Commit Graph

251 Commits

Author SHA1 Message Date
Alexander Kabaev
96ff9a2bf4 Make lookups for relocations from old unversioned binaries return
oldest versioned symbol available. Do not accept hidden symbols for
all other versions.

Use "<obj->path>: <error message>" for all error messages in new
functions to make them more consistent.
2005-12-24 15:37:30 +00:00
Alexander Kabaev
b80d39d0c0 Remove debugging statement that slipped into lone of the previous commits
unintentionally.
2005-12-23 15:30:53 +00:00
Alexander Kabaev
f6e5db226f Initialize object dagmembers list before checking version dependencies. 2005-12-22 16:42:38 +00:00
Alexander Kabaev
0eb88f2029 Implement ELF symbol versioning using GNU semantics. This code aims
to be compatible with symbol versioning support as implemented by
GNU libc and documented by http://people.redhat.com/~drepper/symbol-versioning
and LSB 3.0.

Implement dlvsym() function to allow lookups for a specific version of
a given symbol.
2005-12-18 19:43:33 +00:00
Marcel Moolenaar
757686b115 Make our ELF64 type definitions match standards. In particular this
means:
o  Remove Elf64_Quarter,
o  Redefine Elf64_Half to be 16-bit,
o  Redefine Elf64_Word to be 32-bit,
o  Add Elf64_Xword and Elf64_Sxword for 64-bit entities,
o  Use Elf_Size in MI code to abstract the difference between
   Elf32_Word and Elf64_Word.
o  Add Elf_Ssize as the signed counterpart of Elf_Size.

MFC after: 2 weeks
2005-12-18 04:52:37 +00:00
Marcel Moolenaar
55dfaa9163 Explicitly cast ELF_R_TYPE() to the right type. 2005-12-18 01:38:26 +00:00
John Baldwin
4d5fe96d68 Fix a bug in dlinfo(RTLD_DI_SERINFOSIZE) requests. For each search path
we included the length of the path in the returned size but not the length
of the associated Dl_serpath structure.  Without this fix, programs
attempting to allocate a structure to hold the search path information
would allocate too small of a buffer and rtld would overrun the buffer
while filling it via a subsequent RTLD_DI_SERINFO request.

Submitted by:	"William K. Josephson" wkj at morphisms dot net
Reviewed by:	jdp
MFC after:	2 weeks
2005-11-11 19:57:41 +00:00
Peter Wemm
3b4399f6a7 Clean out the leftovers from the i386_set_gsbase() TLS conversion.
Like on libthr, there is an i386_set_gsbase() stub implementation here
to avoid libc.so.5 issues.  This should likely be a weak symbol and I
expect this will be fixed soon.

Approved by:	re
2005-06-29 23:15:36 +00:00
Ruslan Ermilov
cc4fdb1de9 Removed trailing whitespace.
Approved by:	re (blanket)
2005-06-14 08:41:11 +00:00
Marius Strobl
8698905073 MFNetBSD: src/libexec/ld.elf_so/arch/sparc64/mdreloc.c 1.26 (partial)
Fix an obvious bug in the 64-bit PLT fixup: the SLLX was by 12 bits, when
it should be 32.

MFC after:	1 month
2005-06-04 20:49:15 +00:00
Marcel Moolenaar
d4337d869f Fix the load64 and store64 macros, created to handle 8-byte unaligned
loads and stores (resp.) The ldq_u and stq_u instruction mask off the
lower 3 bits of the final address before loading from or storing to
the address, so as to avoid unaligned loads and stores. They do not
themselves allow loads from or stores to unaligned addresses. Replace
the macro definitions by a packed struct dereference.

Submitted by: Richard Henderson (rth at twiddle dot net)
2005-06-02 05:34:08 +00:00
Doug Rabson
9310a53def Align the stack to a 16 byte boundary before calling _rtld so that we can
safely initialise shared libraries that use SSE in their init sections.

MFC After: 1 week
2005-05-19 07:32:42 +00:00
David Xu
9b0c632a4c Fix compilation problem. 2005-04-27 13:17:23 +00:00
Peter Wemm
8d598c0d01 Stop calling _amd64_set_gsbase() for COMPAT_32BIT. The amd64 kernel
implements i386_set_gsbase(), so there is no need for the variation.
2005-04-26 20:38:44 +00:00
Peter Wemm
8a477e0a7a Attempt to use i386_set_gsbase(), and gracefully fall back to LDT methods
if the direct access methods are not implemented.
2005-04-14 00:04:50 +00:00
Olivier Houchard
2204e78f5e No need to provide atomic_cmpset_32() anymore. 2005-04-07 22:04:49 +00:00
Doug Rabson
3709906a1d When allocating TLS and DTV, make sure that any unused slots in the DTV
are initialised to zero. When freeing TLS, don't attempt to free DTV
slots which were not used.

Pointed out by: Joerg Sonnenberger
X-MFC-After: After the branch, probably
2005-03-30 08:28:26 +00:00
Colin Percival
8fe7df9b68 If "dangerous" environment variables (LD_PRELOAD, LD_LIBMAP,
LD_LIBMAP_DISABLE, LD_LIBRARY_PATH) are used, then make sure the
libraries being loaded aren't on a noexec-mounted filesystem.

This is a compromise position: I'm assuming that nobody will be silly
enough to set the noexec mount flag on part of the default library
path, in order to avoid adding extra overhead into the common case
(where those environment variables aren't used).

Discussed with:	csjp, secteam
MFC after:	1 week
2005-03-24 10:12:29 +00:00
David Xu
c5fa3778bb Add locking code for tls routines. 2005-03-20 23:28:25 +00:00
Doug Rabson
ddab7ee80a Attempt to free any static TLS space used by a shared library when it
is unloaded. This allows applications which load and unload libraries
like libGL.so.1 several times to work properly.

MFC after: 2 days
2005-02-27 12:55:40 +00:00
Olivier Houchard
c6ac5bfcae Only provide the dummy, non-atomic atomic_cmpset_32() if
ARM_HAS_ATOMIC_CMPSET_32 isn't defined.
2005-02-26 22:49:19 +00:00
Matthew N. Dodd
5b08cb0449 Description from Dan:
Another handy libmap patch.  Lets you do stuff like this:

	LD_LIBMAP="libpthread.so.1=libthr.so.1" mythreadedapp

	If you already have a program-specific override in libmap.conf, note
	that you must use a program-specific override in LD_LIBMAP:

	LD_LIBMAP="[mythreadedapp],libpthread.so.1=libthr.so.1" mythreadedapp

PR:		bin/74471
Submitted by:	Dan Nelson <dnelson AT allantgroup.com>
MFC after:	2 weeks
2005-02-04 02:46:41 +00:00
Matthew N. Dodd
d33da23f4c style(9) 2005-02-04 02:13:37 +00:00
Olivier Houchard
25a252899e Implement a dummy atomic_cmpset_32(). It should be safe to use it in rtld as
the signals are masked anyway.
2004-11-23 16:32:34 +00:00
John Baldwin
165204a75f Remove 80386 support from the ELF run time linker. 2004-11-16 20:45:51 +00:00
Ruslan Ermilov
1b1aa7e465 So do it like we do in usr.bin/tip/tip/Makefile. ;) 2004-11-14 22:18:31 +00:00
Jens Schweikhardt
443ceb1c7e Revert previous commit. As ru explains:
In the old world (as the surrounding comment in makefile says), there
 was the /usr/libexec/ld-elf.so.1 binary which is now a symlink to
 /libexec/ld-elf.so.1. To symlink, we need to make sure that the
 _target_ (and the target is /usr/libexec/ld-elf.so.1) doesn't have
 "schg" flag set. A real solution is to protect the chflags call only if
 target exists, like we do in usr.bin/tip/tip/Makefile.

Requested by:	ru
2004-11-14 21:14:06 +00:00
Jens Schweikhardt
6ccc491b4b Avoid an (ignored) error by invoking chflags on the link target, not the
symlink.

PR:		kern/73016
Submitted by:	John E. Hein <jhein@timing.com>
MFC after:	1 week
2004-11-14 12:47:20 +00:00
John Baldwin
2939195e46 Remove these unused files before any other archs include the same bogus
file.
2004-11-12 18:05:30 +00:00
Peter Wemm
24b4ec3d21 The 32 bit compatability ld-elf32.so.1 cannot use i386_set_ldt() when
running on an amd64 kernel.  Use the recently exposed direct %fs/%gs set
routines instead for the TLS setup of 32 bit binaries.
2004-11-06 03:32:07 +00:00
Ruslan Ermilov
dee651eb15 Introduce the PRECIOUSPROG knob in bsd.prog.mk, similar
to PRECIOUSLIB from bsd.lib.mk.  The side effect of this
is making installing the world under jail(8) possible by
using another knob, NOFSCHG.

Reviewed by:	oliver
2004-11-03 18:01:21 +00:00
Suleiman Souhlal
6c2a9753f2 Implement TLS relocations for powerpc.
Approved by:	grehan (mentor)
2004-11-02 09:47:01 +00:00
Suleiman Souhlal
5bbd22ee8d Do the TLS offset allocations before relocations, as otherwise there
can be overlap in the TLS offsets, if the relocations are done in a
certain order.

Approved by:	dfr, grehan (mentor)
2004-11-02 09:42:21 +00:00
Ruslan Ermilov
a35d88931c For variables that are only checked with defined(), don't provide
any fake value.
2004-10-24 15:33:08 +00:00
Olivier Houchard
34be051190 Don't try to relocate the dynamic loader in reloc_non_plt(). It has already
been done before.
2004-09-28 14:43:12 +00:00
Olivier Houchard
0e030636cb Use add instead of saving the sp in a register. 2004-09-28 14:41:15 +00:00
Olivier Houchard
9ac88d19dc Add stubs for TLS.
Arbitraly choose the 2nd variant until I figure out which one I should use.
2004-09-23 23:04:52 +00:00
Doug Rabson
fca32c746e Add stubs for powerpc TLS.
Submitted by: ssouhlal
2004-08-04 19:12:14 +00:00
Doug Rabson
017246d02f Add support for Thread Local Storage. 2004-08-03 08:51:00 +00:00
Stefan Farfeleder
5908d366fb Consistently use __inline instead of __inline__ as the former is an empty macro
in <sys/cdefs.h> for compilers without support for inline.
2004-07-04 16:11:03 +00:00
Thomas Moestl
d05bb9a2a6 Fix the problem that surfaced with the new binutils import on sparc64
(and that is for now being worked around by a binutils patch).

The rtld code tested &_DYNAMIC against 0 to see whether rtld itself
was built as PIC or not. While the sparc64 MD code did not rely
on the preset value of the GOT slot for _DYNAMIC any more due
to previous binutils changes, it still used to not be 0, so
that this check did work. The new binutils do however initialize
this slot with 0. As a consequence, rtld would not properly initialize
itself and crash.
Fix that by introducing a new macro, RTLD_IS_DYNAMIC, to take the role
of this test. For sparc64, it is implemented using the rtld_dynamic()
code that was already there. If an architecture does not provide its
own implementation, we default to the old check.

While being there, mark _DYNAMIC as a weak symbol in the sparc64
rtld_start.S. This is needed in the LDSCRIPT case, which is however
not currently supported for want of an actual ldscript.

Sanity checked with md5 on alpha, amd64, i386 and ia64.
2004-06-18 02:01:37 +00:00
Olivier Houchard
45ab3f5350 This comment should have been removed in the previous commit.
Spotted out by: marcus, simon
2004-06-17 19:01:53 +00:00
Olivier Houchard
f77d42ce5c Woohoo !
the latest binutils import mades this gross hack useless, so just remove it.
2004-06-17 17:53:16 +00:00
Oliver Eikemeier
0a16eb8341 give out a little more information in case of a missing dependency
PR:		56549
Submitted by:	edwin
Reviewed by:	joerg, ru
Approved by:	joerg
MFC after:	2 weeks
2004-05-28 00:05:28 +00:00
Matthew N. Dodd
966efcc767 Support basename and path based constrained matches.
eg:
	[foo]
	...

	matches any executable 'foo'

	[/usr/bin/foo/]
	...

	matches any executable under the directory /usr/bin/foo/

Exact matches continue to function as before.

PR:		 bin/66769
Submitted-by:	 Dan Nelson
2004-05-24 01:24:13 +00:00
Olivier Houchard
3088daddf3 Work around a problem somewhere with binutils (?) on arm, hopefully without
breaking any other arch this time.
2004-05-15 00:13:14 +00:00
Stefan Eßer
55a4ccf3e8 Fix breakage caused by alphabetically sorting SRCS: rtld_start.S must come first!
The previous version made all shared binaries dump core.
2004-05-14 21:01:52 +00:00
Olivier Houchard
e659267f1e Import arm bits for rtld-elf.
Obtained from:	NetBSD
2004-05-14 12:15:51 +00:00
Doug Rabson
f88e6caca2 If we change obj_rtld.path after initialising __progname, make sure we
change __progname to point at the new storage otherwise it ends up
pointing at freed memory which leads to confusing garbled error messages.
2004-03-29 18:37:37 +00:00
Peter Wemm
c707fea10b More stack alignment fixes. Arrange so we call _rtld() in ld-elf.so.1
with the correct alignment.  This is important because this calls to
library static constructors are made from here.  The bug in the old crt*.s
files hid this because in this case, two wrongs do indeed make a right.
Also, call _rtld_bind() with the correct alignment, because it calls back
into the pthread library locking functions.  If things happen just
the wrong way, we get a SIG10 due to the broken stack alignment.
2004-03-21 01:43:39 +00:00