[Bug dynamic-link/15030] New: do_lookup_x returns undefined symbol entry

[Bug dynamic-link/15030] New: do_lookup_x returns undefined symbol entry

[hjl.tools at gmail dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=15030

             Bug #: 15030
           Summary: do_lookup_x returns undefined symbol entry
           Product: glibc
           Version: 2.18
            Status: NEW
          Severity: normal
          Priority: P2
         Component: dynamic-link
        AssignedTo: unassigned@sourceware.org
        ReportedBy: hjl.tools@gmail.com
    Classification: Unclassified


do_lookup_x has

          /* Use the old SysV-style hash table.  Search the appropriate
             hash bucket in this object's symbol table for a definition
             for the same symbol name.  */
          for (symidx = map->l_buckets[*old_hash % map->l_nbuckets];
               symidx != STN_UNDEF;
               symidx = map->l_chain[symidx])
            {
              sym = check_match (&symtab[symidx]);
              if (sym != NULL)
                goto found_it;
            }


When the old SysV-style hash table, we may return undefined
symbol entry.  We don't run into this problem often since
GNU-style hash table is enabled by most GCC.

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug dynamic-link/15030] do_lookup_x returns undefined symbol entry

[hjl.tools at gmail dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=15030

--- Comment #1 from H.J. Lu <hjl.tools at gmail dot com> 2013-01-18 02:13:50 UTC ---
      /* Nested routine to check whether the symbol matches.  */
      const ElfW(Sym) *
      __attribute_noinline__
      check_match (const ElfW(Sym) *sym)
      {    
        unsigned int stt = ELFW(ST_TYPE) (sym->st_info);
        assert (ELF_RTYPE_CLASS_PLT == 1);
        if (__builtin_expect ((sym->st_value == 0 /* No value.  */
                               && stt != STT_TLS)
                              || (type_class & (sym->st_shndx == SHN_UNDEF)),
                              0))
          return NULL;

Should we remove "type_class &"?

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug dynamic-link/15030] do_lookup_x returns undefined symbol entry

[hjl.tools at gmail dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=15030

--- Comment #2 from H.J. Lu <hjl.tools at gmail dot com> 2013-01-18 15:26:05 UTC ---
The reason it works for GNU hash table is GNU hash table
only contains defined symbols.

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug dynamic-link/15030] do_lookup_x returns undefined symbol entry

[hjl.tools at gmail dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=15030

--- Comment #3 from H.J. Lu <hjl.tools at gmail dot com> 2013-01-18 17:58:21 UTC ---
It only happens with size relocation against TLS symbol:

[hjl@gnu-6 sized-tls-2]$ readelf -r dynamic 

Relocation section '.rela.text' at offset 0x4d0 contains 1 entries:
  Offset          Info           Type           Sym. Value    Sym. Name +
Addend
000000400604  000300000020 R_X86_64_SIZE32   0000000000000000 bar + 0

Relocation section '.rela.got' at offset 0x4e8 contains 2 entries:
  Offset          Info           Type           Sym. Value    Sym. Name +
Addend
000000600b18  000300000012 R_X86_64_TPOFF64  0000000000000000 bar + 0
000000600b20  000500000006 R_X86_64_GLOB_DAT 0000000000000000 __gmon_start__ +
0

We don't consider R_X86_64_SIZE32 as ELF_RTYPE_CLASS_PLT class.
We can check STT_TLS instead of TLS relocations for relocations
against TLS symbols.

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug dynamic-link/15030] do_lookup_x returns undefined symbol entry

[hjl.tools at gmail dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=15030

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |WONTFIX

--- Comment #4 from H.J. Lu <hjl.tools at gmail dot com> 2013-01-18 23:01:12 UTC ---
I checked in a patch to resolve size relocation against non-empty
TLS symbol at link-time:

http://sourceware.org/ml/binutils/2013-01/msg00309.html

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug dynamic-link/15030] do_lookup_x returns undefined symbol entry

[hjl.tools at gmail dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=15030

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|WONTFIX                     |

--- Comment #5 from H.J. Lu <hjl.tools at gmail dot com> 2013-02-07 02:20:18 UTC ---
(In reply to comment #4)
> I checked in a patch to resolve size relocation against non-empty
> TLS symbol at link-time:
> 
> http://sourceware.org/ml/binutils/2013-01/msg00309.html

I reverted this patch:

http://sourceware.org/ml/binutils/2013-02/msg00073.html

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug dynamic-link/15030] do_lookup_x returns undefined symbol entry

[hjl.tools at gmail dot com]

http://sourceware.org/bugzilla/show_bug.cgi?id=15030

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |NEW

--- Comment #6 from H.J. Lu <hjl.tools at gmail dot com> 2013-02-07 18:57:06 UTC ---
[hjl@gnu-6 pr15030]$ cat m.c
#include <stdio.h>

extern __thread char bar[];
extern char size_of_bar asm ("bar@SIZE");
extern void set_bar (int, int);

int
main ()
{
  set_bar (1, 20);
  if (10 == (long) &size_of_bar && bar[1] == 20)
    printf ("OK\n");
  else
    printf ("BAD\n");

  return 0;
}
[hjl@gnu-6 pr15030]$ cat b.c
__thread char bar[10];

void
set_bar (int i, int v)
{
  bar[i] = v;
}
[hjl@gnu-6 pr15030]$ make
gcc -B./ -g -O2   -c -o m.o m.c
gcc -B./ -g -O2 -fPIC   -c -o b.o b.c
gcc -B./ -shared -o b.so b.o
gcc -B./ -o x m.o b.so -Wl,-R,. -Wl,--hash-style=gnu
gcc -B./ -o x1 m.o b.so -Wl,-R,. -Wl,--hash-style=sysv
./x1
BAD
./x
OK
[hjl@gnu-6 pr15030]$

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[Bug dynamic-link/15030] do_lookup_x returns undefined symbol entry

[fweimer at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=15030

Florian Weimer <fweimer at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
              Flags|                            |security-

-- 
You are receiving this mail because:
You are on the CC list for the bug.