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.
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
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 ]
The crasher test is used as part of the coredump test suite,
but is being built regardless of whether that support has been
configured or not. This patch changes the build so that it is
only built when coredump support is enabled.
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.
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.
When libunwind is built for a ppc32 target, for example the yocto distro
for qemuppc, and configured with options --enable-debug-frame and
--enable-debug, then the shortfall in the size of the 'opaque' member
of the unw_cursor_t type throws an assert in mi_init().
The calculations for the ppc32 target size of UNW_TDEP_CURSOR_LEN in
the comment block based on the register count plus overhead sum to 280
words. However, that value is not reflected in the ppc32 header, which
defines the size as 200. Fixing the ppc32 header value to 280 makes the
unw_cursor_t type larger than struct cursor type and the assert tests
true at run-time in mi_init().
Signed-off-by Dave Lerner <dave.lerner@windriver.com>
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.
We do not modify `_UPT_reg_offset', so let's set it `const'. Given that
the array is exported, it's technically possible that some library user
is actually modifying the content, but does not seem really all that
likely.
src/ptrace/_UPT_find_proc_info.c: In function 'get_unwind_info':
src/ptrace/_UPT_find_proc_info.c:67: warning: implicit declaration of function 'ia64_find_unwind_table'
Add support for the 32bit SuperH architecture running Linux.
Specifically, support is added for SH4, and support for earlier SH
versions and to the 64bit SH5 are left out. This was tested in qemu with
a little-endian SH4 debian image & GCC 4.7 cross compiler.
Add MIPS support to the coredump library. Explicit support for the MIPS
program counter register is added so that we can start backtracing from
the program counter value we read from a core dump. The PC register
support was not strictly required for local backtracing, and we will in
fact just plug the return address to the PC register in getcontext().
I have only tested the 32bit "OABI" paths.
We do not really need to care if the system provides `backtrace()',
since we will want to test the one provided in libunwind, not the one
that is provided by the system. The `backtrace()' calls should already
be aliased to `unw_backtrace()', but if that is not working for whatever
reason, we can call `unw_backtrace()' explicitly.
We are building only a UNW_LOCAL_ONLY build of `test-nocalloc', and the
"generic" build would not be very interesting. Roll the whole test into
`Ltest-nocalloc.c'.
Only build a "local-only" version of `rs-race'. The "generic" build is
not very entertaining, as the `unw_set_caching_policy()' calls
manipulate the `unw_local_addr_space' in libunwind-$arch.so, while the
`backtrace()' calls use the address space object from libunwind.so
behind the scenes.
Cleanup dynamically allocated memory before exit in tests in a few
places where missing. While such cleanups right before exit do not
usually make much sense (as the operating system would cleanup anyway,
so manual cleanups only burn CPU cycles), we will want to catch any
potential problems in libunwind related to the cleanups. This also stops
valgrind complaining about unreleased memory.
