @@  elf_machine_rela (struct link_map *map, const ElfW(Rela) *reloc,
 # ifndef RTLD_BOOTSTRAP
       const ElfW(Sym) *const refsym = sym;
 # endif
+#if defined RTLD_BOOTSTRAP || defined STATIC_PIE_BOOTSTRAP
+      struct link_map *sym_map = boot_map;
+#else
       struct link_map *sym_map = RESOLVE_MAP (&sym, version, r_type);
+#endif
       ElfW(Addr) value = SYMBOL_ADDRESS (sym_map, sym, true);
 
       if (sym != NULL
@@ -518,7 +523,7 @@  and creates an unsatisfiable circular dependency.\n",
     }
 }
 
-auto inline void
+static inline void
 __attribute ((always_inline))
 elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
 			   void *const reloc_addr_arg)
@@ -537,7 +542,7 @@  elf_machine_rela_relative (ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
     }
 }
 
-auto inline void
+static inline void
 __attribute ((always_inline))
 elf_machine_lazy_rel (struct link_map *map,
 		      ElfW(Addr) l_addr, const ElfW(Rela) *reloc,
@@ -573,7 +578,8 @@  elf_machine_lazy_rel (struct link_map *map,
 
       /* Always initialize TLS descriptors completely at load time, in
 	 case static TLS is allocated for it that requires locking.  */
-      elf_machine_rela (map, reloc, sym, version, reloc_addr, skip_ifunc);
+      elf_machine_rela (map, reloc, sym, version, reloc_addr, skip_ifunc,
+                        NULL);
     }
   else if (__glibc_unlikely (r_type == R_X86_64_IRELATIVE))
     {