1
0
Fork 0
mirror of https://github.com/tobast/libunwind-eh_elf.git synced 2024-11-28 10:07:37 +01:00
Commit graph

1789 commits

Author SHA1 Message Date
Ulrich Weigand
c56fb8f99e ppc64: Fix serious regression (many crashes in test suite)
A recent commit added code to override the unwind location for the
TOC pointer register r2:

    unsigned int *inst = (unw_word_t*)c->dwarf.ip;
    if (*inst == (0xE8410000 + 24)) {
      // @plt call, restoring R2 from CFA+24
      c->dwarf.loc[UNW_PPC64_R2] = DWARF_LOC(c->dwarf.cfa + 24, 0);
    }

It is correct that such code is needed, since DWARF CFI does not
describe the unwind location for r2 on PowerPC.  However, this
particular bit of code has a number of issues, which are fixed
in this patch.

First of all, the location CFA+24 is correct only for the ELFv2
ABI.  In the ELFv1 ABI, the TOC location is actually CFA+40.

More problematically, attempting to read the current instruction
by just dereferencing the address in c->dwarf.ip is wrong, and
may often lead to crashes.  In particular:

- During remote unwinding, this is always wrong since we're in
  the wrong address space.  I've used the fetch32 helper from
  remote.h to use the proper access_mem under the covers.

- c->dwarf.ip may be NULL if we've reached the end-of-stack.
  I've fixed this by moving the c->dwarf.ip == 0 check down
  to after unwinding (instead of before), just like all other
  platforms do.

- Even so, c->dwarf.ip may point to some random location if
  we've gotten confused during unwinding earlier.  One likely
  cause for such confusion is that we did not find DWARF CFI
  for some earlier frame and attempted to use the stack
  backchain.  The problem is that this code currently claims
  all registers remain unchanges in such a frame, which is
  generally wrong.  In particular if the function actually
  saves and modifies r31, and this is used as frame pointer
  by a later frame, things will likely go quite wrong.  While
  it is not really possibly to completely fix this, I've at
  least marked all registers as unavailable after unwinding
  a frame via stack backchain.

Tested on powerpc64-linux and powerpc64le-linux.  The patch fixes
about a dozen test cases that were crashing before.

Signed-off-by: Ulrich Weigand <ulrich.weigand@de.ibm.com>
2015-12-06 11:10:10 -08:00
Keno Fischer
8afc33ce9f Add an option to have start_ip_offset be relative to start_ip
By default, the start_ip_offset in libunwind's table_entry struct is
relative to the unw_dyn_info_t's segbase. This presents a problem
for us in conjunction with using LLVM's MCJIT because it likes to
spread text sections and the corresponding eh_frame sections quite
far apart. This represents my attempt to support this use case in the
simplest manner that is backwards compatible, by adding a new format
kind (UNW_INFO_FORMAT_REMOTE_TABLE2) that indicates that the
`start_ip_offset` should be interpreted as relative to `start_ip`
rather than segbase.
2015-09-15 12:18:30 -07:00
Alexander Esilevich
d1a8ca6d84 powerpc64 implementation 2015-09-07 18:08:26 -07:00
Arun Sharma
923fa78e79 Revert parts of 68a2910
The check breaks a few tests when used with gold.
2015-09-07 17:59:27 -07:00
Vyacheslav Barinov
7701522688 arm: Validate memory before access
Prevent SIGSEGV due to accessing addresses now mapped to current process

Signed-off-by: Vyacheslav Barinov <v.barinov@samsung.com>
2015-06-07 12:17:40 -07:00
Faraz Shahbazker
16bf4e5e49 Stack-unwinding through signal frames on mips
Detect if current frame is a signal frame by checking for preceeding
syscall instruction. For signal frames, update the cursor correctly
from user context before calling dwarf_step.
2015-04-02 11:20:33 -07:00
Peter Wu
68a2910bae Check that the CIE is within the segment
Due to a bug in the gold linker[1], the .eh_frame and .eh_frame_hdr
sections contains garbage. When dwarf_extract_proc_info_from_fde tried
to look up the begin of the CIE subsection, it would underflow the
.eh_frame segment, resulting in a crash[2].

This patch avoids that crash by checking whether the CIE pointer is
located after the begin of the .eh_frame section. The variable "base"
was misused in various places as a boolean (decode as .debug_frame or
decode as .eh_frame). These instances have been renamed to
is_debug_frame where applicable.

Tested on Linux x86_64.

 [1]: https://sourceware.org/bugzilla/show_bug.cgi?id=17639
 [2]: http://lists.nongnu.org/archive/html/libunwind-devel/2014-11/msg00009.html

Signed-off-by: Peter Wu <peter@lekensteyn.nl>
2014-12-05 23:39:48 -08:00
Victor Kamensky
16e98c8cf3 aarch64: fix wrong big_endian flag in aarch64_be case
Set local_addr_space.big_endian flag according to current target
endianness. Before it was set by memset to 0, which corresponds
to little endian and it worked perfectly fine for aarch64.
But it breaks aarch64_be because dwarf_readu16, dwarf_readu32,
etc functions do not read values correctly since they operate
with wrong idea about current target endianness.

Signed-off-by: Victor Kamensky <victor.kamensky@linaro.org>
2014-12-03 23:28:07 -08:00
C. Bergström
3babc1d148 aarch64: Restore x0, x1, x2 and x3
These registers contain the exception pointer and selector.
2014-10-24 22:40:19 -07:00
Paul Pluzhnikov
1bca6eb8ea Remove unnecessary calls to labs()
We've just traced a large memory increase to that patch (Google ref:
b/18069427).
It appears that labs() was there for a good reason.

Sorry about that :-(

For the curious:

  unsigned long u1 = ~0UL;
  unsigned long u2 = labs(u1);

  assert(u1 != u2);  // labs on unsigned *may* have an effect, despite
what Clang says.

Attached patch suppresses the Clang warning, while still keeping the
original behavior (which I believe to be correct).

Thanks,
--
Paul Pluzhnikov
2014-10-24 22:29:15 -07:00
Arun Sharma
781d5d5263 One time whitespace fixup.
for f in $(find src include -name '*.[ch]'); do
  expand -t 8 $f > $tmp; mv $tmp $f;
done
2014-09-27 09:47:23 -07:00
Arun Sharma
06e608d043 Merge branch 'fast-stack-trace-arm' of https://github.com/fillexen/libunwind 2014-09-20 08:09:55 -07:00
Arun Sharma
6046c87d8e Merge branch 'fast-stack-trace-aarch64' of https://github.com/fillexen/libunwind 2014-09-20 08:09:35 -07:00
Arun Sharma
5d79a12ec0 Merge https://github.com/zliu2014/libunwind-tilegx 2014-09-20 00:47:23 -07:00
Zhi-Gang Liu
790be1e40d Add TileGx platform support to libunwind.
"make check" passed.
======================================================
All 34 tests behaved as expected (2 expected failures)
======================================================
Zhi-Gang Liu @ Tilera
2014-09-08 16:21:53 -04:00
Arun Sharma
0808aa6737 Stop unwinding on undefined return address
Milian Wolff <mail@milianw.de> reported that this greatly
improved unwinding perf on some his apps.
2014-09-07 03:32:57 -07:00
Paul Pluzhnikov
e63cadf76d Remove unnecessary calls to labs()
Greetings,

Roman reports that Clang warns on unnecessary calls to labs():

  taking the absolute value of unsigned type 'unw_word_t' (aka
'unsigned long') has no effect [-Werror,-Wabsolute-value]

Since rs->reg[...].val is unw_word_t and unsigned on all platforms,
this patch removes the unnecessary calls to labs().

Tested on Linux x86_64, no regressions.

Thanks,
--
Paul Pluzhnikov
2014-08-15 13:01:47 -07:00
Konstantin Belousov
dbce594d33 Correct name returned by get_proc_name() for some frames.
Fix returning the name of the function containing the frame PC,
for the non-interrupted frames.  The symbol lookup code should
take use_prev_instr value into account, otherwise it could return
the name of the function adjacent to the caller.
2014-08-15 13:01:40 -07:00
Konstantin Belousov
3723511003 Adjust use_prev_instr for syscall and frame-chain frames.
Mark frames which are unwound with the frame-chain walker or
syscall frame code, as non-interrupted.  The return PC in the frame
points to the instruction after the call.
2014-08-15 13:01:36 -07:00
Konstantin Belousov
85946d9766 Use helper to walk past the syscall frames on FreeBSD/x86, similar to FreeBSD/x86_64.
The syscall trampolines in FreeBSD libc do not have unwind annotations.
2014-08-15 13:01:33 -07:00
Filip Nyback
93d8272481 Add fast stack trace on AArch64.
Port of the fast stack trace feature to AArch64.
2014-07-30 14:34:31 +03:00
Filip Nyback
e1804829ca Add fast stack trace on ARM.
Port of the fast stack trace feature to ARM.
2014-07-30 14:25:14 +03:00
Martin Milata
65ac867416 Do not compute load offset for MiniDebugInfo
We already have the value computed based on the executable. Furthermore,
debuginfo need not have valid program header table, thus the result
might be bogus.

Signed-off-by: Martin Milata <mmilata@redhat.com>
2014-06-07 18:07:06 -07:00
Kevin Modzelewski
4509adb853 x86_64: Handle edge conditions with zero length frames
When JITs generate code without unwind information, it may be possible
to continue unwinding via RBP chaining. However, we currently disallow
RBP==RSP condition even though we can make forward progress.

Relax the check a bit in the code where we switch from one type of
unwinding to another to handle this situation. JIT authors
are encouraged to use the dynamic unwind info registration API when
the underlying platform supports it.

Signed-off-by: Kevin Modzelewski <kmod@dropbox.com>
2014-05-18 19:39:26 -07:00
Christopher Ferris
2cb00c1eac AARCH64: Change Debug calls to use %lx instead of %x.
The attached patch simply modifies the Debug statements for aarch64 to use
%lx instead of %x. Tested by compiling for aarch64 using a cross-compiler.

Signed-off-by: Christopher Ferris <cferris@google.com>
2014-03-07 22:52:24 -08:00
Mike Frysinger
385c19f335 link sublibs against liblzma as needed
The coredump/elf32/elf64/elfxx libs use lzma funcs but don't link against
it.  This produces sub-shared libs that don't link against lzma and can
make the linker angry due to underlinking like so:

libtool: link: x86_64-pc-linux-gnu-gcc -O2 -march=amdfam10 -pipe -g \
	-frecord-gcc-switches -Wimplicit-function-declaration -fexceptions \
	-Wall -Wsign-compare -Wl,-O1 -Wl,--hash-style=gnu \
	-o .libs/test-coredump-unwind test-coredump-unwind.o  \
	../src/.libs/libunwind-coredump.so ../src/.libs/libunwind-x86_64.so
../src/.libs/libunwind-coredump.so: error: undefined reference to 'lzma_stream_footer_decode'
../src/.libs/libunwind-coredump.so: error: undefined reference to 'lzma_index_buffer_decode'
../src/.libs/libunwind-coredump.so: error: undefined reference to 'lzma_index_size'
../src/.libs/libunwind-coredump.so: error: undefined reference to 'lzma_index_end'
../src/.libs/libunwind-coredump.so: error: undefined reference to 'lzma_index_uncompressed_size'
../src/.libs/libunwind-coredump.so: error: undefined reference to 'lzma_stream_buffer_decode'
collect2: error: ld returned 1 exit status

So add LIBLZMA to the right LIBADD for each of these libraries.

URL: https://bugs.gentoo.org/444050
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
2014-02-22 10:53:39 -08:00
Tim Deegan
4eb880e1b5 Implement DWARF DW_CFA_val_expression for x86_64
Ubuntu's libc-bin (2.15-0ubuntu20.2) on x86_64 uses DW_CFA_val_expression
in describing the pthread spinlock operations __lll_unlock_wake() and
__lll_lock_wait().  libunwind 1.1 doesn't understand that opcode and
so backtraces from those operations are truncated.

This changeset adds basic support for it, by adding a new type to
dwarf_loc_t that describes the register's actual contents rather than
its location.  I've only implemented the new type for x86_64, and
stubbed it out for all other architectures -- it looks like a lot
of that code is duplicated so oughtn't to be that hard, but I don't
have test cases for them.

Tested that DW_CFA_val_expression works on x86_64 (by using
https://code.google.com/p/gperftools/ on a lock-heavy program).
Build-tested on x86, x86_64 and arm.  The unit tests don't pass for me
on any of those archs, but this cset doesn't break anything that was
passing before.

Signed-off-by: Tim Deegan <tjd@phlegethon.org>
2014-02-08 20:19:20 -08:00
Tim Deegan
11a7d98e39 Missing header in dist tarball.
Signed-off-by: Tim Deegan <tjd@phlegethon.org>
2014-02-08 19:47:17 -08:00
Ulrich Weigand
da0b1a146f Support powerpc64le-linux platform
This patch adds support for the powerpc64le-linux platform.  It consists
of two main features:

- Support little-endian byte order
  This is done via a "big_endian" member of struct unw_addr_space,
  which is evaluated by common code via the dwarf_is_big_endian
  macro, and also in endian-aware code in unw_is_signal_frame.

- Support the ELFv2 ABI
  This is done via an "abi" member of struct unw_addr_space.  This
  is currently only needed in tdep_get_func_addr, since the ELFv2
  ABI does not use function descriptors.

Both new members are initialized in unw_create_addr_space and
ppc64_local_addr_space_init, following the mips precedent.

Since ppc32 and ppc64 now no longer share the unw_create_addr_space
implementation, the file is duplicated from the ppc directory into
ppc32/ppc64.

Tested on powerpc64-linux and powerpc64le-linux.  Support on LE
seems to be as good as existing BE support; I have not attempted to
fix the existing shortcomings of PPC support that already cause a
number to tests to fail due to unimplemented features.

Signed-off-by: Ulrich Weigand <uweigand@de.ibm.com>
2013-12-27 07:09:06 -08:00
Ulrich Weigand
844f1b84cc Pass relocated address to tdep_get_func_addr
The ppc64 implementation of tdep_get_func_addr would crash when
attempting to retrieve the address of a function in a shared
library.  The problem was that it needs to dereference the
function descriptor, but common code was passing the *unrelocated*
adddress of the descriptor to the tdep_get_func_addr routine.

Instead, common code would attempt to relocate the *result* of
tdep_get_func_addr, which is also wrong: the ppc64 implementation
reads the function address from the in-memory copy of the
descriptor, which is already relocation and contains the final
address.

This patch fixes the problem by relocating the descriptor address
before passing it to tdep_get_func_addr, instead of relocating
the result of tdep_get_func_addr.  Since ppc64 is the only
non-trivial implementation of tdep_get_func_addr, this cannot
affect any other platform.

Signed-off-by: Ulrich Weigand <uweigand@de.ibm.com>
2013-12-27 07:08:43 -08:00
Ulrich Weigand
5710c98fbf Fix undefined behavior in dwarf_eval_expr
The dwarf_eval_expr routine uses macros push, pop, and pick to
manipulate the DWARF expression stack.  When these macros are
nested, e.g. in the implementation of DW_OP_dup:
          push (pick (0));
the combination can lead to unfortunate results.

In particular, when substituting into:
do {
  if (tos >= MAX_EXPR_STACK_SIZE)
    {
      Debug (1, "Stack overflow\n");
      return -UNW_EINVAL;
    }
  stack[tos++] = (x);
} while (0)
a value of "x" that makes use of "tos" (as instances of the
pick or pop macros do), the resulting expression will both
use and modify tos without an intervening sequence point,
which is undefined behavior according to the C standard.

And in fact with current GCC on PowerPC, this leads to a
miscompilation of the DW_OP_dup implementation.

This patch fixes the problem by assigning "x" to a
temporary variable before modifying tos.

Signed-off-by: Ulrich Weigand <uweigand@de.ibm.com>
2013-12-27 07:07:48 -08:00
Masatake YAMATO
2c3444d0f4 Remove one of duplicated declarations of `get_proc_name'
Declarations of `get_proc_name' of is duplicated.

Signed-off-by: Masatake YAMATO <yamato@redhat.com>
2013-09-25 07:39:16 -07:00
Masatake YAMATO
dee708f4bc Missing -llzma in libunwind.pc
liblzuma used in decoding MiniDebuginfo is not listed in
libunwind.pc.

Changes in version 2 of patch:

* Don't check HAVE_LZMA. It is redundant.
* Make liblzma as private libraries in use.

Both are suggested by Mike Frysinger <vapier@gentoo.org>.

Signed-off-by: Masatake YAMATO <yamato@redhat.com>
2013-09-11 00:11:17 -07:00
Luca Clementi
af23f0a04b Add Cflags keyword to pkg-config metadata files
pkg-config metadata template files are missing Cflags keyword
which is required to properly compile against libunwind using
PKG_* autoconf macros

Acked-by: Mike Frysinger <vapier@gentoo.org>
2013-08-26 05:40:19 -07:00
Matt Fischer
3b8254d3b9 Fix unwind info freeing code in DWARF parser
The DWARF code allocates its unwind_info objects out of a
memory pool.  The code which frees the object therefore calls
the mempool freeing code.  However, there are cases where the
free code will be run with an unwind_info that was allocated
through a different mechanism (e.g. an ARM exidx table entry).
In these cases, the object should not be freed through the
mempool code.

To correct this, a check was added to ensure that the unwind_info
is of the appropriate type before passing the object along to the
mempool to be freed.
2013-05-19 00:10:18 -07:00
Arun Sharma
d41a453bbf Fix a build error
<elf.h> needs to be included before the check for ELF_CLASS.
2013-05-19 00:04:15 -07:00
Arun Sharma
5eba0f816d Fixup long lines. 2013-05-18 23:20:03 -07:00
Matt Fischer
eac65dc9b8 Add basic support for the QNX operating system
This change adds some special cases to allow libunwind to compile
for QNX.

* QNX's copy of <elf.h> and <link.h> reside in sys/ instead.  To deal
  with this, an AC_CHECK_HEADERS() was added to check for the files
  in both locations.
* Similarly, QNX does not have <endian.h>.  In cases where the file is
  not found, logic was added to refer to QNX-specific macros to determine
  endianness.
* The QCC compiler, which is a wrapper around GCC, cannot handle some
  standard GCC options.  Therefore, logic was added to check for QCC,
  and when it is found, to suppress the use of -lgcc, and to express the
  option -nostartfiles as -Wc,-nostartfiles instead, which is correctly
  passed on to the underlying GCC.
* Finally, the support file os-qnx.c was added, patterned after the existing
  os-*.c files.  Only local image lookup is currently supported (see the
  comments for more information), but this is sufficient for QNX, since
  ptrace is not supported there anyway, and that is the only case where the
  function is required to do remote image lookup.

Change-Id: Ie7934f94a7317bdde59335f2acd4c3a97c0384c1
2013-05-13 10:50:17 -05:00
Yvan Roux
ac6c0a6535 [PATCH] AArch64 port. 2013-05-11 09:20:28 -06:00
Matt Fischer
612e1056f6 Fix symlink install hook
During the install, symlinks are added from libunwind-<arch> to
libunwind-generic.  However, on platforms that don't support
symlinking (such as Windows), the $(LN_S) macro is defined
as 'cp -p' instead.  This works fine, except that since the
target of the symlink is a relative path, the copy will only
succeed if the current directory is the directory that contains
the file.

The solution to this problem suggested in the Autotools manual
(see http://www.gnu.org/software/automake/manual/automake.html#Extending)
is to simply cd into the correct directory first.  This patch
makes that change for the symlinks that are being made during install.

[ edit: use relative path for the link name as well ]
2013-04-19 20:07:02 -07:00
Matt Fischer
563b0ff314 Added --enable-setjmp
Currently, libunwind-setjmp is built whenever local unwinding is
built.  This patch adds an explicit flag to control it instead.
The default if not specified is to follow the old behavior.
2013-04-19 17:08:03 -05:00
Matt Fischer
abb0957198 Added --enable-ptrace
This change adds a manual override to control building of the ptrace
library, similar to the existing --enable-coredump option.  The
default is set based on the existence of sys/ptrace.h, allowing it
to be automatically disabled for platforms that do not have ptrace.
2013-04-19 16:46:24 -05:00
Arun Sharma
f92ecb6127 dwarf: Fix memory leak
Need to clean up by calling put_unwind_info() on code paths where fetch_proc_info() fails.

Signed-off-by: Arun Sharma <asharma@fb.com>
2013-04-02 21:19:56 -07:00
dave lerner
8028a537dd unw_is_signal_frame should return false/0 for bad addresses
access_mem() could fail and return a non-zero value, which callers
interpret as boolean true.

Signed-off-by: Dave Lerner <dave.lerner@windriver.com>
2013-03-10 14:59:44 -07:00
Chris January
679b65cd22 dwarf: Fix the order of the operands to the relational operators
dwarf_eval_expr are reversed compared with the DWARF spec.
2013-02-04 19:36:33 -08:00
Ladislav Michl
10b064ffe9 Support building with older compilers.
Add a check for __builtin_unreachable.
2012-11-24 21:37:07 -08:00
Arun Sharma
d7322f0f64 Update to v1.1 2012-10-05 21:54:17 -07:00
Tommi Rantala
adae66d2e2 Fix x86-64 debug build -Wformat warning
src/x86_64/Gstep.c: In function '_ULx86_64_step':
src/x86_64/Gstep.c:204:4: warning: unknown conversion type character 'r' in format [-Wformat]
2012-10-05 21:24:36 -07:00
Tommi Rantala
96c1bca2b0 Fix header inclusion order in `src/elfxx.c'
Make sure that we have included `config.h' before checking HAVE_LZMA.
This makes a difference at least on MIPS, where we include `elfxx.c'
directly. For other platforms this does not make much difference, as we
are already indirectly including `config.h' from other headers.
2012-10-05 21:24:30 -07:00
Arun Sharma
cb3fbbb172 Reduce the number of ifdefs in machine independent code
Hopefully we don't have too many GNUC users who don't have
fetch and add.
2012-09-29 12:22:47 -07:00