Summary:
This improves unwind performance quite substantially, and follows
a somewhat similar approach used in libgcc_s as described in the
thread here:
https://gcc.gnu.org/ml/gcc/2005-02/msg00625.html
On certain extremely exception heavy internal tests, the time
drops from about 80 minutes to about five minutes.
Subscribers: libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D75954
Summary:
Copying all of the saved register state on every entry to
parseInstruction is a severe performance contraint, especially
because most of this saved state is never used. On x86 linux
this is about 560 bytes, and will be more on other platforms.
When performance testing libunwind, this memcpy appears at the
top of nearly all our tests.
By only saving this state as needed, we see increasing in performance
of around 2.5% for the ctak test here.
https://github.com/clasp-developers/ctak
Certain internal extremely exception-heavy tasks run in about 2/3
the time.
Note that by stashing the new boolean inside what had been padding in
the original structure, this uses no additional memory.
Subscribers: fedor.sergeev, libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D75692
We were seeing non-deterministic binary size differences depending on which
toolchain was used to build fuchsia. This is because libunwind embeded the
FILE path into a logging macro, even for release builds, which makes the code
dependent on the build directory.
This removes the file and line number from the error message. This is
consistent with how other runtimes report error, e.g.
https://github.com/llvm/llvm-project/blob/master/libcxxabi/src/abort_message.cpp#L30.
Differential Revision: https://reviews.llvm.org/D75890
Summary:
- Executable segment is usually segment 3. Look there for the address first.
- GNU_EH_FRAME_HEADER segment is usually near the end. Iterate from the end.
- Exit early if both phdrs have been found.
This is the last cl before a patch to cache the information this function
finds.
Subscribers: libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D75781
Summary:
This further cleans up the control flow and makes it easier to
optimize and replace portions in a subsequent patch.
This should be NFC, but given the amount of #ifdeffing here,
it may not be. So will watch the buildbots closely.
Also, as this is purely moving existing code around, I plan to
ignore the lint errors.
Reviewers: compnerd, miyuki, mstorsjo
Subscribers: libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D75705
Summary:
This cleans up control flow inside findUnwindSections, and will make
it easier to replace this code in a following patch. Also, expose the
data structure to allow use by a future replacment function.
Reviewers: mstorsjo, miyuki
Subscribers: krytarowski, libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D75637
This reverts commit d93371238e.
The commit broke the build in several configurations (including
Windows and bare-metal). For details see comments in
https://reviews.llvm.org/D75480
to allow use by a future replacment function.
Summary: [Refactor] Promote nameless lambda to fully named function, allowing easy replacement in following patch.
Subscribers: krytarowski, libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D75480
The OSX_ARCHITECTURES property is supposed to add the -arch flag when
targeting Apple platforms. However, due to a bug in CMake
(https://github.com/Kitware/CMake/blob/master/Source/cmLocalGenerator.cxx#L1780),
this does not apply to assembly files. This results in a linker error
when trying to build libunwind for i386 on MacOS.
rdar://59642189
parseInstructions() doesn't always process the whole set of DWARF
instructions for a frame. It will stop once the target PC is reached, or
if malformed instructions are found. So, for example, if we have an
instruction sequence like this:
```
<start>
...
DW_CFA_remember_state
...
DW_CFA_advance_loc past the location we're unwinding at (pcoffset in parseInstructions() main loop)
...
DW_CFA_restore_state
<end>
```
... the saved state will never be freed, even though the
DW_CFA_remember_state opcode has a matching DW_CFA_restore_state later
in the sequence.
This change adds code to free whatever is left on rememberStack after
parsing the CIE and the FDE instructions.
Differential Revision: https://reviews.llvm.org/D66904
This patch renames `__personality_routine` to `_Unwind_Personality_Fn`
in `unwind.h`. Both `unwind.h` from clang and GCC headers use this name
instead of `__personality_routine`. With this patch one is also able to
build libc++abi with libunwind support on Windows.
Patch by Markus Böck!
```
src/UnwindCursor.hpp:1344:51: error: operator '?:' has lower precedence than '|';
'|' will be evaluated first [-Werror,-Wbitwise-conditional-parentheses]
_info.flags = isSingleWordEHT ? 1 : 0 | scope32 ? 0x2 : 0; // Use enum?
~~~~~~~~~~~ ^
src/UnwindCursor.hpp:1344:51: note: place parentheses around the '|' expression
to silence this warning
_info.flags = isSingleWordEHT ? 1 : 0 | scope32 ? 0x2 : 0; // Use enum?
^
( )
src/UnwindCursor.hpp:1344:51: note: place parentheses around the '?:' expression
to evaluate it first
_info.flags = isSingleWordEHT ? 1 : 0 | scope32 ? 0x2 : 0; // Use enum?
^
( )
```
But `0 |` is a no-op for either of those two interpretations, so I think
what was meant here was
```
_info.flags = (isSingleWordEHT ? 1 : 0) | (scope32 ? 0x2 : 0); // Use enum?
```
Previously, if `isSingleWordEHT` was set, bit 2 would never be set. Now
it is. From what I can tell, the only thing that checks these bitmask is
ProcessDescriptors in Unwind-EHABI.cpp, and that only cares about bit 1,
so in practice this shouldn't have much of an effect.
Differential Revision: https://reviews.llvm.org/D73890
When targeting mingw, current CMake (3.16) fails to get the right
flags for assembly source files for windows gnu/clang targets
(see https://gitlab.kitware.com/cmake/cmake/merge_requests/4287
for a fix), causing builds to fail due to `-fPIC` being unsupported
in clang for mingw targets
In the meantime, restore the behaviour from before c48974ffd7
selectively on mingw targets, treating the assembly files as C.
Differential Revision: https://reviews.llvm.org/D73436
After this change, we need to explicitly list the languages the
project uses, otherwise the assembly source files won't get built
at all.
Previously (before that commit), the assembly source files were
simply treated as C.
The toplevel llvm CMakeLists.txt adds these three languages, so
when building libunwind integrated as part of that, it works fine.
I believe this is an oversight from the import of libunwind into its own
library from libc++abi.
In libc++abi, these files had the .s suffix, which indicates that the file
is a preprocessed assembly source file. This caused problems because the
files rely upon preprocessors to guard target-specific blocks.
To fix this, the CMakeLists file marked these files as C so that the
preprocessor would be run over them, but then the compiler would correctly
identify the files as assembly and compile them as such.
When imported to libunwind, these files were (correctly) renamed with .S
suffixes, which are non-preprocessed assembly. Thus, we no longer need the
C language property.
The benefit here is that the files can now benefit from CMAKE_ASM_FLAGS
rather than CMAKE_C_FLAGS.
Patch By: JamesNagurne
Differential Revision: https://reviews.llvm.org/D72952
reg is unsigned type and used here for getting array element from the end by
negating it. negation of unsigned can result in large number and array access
with that index will result in segmentation fault.
Fixes: https://bugs.llvm.org/show_bug.cgi?id=43872
Patched by: kamlesh kumar
Differential Revision: https://reviews.llvm.org/D69893
Summary:
Add unwinding support for 64-bit RISC-V.
This is from the FreeBSD implementation with the following minor
changes:
- Renamed and renumbered DWARF registers to match the RISC-V ABI [1]
- Use the ABI mneumonics in getRegisterName() instead of the exact
register names
- Include checks for __riscv_xlen == 64 to facilitate adding the 32-bit
ABI in the future.
[1] https://github.com/riscv/riscv-elf-psabi-doc/blob/master/riscv-elf.md
Patch by Mitchell Horne (mhorne)
Reviewers: lenary, luismarques, compnerd, phosek
Reviewed By: lenary, luismarques
Subscribers: arichardson, sameer.abuasal, abidh, asb, aprantl, krytarowski, simoncook, kito-cheng, christof, shiva0217, rogfer01, rkruppe, PkmX, psnobl, benna, lenary, s.egerton, luismarques, emaste, cfe-commits
Differential Revision: https://reviews.llvm.org/D68362
Summary:
Relands https://reviews.llvm.org/D70815.
The original commit set `CMAKE_TRY_COMPILE_TARGET_TYPE` to
`STATIC_LIBRARY` globally in libunwind/CMakeLists.txt, which effectively
disabled the linking step in CMake checks.
This broke some builds (see 938c70b86c).
Here we set CMAKE_TRY_COMPILE_TARGET_TYPE to
STATIC_LIBRARY only when checking for presence of the `-funwind-tables`
flag, and then set it back to the original value so it doesn't affect
other checks.
Reviewers: mstorsjo, jfb
Subscribers: mgorny, christof, libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D71117
This reverts commit b3fdf33ba6.
This change broke building libunwind for Windows/MinGW, and broke
on aspect of the CMake tests in libunwind in general.
After set(CMAKE_TRY_COMPILE_TARGET_TYPE STATIC_LIBRARY), CMake
skips the linking step in tests, but cmake/config-ix.cmake also
does a few checks for functions in libraries (looking for whether
-lc provides fopen and whether -ldl provides dladdr).
As CMake only tests building a static library, these tests
incorrectly succeed and CMake concludes "Looking for fopen in c -
found" and "Looking for dladdr in dl - found", while building
then fails at the end with errors about unable to find -lc and -ldl.
Summary:
Or, rather, don't accidentally forget to pass it.
This is aimed to solve the problem discussed in [this thread](http://lists.llvm.org/pipermail/llvm-dev/2019-November/136890.html), and to fix [a year-old bug](https://bugs.llvm.org/show_bug.cgi?id=38468).
TL;DR: when building libunwind for ARM Linux, we **need** libunwind to be built with the `-funwind-tables` flag, because, well ARM EHABI needs unwind info produced by this flag. Without the flag all the procedures in libunwind are marked `.cantunwind`, which causes all sorts of bad things. From `_Unwind_Backtrace` not working, to C++ exceptions not being caught (which is the aforementioned bug is about).
Previously, this flag was not added because the CMake check `add_compile_flags_if_supported(-funwind-tables)` produced a false negative. Why? With this flag, the compiler generates calls to the `__aeabi_unwind_cpp_pr0` symbol, which is defined in libunwind itself and obviously is not available at configure time, before libunwind is built. This led to failure at link time during the CMake check. We handle this by disabling the linker for CMake checks in linbunwind.
Also, this patch introduces a lit feature `libunwind-arm-ehabi`, which is used to mark the `signal_frame.pass.cpp` test as unsupported (as was advised by @miyuki in D70397).
Reviewers: peter.smith, phosek, EricWF, compnerd, jroelofs, saugustine, miyuki, jfb
Subscribers: mgorny, kristof.beyls, christof, libcxx-commits, miyuki
Tags: #libc
Differential Revision: https://reviews.llvm.org/D70815
996e62eef7 added Linux-specific dependent libraries to libunwind
sources. As a result, building libunwind with modern LLD on *BSD
started failing due to trying to link libdl. Instead, add those
libraries only if they were detected by CMake.
While technically we could create a long list of systems that need -ldl
and -lpthread, maintaining a duplicate list makes little sense when
CMake needs to detect it for non-LLD systems anyway. Remove existing
system exceptions since they should be covered by the CMake check
anyway.
Remove -D_LIBUNWIND_HAS_COMMENT_LIB_PRAGMA since it is no longer
explicitly needed, if we make the library-specific defines dependent
on presence of this pragma support.
Differential Revision: https://reviews.llvm.org/D70868
Summary:
This patch adjusts the signal_frame.pass.cpp to pass on Arm targets:
* When Arm EHABI is used the unwinder does not use DWARF, hence the
DWARF-specific check unw_is_signal_frame() must be disabled.
* Certain C libraries don't include EH tables, so the unwinder must
not try to step out of main(). The patch moves the test code out of
main() into a separate function to avoid this.
Reviewers: saugustine, ostannard, phosek, jfb, mclow.lists
Reviewed By: saugustine
Subscribers: dexonsmith, aprantl, kristof.beyls, christof, libcxx-commits, pbarrio, labrinea
Tags: #libc
Differential Revision: https://reviews.llvm.org/D70397
A "signal frame" is a function or block of code where execution arrives via a signal or interrupt, rather than via a normal call instruction. In fact, a particular instruction is interrupted by the signal and needs to be restarted. Therefore, when the signal handler is complete, execution needs to return to the interrupted instruction, rather than the instruction immediately following the call instruction, as in a normal call.
Stack unwinders need to know this to correctly unwind signal frames. Dwarf handily provides an "S" in the CIE augmentation string to describe this case, and the libunwind API provides various functions to for unwinders to determine it,.
The llvm libunwind implementation correctly sets it's internal variable "isSignalFrame" when initializing an unwind context. However, upon stepping up the stack, the current implementation correctly reads the augmentation string and sets it in the CIE info (which it then discards), libunwind doesn't update it's internal unwind context data structure.
This change fixes that, and provides compatibility with both the canonical libunwind and the libgcc implementation.
Reviewers: jfb
Subscribers: christof, libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D69677
Have CMake treat the unwind libraries as C libraries rather than C++.
There is no C++ runtime dependency at runtime. This ensures that we do
not accidentally end up with a link against the C++ runtime.
We need to explicitly reset the implicitly linked libraries for C++ to
ensure that we do not have CMake force the link against the C++ runtime.
This adjustment should enable the NetBSD bots to be happy with this
change.
Disable the type information emission for libunwind. libunwind does not
use `dynamic_cast`. This results in a smaller binary, and more
importantly, avoids the dependency on libc++abi. This ensures that we
have complete symbol resolution of symbols on ELF targets without
linking to the C++ runtime support library. This change avoids the
emission of a reference to `__si_class_type_info`.
The unwinder should not depend on libc++. In fact, we do not end up
with a link against libc++ as we do not have a dependency on libc++ at
runtime. This ensures that we link with `clang` rather than `clang++`.
Summary:
Fix the arm_section_length count. The meaning of the arm_section_length
field changed from num-of-elements to num-of-bytes when the
dl_unwind_find_exidx special case was removed (D30306 and D30681). The
special case was restored in D39468, but that patch didn't account for the
change in arm_section_length's meaning.
That patch worked when it was applied to the NDK's fork of libunwind,
because it never removed the special case in the first place, and the
special case is probably disabled in the Android platform's copy of
libunwind, because __ANDROID_API__ is greater than 21.
Turn the dl_unwind_find_exidx special case on unconditionally for Bionic.
Bionic's dl_unwind_find_exidx is much faster than using dl_iterate_phdr.
(e.g. Bionic stores exidx info on an internal soinfo object.)
Reviewers: thomasanderson, srhines, danalbert, ed, keith.walker.arm, mclow.lists, compnerd
Reviewed By: srhines, danalbert
Subscribers: srhines, kristof.beyls, christof, libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D68972
llvm-svn: 375275
Summary:
The workaround added in https://reviews.llvm.org/rL299575 appears to be
working around a bug in Android JB 4.1.x and 4.2.x (API 16 and 17).
Starting in API 16, Android added support for PIE binaries, but the
dynamic linker failed to initialize dlpi_addr to the address that the
executable was loaded at. The bug was fixed in Android JB 4.3.x (API 18).
Improve the true load bias calculation:
* The code was assuming that the first segment would be the PT_PHDR
segment. I think it's better to be explicit and search for PT_PHDR. (It
will be almost as fast in practice.)
* It's more correct to use p_vaddr rather than p_offset. If a PIE
executable is linked with a non-zero image base (e.g. lld's
-Wl,--image-base=xxxx), then we must use p_vaddr here.
The "phdr->p_vaddr < image_base" condition seems unnecessary and maybe
slightly wrong. If the OS were to load a binary at an address smaller than
a vaddr in the binary, we would still want to do this workaround.
The workaround is safe when the linker bug isn't present, because it
should calculate an image_base equal to dlpi_addr. Note that with API 21
and up, this workaround should never activate for dynamically-linked
objects, because non-PIE executables aren't allowed.
Consolidate the fix into a single block of code that calculates the true
image base, and make it clear that the fix no longer applies after API 18.
See https://github.com/android/ndk/issues/505 for details.
Reviewers: mclow.lists, srhines, danalbert, compnerd
Reviewed By: compnerd
Subscribers: srhines, krytarowski, christof, libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D68971
llvm-svn: 374969
There are few differences in compile flags introduced in r374606
which are causing libcxx-libcxxabi-libunwind-armv8-linux to fail.
This change should address all of those, I've compared the generated
build file from before r374606 and with this change and the set of
flags is the same modulo order.
llvm-svn: 374624
libunwind was using its own set of macros/functions for flag checking
which was similar but different from libc++ and libc++abi. This made
it difficult to replicate the same checks across projects, in fact
there were some checks that appear to have been copy & pasted from
another project and that were broken in the standalone libunwind build.
This change refactors flag checks to match libc++ and libc++abi using
a copy of HandleLibunwindFlags.cmake which is derived from the versions
used by the other projects. This also paves a road to deduplicating and
unifying HandleLibunwindFlags.cmake, HandleLibcxxabiFlags.cmake and
HandleLibcxxFlags.cmake post monorepo switch.
Differential Revision: https://reviews.llvm.org/D68855
llvm-svn: 374606
ARM EHABI unwinding tables only store the start address of each function, so the
last function is assumed to cover the entire address space after it. The test
picks an address on the stack assuming that it's in no function, but because of
the above it's actually resolved to the last function. Fix this by using address
0 instead.
Differential Revision: https://reviews.llvm.org/D68387
llvm-svn: 373628
Move the definition of Elf_Addr typedef to the only place it is used, to avoid:
```
llvm-project/libunwind/src/AddressSpace.hpp:501:28: warning: unused typedef 'Elf_Addr' [-Wunused-local-typedef]
```
when compiling for Android with _LIBUNWIND_ARM_EHABI defined and
_LIBUNWIND_SUPPORT_DWARF_UNWIND not defined.
Patch by Joel Klinghed!
llvm-svn: 372427
If unwind info is not available at the current IP, unw_get_proc_info should
return a zero-filled structure rather than the info of the previous IP.
This change also makes unw_get_proc_info return UNW_ENOINFO instead of
UNW_ESUCCESS.
Patch by Amanieu d'Antras!
llvm-svn: 372407
r362048 added support for ELF dependent libraries, but broke Android
build since Android does not have libpthread. Remove the dependency on
the Android build.
Differential Revision: https://reviews.llvm.org/D65098
llvm-svn: 366734
This is a small fix for https://reviews.llvm.org/D64996. The types of
w0 and w1 in _Unwind_VRS_Get must be uint64_t, not uint32_t.
Committing as obvious.
llvm-svn: 366701
Summary:
The function Unwind-EHABI.cpp:_Unwind_VRS_Pop loads the saved values of
64-bit FP registers as two 32-bit words because they might not be
8-byte aligned. Combining these words into a 64-bit value has to be
done differently on big-endian platforms.
Reviewers: ostannard, john.brawn, dmgreen
Reviewed By: ostannard
Subscribers: kristof.beyls, christof, libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D64996
llvm-svn: 366587
Summary:
The function getByte is dependent on endianness and the current
behavior is incorrect on big-endian targets.
This patch fixes the issue.
Reviewers: phosek, ostannard, dmgreen, christof, chill
Reviewed By: ostannard, chill
Subscribers: chill, christof, libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D64402
llvm-svn: 365505
Summary:
The AArch64 version of the libunwind function which restores the
CPU state and resumes execution is not interrupt-safe. It restores
the target value of SP before loading the floating-point registers
from the context struct, but that struct is allocated on the stack
which is being deallocated. This means that if an interrupt occurs
during this function, and uses a lot of stack space, it could
overwrite the values about to be loaded into the floating-point
registers.
This patch fixes the issue.
Patch by Oliver Stannard.
Reviewers: phosek, chill
Reviewed By: chill
Subscribers: chill, javed.absar, kristof.beyls, christof, LukeCheeseman, pbarrio, olista01, libcxx-commits
Tags: #libc
Differential Revision: https://reviews.llvm.org/D63006
llvm-svn: 363545
Summary:
This updates all places in documentation that refer to "Mac OS X", "OS X", etc.
to instead use the modern name "macOS" when no specific version number is
mentioned.
If a specific version is mentioned, this attempts to use the OS name at the time
of that version:
* Mac OS X for 10.0 - 10.7
* OS X for 10.8 - 10.11
* macOS for 10.12 - present
Reviewers: JDevlieghere
Subscribers: mgorny, christof, arphaman, cfe-commits, lldb-commits, libcxx-commits, llvm-commits
Tags: #clang, #lldb, #libc, #llvm
Differential Revision: https://reviews.llvm.org/D62654
llvm-svn: 362113
This fixes the issue introduced by r362048 where we always use
pragma comment(lib, ...) for dependent libraries when the compiler
is Clang, but older Clang versions don't support this pragma so
we need to check first if it's supported before using it.
llvm-svn: 362055
As of r360984, LLD supports dependent libraries feature for ELF.
libunwind, libc++abi and libc++ have library dependencies: libdl librt
and libpthread, which means that when libunwind and libc++ are being
statically linked (using -static-libstdc++ flag), user has to manually
specify -ldl -lpthread which is onerous.
This change includes the lib pragma to specify the library dependencies
directly in the source that uses those libraries. This doesn't make any
difference when using linkers that don't support dependent libraries.
However, when using LLD that has dependent libraries feature, users no
longer have to manually specifying library dependencies when using
static linking, linker will pick the library automatically.
Differential Revision: https://reviews.llvm.org/D62090
llvm-svn: 362048
Fix two issues that caused libcxx source path not to be inferred
correctly when not specified explicitly:
1. get_lit_conf() uses default value only if the lit variable is set
to None. Due to the mehod of substituting lit.site.cfg, they were
"" rather than None when unset, effectively causing the default never
to apply. Instead, use 'or' construct to use the default whenever
get_lit_conf() returns a false value.
2. If os.path.join() is given a component starting with '/', it takes
it to be an absolute path and ignores everything preceding it.
Remove the slash to correctly append subdirectory.
With these two fixes, libunwind tests start working on NetBSD buildbot
again.
Differential Revision: https://reviews.llvm.org/D62005
llvm-svn: 361931
This change is a consequence of the discussion in "RFC: Place libs in
Clang-dedicated directories", specifically the suggestion that
libunwind, libc++abi and libc++ shouldn't be using Clang resource
directory. Tools like clangd make this assumption, but this is
currently not true for the LLVM_ENABLE_PER_TARGET_RUNTIME_DIR build.
This change addresses that by moving the output of these libraries to
lib/$target/c++ and include/c++ directories, leaving resource directory
only for compiler-rt runtimes and Clang builtin headers.
Differential Revision: https://reviews.llvm.org/D59168
llvm-svn: 361432
Clang integrated assembler was unable to build libunwind PPC32 assembly code,
present in functions used to save/restore register context.
This change consists in replacing the assembly style used in libunwind source,
to one that is compatible with both Clang integrated assembler as well as
GNU assembler.
Patch by Leandro Lupori!
Differential Revision: https://reviews.llvm.org/D61792
llvm-svn: 360862
This change makes each unwind step inspect the instruction at the
return address and, if needed, read r2 from its saved location and
modify the context appropriately.
The unwind logic is able to handle both ELFv1 and ELFv2 stacks.
Reported by Bug 41050
Patch by Leandro Lupori!
Differential Revision: https://reviews.llvm.org/D59694
llvm-svn: 360861
calls into the pthread library use weak symbols.
This option allows libpthread to be a weak dependency rather
than a hard one.
Differential Revision: https://reviews.llvm.org/D60285
llvm-svn: 360610
This change introduces support for building libuwind. The library
build should be complete, but not all CMake options have been
replicated in GN. We also don't support tests yet.
We only support two stage build at the moment.
Differential Revision: https://reviews.llvm.org/D60370
llvm-svn: 359804
For builds with SJLJ, there is no __unw_getcontext symbol. On Windows,
the weak alias macro also expands to a dllexport directive, which fails
if the symbol doesn't exist.
Differential Revision: https://reviews.llvm.org/D60251
llvm-svn: 357711
This is not necessary for ELF since .globl and .weak are mutually
exclusive, but is necessary for Mach-O otherwise the symbol isn't
visible externally.
Differential Revision: https://reviews.llvm.org/D60245
llvm-svn: 357671
libunwind defines the _Unwind_* ABI used by libc++abi. This ABI is a
stable quasi-standard common between multiple implementations such as
LLVM and GNU. The _U* symbol name space is also safely within the symbol
name space that standard C & C++ reserve for the implementation.
Futhermore, libunwind also defines several unw_* symbols, and references
these from the _Unwind_* entry points so the standard/reserved part of
the ABI is dependent on the unw_* part of the ABI. This is not OK for a
C or C++ implementation. The unw_* symbols are reserved for C and extern
"C" used by application code.
This change renames each unw_* function to __unw* and adds a weak alias
unw_* to keep the public <libunwind.h> ABI unchanged for backwards
compatibility. Every reference to unw_* in the implementation has been
changed to use __unw* so that if other unw_* definitions are in force
because nothing uses <libunwind.h> in a particular program, no _Unwind*
code path depends on any unw_* symbol. Furthemore, __unw_* symbols are
hidden, which saves PLT overhead in the shared library case.
In the future, we should cconsider untangling the unw_* API/ABI from the
_Unwind_* API/ABI. The internal API backing the _Unwind_* ABI
implementation should not rely on any nonstandard symbols not in the
implementation-reserved name space. This would then allow separating the
_Unwind_* API/ABI from unw_* entirely, but that's a more substantial
change that's going to require more significant refactoring.
Differential Revision: https://reviews.llvm.org/D59921
llvm-svn: 357640
This change is a consequence of the discussion in "RFC: Place libs in
Clang-dedicated directories", specifically the suggestion that
libunwind, libc++abi and libc++ shouldn't be using Clang resource
directory. Tools like clangd make this assumption, but this is
currently not true for the LLVM_ENABLE_PER_TARGET_RUNTIME_DIR build.
This change addresses that by moving the output of these libraries to
lib/<target> and include/ directories, leaving resource directory only
for compiler-rt runtimes and Clang builtin headers.
Differential Revision: https://reviews.llvm.org/D59013
llvm-svn: 355665
If we're not in a standalone build, this variable should be already
set, so there's no need to set it again or to cache it.
Differential Revision: https://reviews.llvm.org/D57993
llvm-svn: 353915
We're building tests with -nostdlib which means that we need to
explicitly include the builtins library. When using libgcc (default)
we can simply include -lgcc_s on the link line, but when using
compiler-rt builtins we need a complete path to the builtins library.
This path is already available in CMake as <PROJECT>_BUILTINS_LIBRARY,
so we just need to pass that path to lit and if config.compiler_rt is
true, link it to the test.
Prior to this patch, running tests when compiler-rt is being used as
the builtins library was broken as all tests would fail to link, but
with this change running tests when compiler-rt bultins library is
being used should be supported.
Differential Revision: https://reviews.llvm.org/D56701
llvm-svn: 353208
There are several changes:
- Don't stringify Pythonized bools (that's why we're Pythonizing them)
- Support specifying target and sysroot via CMake variables
- Use consistent spelling for --target, --sysroot, --gcc-toolchain
llvm-svn: 353137
CMake has a standard way of setting target triple, sysroot and external
toolchain through CMAKE_<LANG>_COMPILER_TARGET, CMAKE_SYSROOT and
CMAKE_<LANG>_COMPILER_EXTERNAL_TOOLCHAIN. These are turned into
corresponding --target=, --sysroot= and --gcc-toolchain= variables add
included appended to CMAKE_<LANG>_FLAGS.
libunwind, libc++abi, libc++ provides their own mechanism through
<PROJECT>_TARGET_TRIPLE, <PROJECT>_SYSROOT and <PROJECT>_GCC_TOOLCHAIN
variables. These are also passed to lit via lit.site.cfg, and lit config
uses these to set the corresponding compiler flags when building tessts.
This means that there are two different ways of setting target, sysroot
and toolchain, but only one is properly supported in lit. This change
extends CMake build for libunwind, libc++abi and libc++ to also support
the CMake variables in addition to project specific ones in lit.
Differential Revision: https://reviews.llvm.org/D57670
llvm-svn: 353084
While Clang automatically generates the code for placement new,
g++ doesn't do that so we need to provide our own definition.
Differential Revision: https://reviews.llvm.org/D57455
llvm-svn: 352966
This is unfinished, unused and incomplete. This could be brought back in
the future if there's a desire to build a more complete implementation,
but at the moment it's just bitrotting.
Differential Revision: https://reviews.llvm.org/D57252
llvm-svn: 352965
When linking library dependencies, we shouldn't need to export linked
libraries to dependents. We should be explicit about this in
target_link_libraries, otherwise other targets that depend on these such
as sanitizers get repeated (and possibly even conflicting) dependencies.
Differential Revision: https://reviews.llvm.org/D57456
llvm-svn: 352688
When linking library dependencies, we shouldn't need to export linked
libraries to dependents. We should be explicit about this in
target_link_libraries, otherwise other targets that depend on these such
as sanitizers get repeated (and possibly even conflicting) dependencies.
Differential Revision: https://reviews.llvm.org/D57456
llvm-svn: 352654
This is useful when the static libunwind library is being linked into
shared libraries that may be used in with other shared libraries that
use different unwinder. We want to avoid avoid exporting libunwind
symbols in those cases. This achieved by a new CMake option which can be
enabled by libunwind vendors as needed.
The same CMake option has already been added to libc++ and libc++abi in
D55404 and D56026.
Differential Revision: https://reviews.llvm.org/D57107
llvm-svn: 352559
We haven't eliminated C++ library dependency altogether in D57251,
UnwindCursor.hpp had an unused dependency on <algorithm> which was
pulling in other C++ headers. Removing that dependency also revealed
(correctly) that we need our own global placement new declaration. Now
libunwind should be independent of the C++ library.
Differential Revision: https://reviews.llvm.org/D57262
llvm-svn: 352553
This fixes most references to the paths:
llvm.org/svn/
llvm.org/git/
llvm.org/viewvc/
github.com/llvm-mirror/
github.com/llvm-project/
reviews.llvm.org/diffusion/
to instead point to https://github.com/llvm/llvm-project.
This is *not* a trivial substitution, because additionally, all the
checkout instructions had to be migrated to instruct users on how to
use the monorepo layout, setting LLVM_ENABLE_PROJECTS instead of
checking out various projects into various subdirectories.
I've attempted to not change any scripts here, only documentation. The
scripts will have to be addressed separately.
Additionally, I've deleted one document which appeared to be outdated
and unneeded:
lldb/docs/building-with-debug-llvm.txt
Differential Revision: https://reviews.llvm.org/D57330
llvm-svn: 352514
The existing typedef of unw_fpreg_t to uint64_t might work and be
correct for the ARM_EHABI case, but for dwarf, some cases in e.g.
DwarfInstructions.hpp convert between double and unw_fpreg_t.
When converting implicitly between double and unw_fpreg_t (uint64_t),
the values get interpreted as integers and converted to float and vice
versa, while the correct thing would be to keep the same bit pattern.
Avoid the whole issue by using the same definition of unw_fpreg_t
as all other architectures, when using dwarf unwinding on ARM.
Change assembler functions to take a void pointer instead of
unw_fpreg_t pointer, to avoid having a different mangled symbol name
depending on the actual value of this typedef.
Differential Revision: https://reviews.llvm.org/D57001
llvm-svn: 352461
We haven't eliminated C++ library dependency altogether in D57251,
UnwindCursor.hpp had an unused dependency on <algorithm> which was
pulling in other C++ headers. Removing that dependency also revealed
(correctly) that we need our own global placement new declaration. Now
libunwind should be independent of the C++ library.
Differential Revision: https://reviews.llvm.org/D57262
llvm-svn: 352384
The check_library_exists CMake uses a custom symbol definition. This
is a problem when checking for C library symbols because Clang
recognizes many of them as builtins, and returns the
-Wbuiltin-requires-header (or -Wincompatible-library-redeclaration)
error. When building with -Werror which is the default, this causes
the check_library_exists check fail making the build think that C
library isn't available.
To avoid this issue, we should use a symbol that isn't recognized by
Clang and wouldn't cause the same issue. __libc_start_main seems like
reasonable choice that fits the bill.
Differential Revision: https://reviews.llvm.org/D57142
llvm-svn: 352341
The rest of libunwind already uses placement new, these are the only
places where non-placement new is being used introducing undesirable
C++ library dependency.
Differential Revision: https://reviews.llvm.org/D57251
llvm-svn: 352245
Recent Linux kernel release has introduced a bug as part of the ORC
rollout where the vDSO has a valid .eh_frame section, but it's missing
the .eh_frame_hdr section and GNU_EH_FRAME segment has zero size. This
causes libunwind to abort which breaks programs that use libunwind.
The other unwinder implementation (libgcc, non-gnu) instead silently
bail out unless being compiled as debug. This change modifies libunwind
to use the same strategy.
Differential Revision: https://reviews.llvm.org/D57081
llvm-svn: 352016
Either adjust the format string to use a more exact type, or add casts
(for cases when printing pointers to structs/objects with a %p
format specifier).
Differential Revision: https://reviews.llvm.org/D56982
llvm-svn: 351876
When built within the llvm runtimes directory, the runtimes
CMakeLists.txt adds the same.
Differential Revision: https://reviews.llvm.org/D56981
llvm-svn: 351875
all missed!
Thanks to Alex Bradbury for pointing this out, and the fact that I never
added the intended `legacy` anchor to the developer policy. Add that
anchor too. With hope, this will cause the links to all resolve
successfully.
llvm-svn: 351731
to reflect the new license. These used slightly different spellings that
defeated my regular expressions.
We understand that people may be surprised that we're moving the header
entirely to discuss the new license. We checked this carefully with the
Foundation's lawyer and we believe this is the correct approach.
Essentially, all code in the project is now made available by the LLVM
project under our new license, so you will see that the license headers
include that license only. Some of our contributors have contributed
code under our old license, and accordingly, we have retained a copy of
our old license notice in the top-level files in each project and
repository.
llvm-svn: 351648
This installs the new developer policy and moves all of the license
files across all LLVM projects in the monorepo to the new license
structure. The remaining projects will be moved independently.
Note that I've left odd formatting and other idiosyncracies of the
legacy license structure text alone to make the diff easier to read.
Critically, note that we do not in any case *remove* the old license
notice or terms, as that remains necessary until we finish the
relicensing process.
I've updated a few license files that refer to the LLVM license to
instead simply refer generically to whatever license the LLVM project is
under, basically trying to minimize confusion.
This is really the culmination of so many people. Chris led the
community discussions, drafted the policy update and organized the
multi-year string of meeting between lawyers across the community to
figure out the strategy. Numerous lawyers at companies in the community
spent their time figuring out initial answers, and then the Foundation's
lawyer Heather Meeker has done *so* much to help refine and get us ready
here. I could keep going on, but I just want to make sure everyone
realizes what a huge community effort this has been from the begining.
Differential Revision: https://reviews.llvm.org/D56897
llvm-svn: 351631
GCC and Clang in MinGW mode don't support __declspec(thread), which
after expanding macros ends up as __attribute__((thread)). Use the
GCC specific attribute __thread instead (the next one in the chain
of alternatives).
Differential Revision: https://reviews.llvm.org/D56905
llvm-svn: 351587
Summary:
Adds the register class implementation for Sparc.
Adds support for DW_CFA_GNU_window_save.
Adds save and restore context functionality.
Adds getArch() function to each Registers_ class to be able to separate
between DW_CFA_AARCH64_negate_ra_state and DW_CFA_GNU_window_save which
are both represented by the same constant.
On Sparc the return address is the address of the call instruction, so
an offset needs to be added when returning to skip the call instruction
and its delay slot. If the function returns a struct it is also necessary
to skip one extra instruction on Sparc V8.
Reviewers: jyknight, mclow.lists, mstorsjo, compnerd
Reviewed By: jyknight, compnerd
Subscribers: jgorbe, mgorny, christof, llvm-commits, fedor.sergeev, JDevlieghere, ldionne, libcxx-commits
Differential Revision: https://reviews.llvm.org/D55763
llvm-svn: 351044
Summary:
Adds the register class implementation for Sparc.
Adds support for DW_CFA_GNU_window_save.
Adds save and restore context functionality.
On Sparc the return address is the address of the call instruction,
so an offset needs to be added when returning to skip the call instruction
and its delay slot. If the function returns a struct it is also necessary
to skip one extra instruction.
Reviewers: jyknight, mclow.lists, mstorsjo, compnerd
Reviewed By: compnerd
Subscribers: fedor.sergeev, JDevlieghere, ldionne, libcxx-commits
Differential Revision: https://reviews.llvm.org/D55763
llvm-svn: 350705
This doesn't yet implement inspecting the .pdata/.xdata to find the
LSDA pointer (in UnwindCursor::getInfoFromSEH), but normal C++
exception handling seems to run just fine without it. (The only
place I can see where it's even referenced is in
unwind_phase2_forced, and I can't find a codepath where libcxxabi
would end up calling that.)
Differential Revision: https://reviews.llvm.org/D55674
llvm-svn: 349532
- Support for the case where the return address has been signed with the B key
- When the B key is used, a 'B' character is present in the augmentation string
of CIE associated with the FDE for the function.
Differential Revision: https://reviews.llvm.org/D55704
llvm-svn: 349339
- Follow up to revision r342895
- gcc would not build libunwind with the earlier patch as the autia1716
instruction wasn't allowed to be assembled for pre armv8.3a targets
- The autia1716 instruction lives in the hint space encodings so is a valid
instruction for all armv8a targets
- To work around this I have swapped out the autia1716 instruction for the hint
instruction
Differential Revision: https://reviews.llvm.org/D55700
llvm-svn: 349140
Rename the 'append_if' macro used in libunwind to 'unwind_append_if'.
Otherwise, when used in a combined LLVM+libunwind build, it overrides
the *incompatible* 'append_if' function from LLVM and breaks projects
following libunwind, e.g. OpenMP.
Differential Revision: https://reviews.llvm.org/D55476
llvm-svn: 348852
This is needed when cross-compiling for a different target since
CFLAGS may contain additional flags like -resource-dir which
change the location in which compiler-rt builtins are found.
Differential Revision: https://reviews.llvm.org/D54371
llvm-svn: 346820
Summary:
If `-nodefaultlibs` is given, we weren't actually linking to it. This
was true irrespective of passing `-rtlib=compiler-rt` (see previous
patch). Now we explicitly link it to handle that case.
I wonder if we should be linking these libraries only if we're using
`-nodefaultlibs`...
Reviewers: beanz
Subscribers: dberris, mgorny, christof, chrib, cfe-commits
Differential Revision: https://reviews.llvm.org/D51657
llvm-svn: 343990
- When return address signing is enabled, the LR may be signed on function entry
- When an exception is thrown the return address is inspected used to unwind the call stack
- Before this happens, the return address must be correctly authenticated to avoid causing an abort by dereferencing the signed pointer
Differential Revision: https://reviews.llvm.org/D51432
llvm-svn: 342895
Summary:
This switch only has an effect at link time. It changes the default
compiler support library to `compiler-rt`. With `-nodefaultlibs`, this
library won't get linked anyway; Clang actually warns about that.
Reviewers: mstorsjo, rnk
Subscribers: dberris, mgorny, christof, cfe-commits
Differential Revision: https://reviews.llvm.org/D51645
llvm-svn: 341404
Summary:
This variable is never defined, so its value is always empty. Since
`libunwind` is needed to build the C++ ABI library in the first place,
it should never be linked to the C++ ABI library anyway.
Reviewers: mstorsjo, rnk
Subscribers: mgorny, christof, cfe-commits
Differential Revision: https://reviews.llvm.org/D51644
llvm-svn: 341388
Summary:
By default, symbols aren't visible outside of the module that defines
them. To make them visible, they must be exported. The easiest way to do
that is to embed an `-export:symname` directive into the object file.
Reviewers: mstorsjo, rnk
Subscribers: christof, cfe-commits
Differential Revision: https://reviews.llvm.org/D51508
llvm-svn: 341232
Even though SEH for ARM is incomplete, make what code already exists
at least compile correctly.
The _LIBUNWIND_CURSOR_SIZE wasn't correct.
ARM (and AArch64) have a DISPATCHER_CONTEXT field named TargetPc
instead of TargetIp.
For the libunwind.h UNW_* constants, there is no UNW_ARM_PC, only
UNW_ARM_IP.
Don't use 'r' as loop variable when 'r' already is a Registers_arm
member.
Differential Revision: https://reviews.llvm.org/D51530
llvm-svn: 341217
Summary:
That attribute has no effect on Windows anyway--classes are hidden by
default.
Reviewers: mstorsjo, rnk
Subscribers: christof, cfe-commits
Differential Revision: https://reviews.llvm.org/D51509
llvm-svn: 341210
Summary:
I've tested this implementation on x86-64 to ensure that it works. All
`libc++abi` tests pass, as do all `libc++` exception-related tests. ARM
still remains to be implemented (@compnerd?).
Special thanks to KJK::Hyperion for his excellent series of articles on
how EH works on x86-64 Windows. (Seriously, check it out. It's awesome.)
I'm actually not sure if this should go in as is. I particularly don't
like that I duplicated the UnwindCursor class for this special case.
Reviewers: mstorsjo, rnk, compnerd, smeenai, javed.absar
Subscribers: mgorny, kristof.beyls, christof, chrib, cfe-commits, compnerd, llvm-commits
Differential Revision: https://reviews.llvm.org/D50564
llvm-svn: 341125
Testing commit access from a new machine, so using this as an opportunity to revert my old test access commit (r336006) that I never cleaned up.
llvm-svn: 339899
When compiling with optimizations, mips requires various helper routines(__ashldi3 and the like) contained in libgcc_s.
Conditionally include libgcc_s in the set of libraries to be linked to.
Differential Revision: https://reviews.llvm.org/D50243
llvm-svn: 339878
Mipsr6 does not possess HI and LO accumulator registers, adjust validRegister functions to respect that.
Differential Revision: https://reviews.llvm.org/D50244
llvm-svn: 339849
The __mips_fpr macro can take the value of 0 as well, change to account for that case.
Differential Revision: https://reviews.llvm.org/D50245
llvm-svn: 339848
This is essential when building with -nodefaultlibs.
In some CMake versions (noticed in 3.5.1), the same libraries are
picked up from CMAKE_REQUIRED_LIBRARIES in some exceptional situations
(if CXX probing failed, due to libc++ not being built yet, the libraries
from CMAKE_REQUIRED_LIBRARIES are used for linking the target library),
but not at all in other newer CMake versions (3.10).
This is similar to what already is done in libcxxabi in SVN r302760
and libcxx in SVN r312498.
Differential Revision: https://reviews.llvm.org/D50663
llvm-svn: 339642
Summary:
Make the `_Unwind_Exception` struct correct under SEH. Add a
declaration of `_GCC_specific_handler()`, which is used by SEH versions
of Itanium personality handlers to do common setup. Roughly corresponds
to Clang's D50380.
Reviewers: mstorsjo, rnk, compnerd, smeenai
Subscribers: christof, chrib, cfe-commits, llvm-commits
Differential Revision: https://reviews.llvm.org/D50414
llvm-svn: 339258
Summary:
`long` is too short on LLP64. We have to use `intptr_t` to
avoid truncating pointers.
Reviewers: mstorsjo, rnk, compnerd, smeenai
Subscribers: christof, cfe-commits, llvm-commits
Differential Revision: https://reviews.llvm.org/D50412
llvm-svn: 339217
This should resolve the breakage introduced in r337867 which introduced
the use of cmake_dependent_option without include the necessary file.
llvm-svn: 337868
Currently it's only possible to control whether shared or static library
build of libc++, libc++abi and libunwind is enabled or disabled and
whether to install everything we've built or not. However, it'd be
useful to have more fine grained control, e.g. when static libraries are
merged together into libc++.a we don't need to install libc++abi.a and
libunwind.a. This change adds this option.
Differential Revision: https://reviews.llvm.org/D49573
llvm-svn: 337867
years. Adopt the new convention that it is call-site specific and that
it should be applied before moving the IP by personality routines, but
not during normal unwinding.
Differential Revision: https://reviews.llvm.org/D38680
llvm-svn: 337312
Do not use LLVM_RUNTIMES_LIBDIR_SUFFIX variable which is an internal
variable used by the runtimes build from individual runtimes, instead
set per-runtime librarhy directory suffix variable which is necessary
for the sanitized runtimes build to install libraries into correct
location.
Differential Revision: https://reviews.llvm.org/D49121
llvm-svn: 336713
This change adds a support for multiarch style runtimes layout, so in
addition to the existing layout where runtimes get installed to:
lib/clang/$version/lib/$os
Clang now allows runtimes to be installed to:
lib/clang/$version/$target/lib
This also includes libc++, libc++abi and libunwind; today those are
assumed to be in Clang library directory built for host, with the
new layout it is possible to install libc++, libc++abi and libunwind
into the runtime directory built for different targets.
The use of new layout is enabled by setting the
LLVM_ENABLE_RUNTIME_TARGET_DIR CMake variable and is supported by both
projects and runtimes layouts. The runtimes CMake build has been further
modified to use the new layout when building runtimes for multiple
targets.
Differential Revision: https://reviews.llvm.org/D45604
llvm-svn: 335809
The paths output from llvm-config --cmakedir and from clang
--print-libgcc-file-name can contain backslashes, while CMake
can't handle the paths in this form.
This matches what compiler-rt already does (since SVN r203789
and r293195).
Differential Revision: https://reviews.llvm.org/D48353
llvm-svn: 335169
This makes it possible to unwind hardware exception stack frames,
which necessarily save every register and so need an extra column
for storing the return address. CFI for the exception handler could
then look as follows:
.globl exception_vector
exception_vector:
.cfi_startproc
.cfi_signal_frame
.cfi_return_column 32
l.addi r1, r1, -0x100
.cfi_def_cfa_offset 0x100
l.sw 0x00(r1), r2
.cfi_offset 2, 0x00-0x100
l.sw 0x04(r1), r3
.cfi_offset 3, 0x04-0x100
l.sw 0x08(r1), r4
.cfi_offset 4, 0x08-0x100
l.mfspr r3, r0, SPR_EPCR_BASE
l.sw 0x78(r1), r3
.cfi_offset 32, 0x78-0x100
l.jal exception_handler
l.nop
l.lwz r2, 0x00(r1)
l.lwz r3, 0x04(r1)
l.lwz r4, 0x08(r1)
l.jr r9
l.nop
.cfi_endproc
This register could, of course, also be accessed by the trace
callback or personality function, if so desired.
llvm-svn: 332513
Before this commit, R9, the link register, was used as PC register.
However, a stack frame may have R9 not set to PC on entry, either
because it uses a custom calling convention, or, more likely,
because this is a signal or exception stack frame. Using R9 as
PC register made it impossible to unwind such frames.
All other architectures similarly use a dedicated PC register.
llvm-svn: 332512
Summary:
For MIPS ABIs with 64-bit floating point registers including newabi
and O32 with 64-bit floating point registers, just save and restore the
32 floating-point registers as doubles.
For O32 MIPS with 32-bit floating-point registers, save and restore the
individual floating-point registers as "plain" registers. These registers
are encoded as floats rather than doubles, but the DWARF unwinder
assumes that floating-point registers are stored as doubles when reading
them from memory (via AddressSpace::getDouble()). Treating the
registers as "normal" registers instead causes the DWARF unwinder to
fetch them from memory as a 32-bit register. This does mean that for
O32 with 32-bit floating-point registers unw_get_fpreg() and
unw_set_fpreg() do not work. One would have to use unw_get_reg()
and unw_set_reg() instead. However, DWARF unwinding works
correctly as the DWARF CFI emits records for individual 32-bit
floating-point registers even when they are treated as doubles stored
in paired registers. If the lack of unw_get/set_fpreg() becomes a pressing
need in the future for O32 MIPS we could add in special handling to
make it work.
Reviewers: sdardis, compnerd
Reviewed By: sdardis
Differential Revision: https://reviews.llvm.org/D41968
llvm-svn: 332414
CMAKE_CXX_FLAGS and CMAKE_C_FLAGS are added twice to the command line.
This causes the command line options to be doubled which works until
it doesn't as not all options can be specified twice.
For example,
clang-cl foo.c /GS- /GS- -mllvm -small-loop-cost=1 -mllvm -small-loop-cost=1
clang (LLVM option parsing): for the -small-loop-cost option: may only occur zero or one times!
llvm-svn: 329340
Summary:
N32 uses the same register context as N64. However, N32 requires one
change to properly fetch addresses from registers stored in memory.
Since N32 is an ILP32 platform, getP() only fetches the first 32-bits
of a stored register. For a big-endian platform this fetches the
upper 32-bits which will be zero. To fix this, add a new
getRegister() method to AddressSpace which is responsible for
extracting the address stored in a register in memory. This matches
getP() for all current ABIs except for N32 where it reads the 64-bit
register and returns the low 32-bits as an address. The
DwarfInstructions::getSavedRegister() method uses
AddressSpace::getRegister() instead of AddressSpace::getP().
Reviewers: sdardis, compnerd
Reviewed By: sdardis
Differential Revision: https://reviews.llvm.org/D39074
llvm-svn: 326250
Summary:
This is done via new LIBUNWIND_TEST_COMPILER_FLAGS and
LIBUNWIND_TEST_LINKER_FLAGS variables.
Reviewed By: sdardis
Differential Revision: https://reviews.llvm.org/D43585
llvm-svn: 326223
Otherwise, a shared library build with SJLJ APIs enabled would
end up with duplicate symbols.
This didn't occur for the apple && arm case due to specifically
checking for that in the surrounding ifdef.
Differential Revision: https://reviews.llvm.org/D42555
llvm-svn: 323499
Clang and llvm already use llvm_setup_rpath(), so this change will
help standarize rpath usage across all projects.
Differential Revision: https://reviews.llvm.org/D42461
llvm-svn: 323496
When CMAKE_SYSROOT or CMAKE_FIND_ROOT_PATH is set, cmake
recommends setting CMAKE_FIND_ROOT_PATH_MODE_INCLUDE=ONLY
globally which means find_path() always prepends CMAKE_SYSROOT or
CMAKE_FIND_ROOT_PATH to all paths used in the search.
However, this find_path() invocation is looking for a path in the
libcxx project on the host system, not the target system,
which can be done by passing NO_CMAKE_FIND_ROOT_PATH.
Differential Revision: https://reviews.llvm.org/D41621
llvm-svn: 323141
The Registers_ppc64 class needed a couple of changes, both to accommodate the
new registers as well as to handle the overlaps of VS register set
without wasting space.
The save/restore code of V and VS registers was added.
As VS registers depend on the VMX extension, they are processed only if
VMX support is detected (_ARCH_PWR8 for now).
Patch by Leandro Lupori!
Differential Revision: https://reviews.llvm.org/D41906
llvm-svn: 322596
This is in preparation for adding support for N32 unwinding which reuses
the newabi register class.
Reviewed By: compnerd
Differential Revision: https://reviews.llvm.org/D41842
llvm-svn: 322093
Initial working version of libunwind for PowerPC 64. Tested on
little-endian ppc64 host only.
Based on the existing PowerPC 32 code.
It supports:
- context save/restore (unw_getcontext, unw_init_local, unw_resume)
- read/write from/to saved registers
- backtrace (unw_step)
Patch by Leandro Lupori!
Differential Revision: https://reviews.llvm.org/D41386
Now builds with LIBUNWIND_ENABLE_CROSS_UNWINDING=ON should
work.
llvm-svn: 321680
For builds that only target one architecture, this was required to
be an exact match, while it previously required the allocation to be
strictly larger than the largest concrete one. Requiring it to be
larger than on equal should be enough.
This makes it more straightforward to update _LIBUNWIND_CONTEXT_SIZE
and _LIBUNWIND_CURSOR_SIZE.
llvm-svn: 321679
Initial working version of libunwind for PowerPC 64. Tested on
little-endian ppc64 host only.
Based on the existing PowerPC 32 code.
It supports:
- context save/restore (unw_getcontext, unw_init_local, unw_resume)
- read/write from/to saved registers
- backtrace (unw_step)
Patch by Leandro Lupori!
Differential Revision: https://reviews.llvm.org/D41386
llvm-svn: 321667
Instead, cast them to `void *` which has a portable format string syntax
of `%p`.
This fixes a -Wformat error when building libunwind.
llvm-svn: 321469
Right now, ARM EHABI unwind on bare metal expects to find the symbols
__exidx_start and __exidx_end defined, and uses those to locate
the EH tables. However, DWARF unwind on bare metal expects to find
dl_iterate_phdr, which, although possible to provide, is inconvenient
and mildly absurd.
This commit provides feature parity with ARM EHABI unwind by looking
for symbols __eh_frame_start, __eh_frame_end, __eh_frame_hdr_start
and __eh_frame_hdr_end, denoting the start and end of the sections
with corresponding names. As far as I know, there is no de jure or
de facto ABI providing any such names, so I chose the obvious ones.
The .eh_frame_hdr support is optional for maximum flexibility and
possible space savings (e.g. if libunwind is only used to provide
backtraces when a device crashes, providing the .eh_frame_hdr, which
is an index for rapid access to EH tables, would be a waste.)
The support for .eh_frame_hdr/DWARF index in the first place is
conditional on defined(_LIBUNWIND_SUPPORT_DWARF_INDEX), although
right now config.h will always define this macro.
The support for DWARF unwind on bare metal has been validated within
the ARTIQ environment[1].
[1]: https://m-labs.hk/artiq/
llvm-svn: 321445
Use the `_LIBUNWIND_LOG` and `_LIBUNWIND_LOG0` macros instead of
the explicit `fprintf` call.
This was previously done in r292721 as a cleanup and then reverted
in r293257 because the implementation in r292721 relied on a GNU
extension. This implementation avoids the use of an extension
by using a second macro instead, and allows to avoid the dependency
on fprintf if _LIBUNWIND_BARE_METAL is defined.
llvm-svn: 321441
This supports the soft-float ABI only and has been tested with both clang
and gcc on FreeBSD.
Reviewed By: sdardis, compnerd
Differential Revision: https://reviews.llvm.org/D38110
llvm-svn: 320528
This supports using a newer libunwind with an older installation of LLVM
(whose cmake modules wouldn't have add_llvm_install_targets).
llvm-svn: 320163
This follows the setup used by other runtimes and is expected by
the lit configuration.
Differential Revision: https://reviews.llvm.org/D40820
llvm-svn: 319830
This gains us the install-unwind-stripped target, to perform stripping
during installation.
Differential Revision: https://reviews.llvm.org/D40685
llvm-svn: 319498
This is useful in cases where we only build static library and
libunwind.a is combined with libc++abi.a into a single archive in which
case we don't want to have libunwind.a installed separately. The same
option is already provided by libcxx CMake build.
This change also adds the install-unwind target for consistency with the
libcxxabi and libcxx CMake build.
Differential Revision: https://reviews.llvm.org/D40195
llvm-svn: 318569
If the linker chose to store the full section name instead of truncating
it, this field doesn't contain a truncated name, but an offset into
the string table of the binary. The string table isn't loaded/mapped
into memory during runtime though, so it's not possible to read the
full section name, unless we try to locate the DLL/EXE on disk that
the HMODULE corresponds to and load that manually.
Due to this, lld now always prefers writing a truncated section name
for sections that will be mapped at runtime, even when debug info is
enabled.
Differential Revision: https://reviews.llvm.org/D39918
llvm-svn: 318446
This didn't require any further changes to libunwind as long as win64
in general is handled correctly.
Differential Revision: https://reviews.llvm.org/D39632
llvm-svn: 318383
The previous definition of _LIBUNWIND_HIGHEST_DWARF_REGISTER seems
to be a copy of the ARM64 value (introduced in SVN r276128); since
the code actually hasn't compiled properly for arm in dwarf mode
before, this hasn't actually been used. Set it to the correct value
based on the UNW_ARM_* enum values.
The iwmmx control variables have to be made mutable, since they are
touched from within getRegister (which previously wasn't const), and
getRegister is used on a const Registers object in DwarfInstructions.hpp.
Differential Revision: https://reviews.llvm.org/D39251
llvm-svn: 317192
This matches the original libunwind API. This also unifies the
type between ARM EHABI and the other configurations, and allows
getting rid of a number of casts in log messages.
The cursor size updates for ppc and or1k are untested, but
unw_proc_info_t shrinks by 4 uint64_t units on i386 at least.
Differential Revision: https://reviews.llvm.org/D39365
llvm-svn: 316942
This restores the previous behaviour of the Registers_* classes
after SVN r316745.
Differential Revision: https://reviews.llvm.org/D39382
llvm-svn: 316843
Clang doesn't currently support building for windows/x86_64 with
dwarf by setting command line parameters, but if manually modified
to use dwarf, we can make libunwind work in this configuration
as well.
Also include i386 in the docs when adding this as a supported
configuration; libunwind already works for i386 windows, but
can fail due to an issue unrelated to windows itself.
Differential Revision: https://reviews.llvm.org/D38819
llvm-svn: 316747
This avoids having to keep the same information duplicated in multiple
places.
Adjust _LIBUNWIND_HIGHEST_DWARF_REGISTER to actually have the value
of the highest used register and only use the value
_LIBUNWIND_HIGHEST_DWARF_REGISTER + 1 (kMaxRegisterNumber + 1) for
allocating the savedRegisters array.
Differential Revision: https://reviews.llvm.org/D39281
llvm-svn: 316745
For many targets setting -fno-exceptions will prevent unwinding tables from
being generated for the test programs. As libunwind depends on the tables to
unwind the stack several tests will fail.
This change always adds -funwind-tables so that even when -fno-exceptions
is set unwind tables are generated.
fixes PR33858
Differential Revision: https://reviews.llvm.org/D37484
llvm-svn: 316657
This was missed in SVN r274744 when the WMMX part was made optional;
when made optional, some struct fields were reordered, which caused
the total struct size to grow due to padding/alignment.
llvm-svn: 316559
Most other cases that touch savedRegisters[reg] have got this check,
but these three seemed to lack it.
Differential Revision: https://reviews.llvm.org/D39206
llvm-svn: 316415
This requires _WIN32_WINNT >= 0x0600.
If someone wants to spend effort on supporting earlier versions,
one can easily add another fallback implementation based on
critical sections, or try to load SRW lock functions dynamically.
This makes sure that the FDE cache is thread safe on windows.
Differential Revision: https://reviews.llvm.org/D38704
llvm-svn: 316364
This avoids a hack for making it a no-op for windows.
Also explicitly check for _WIN32 instead of assuming it.
Differential Revision: https://reviews.llvm.org/D39156
llvm-svn: 316300
If there is no binary search table computed, the FDECount encoding is
DW_EH_PE_omit. Do not attempt to read the FDECount in such a situation
as we will read an incorrect value. binutils only writes out the
FDECount if the encoding is not DW_EH_PE_omit.
llvm-svn: 316224
libunwind is known to work on FreeBSD i386, amd64 (x86_64) and arm64.
It is the unwinder provided by the base system on all of those
architectures.
While here sort the OS list.
Differential Revision: https://reviews.llvm.org/D38900
llvm-svn: 315814
In practice, with code built with clang, there are still unresolved
issues with DW_CFA_GNU_args_size though.
Differential Revision: https://reviews.llvm.org/D38679
llvm-svn: 315498
Rather than hardcode the flag and check if the compiler supports it, use
the CMake property to get the right flag spelling for the compiler.
This makes it generally more portable.
llvm-svn: 314834
That commit incorrectly expanded the assumption that defined(__APPLE__)
implies SjLj exception handling, which only is true within ARM
code sections.
llvm-svn: 314695
This moves the definition of the internal helpers
`__Unwind_SjLj_GetTopOfFunctionStack` and
`__Unwind_SjLj_SetTopOfFunctionStack` into `Unwind-sjlj.c`. These are
not extra functions specific to Apple, but rather are internal
implementation details of SjLj support in the LLVM libunwind
implementation.
This allows us to remove the internal header unwind_ext.h, as these
functions are not meant to be used as SPI either. Because they are
static, they will be given implicit hidden visibility, but due to the
simplicity should get fully inlined into the actual use.
Use the C11 standard static TLS annotation (`_Thread_local`) if
possible, otherwise, use the Windows specific `__declspec(thread)` when
targeting Windows or the GNU `__thread` extension. In theory, it should
be possible for this implementation to use a `pthread_setspecific` and
`pthread_getspecific` on platforms with pthreads or `SetFlsValue` and
`GetFlsValue` on Windows for non-static TLS. However, static TLS tends
to be significantly faster, so we really should prefer that over the
dynamic TLS approach. On Apple environments, when not building for the
loader (dyld), use the pre-allocated TLS slot in the loader rather than
the local static TLS variable.
Note that the un-threaded support of libunwind is still present as
before, however, it is unsafe to use in a threaded environment as the
cleanup stack may be mutated incorrectly due to lack of locking across
threads. In the static TLS model, the lock is unneeded as each thread
of execution retains its own copy of the cleanup stack.
Take the opportunity to clean up the comment block, removing the iOS
specific note as the SjLj implementation can be used outside of the
context of iOS. Convert the rest of the explanation to a doxygen style
comment block.
llvm-svn: 314632
When SJLJ exceptions are used, those functions aren't used.
This fixes build failures on ARM with SJLJ enabled (e.g. on armv7/iOS)
when built using the CMake project files.
Differential Revision: https://reviews.llvm.org/D38249
llvm-svn: 314197
This makes it match the definition used within llvm and in libgcc,
we previously got the wrong layout in 64 bit environments.
Differential Revision: https://reviews.llvm.org/D38247
llvm-svn: 314196
The changes in r297174 moved the #include of <link.h> on FreeBSD (and
probably other systems) inside of the open 'libunwind' namespace
causing various system-provided types such as pid_t to be declared in
this namespace rather than the global namespace. Fix this by moving
the relevant declarations before the 'libunwind' namespace is opened,
but still using the cleaned up declarations from r297174.
Reviewed By: ed, compnerd
Differential Revision: https://reviews.llvm.org/D38108
llvm-svn: 313920
The buildbots have shown that -Wstrict-prototypes behaves differently in GCC
and Clang so we should keep it disabled until Clang follows GCC's behaviour
llvm-svn: 312246
Clang 5 supports -Wstrict-prototypes. We should use it to catch any C
declarations that declare a non-prototype function.
rdar://33705313
Differential Revision: https://reviews.llvm.org/D36669
llvm-svn: 312240
It seems that GCC interprets `__attribute__((__aligned__))` as 8-byte
alignment on ARM, but clang does not. Explicitly specify the
double-word alignment value to ensure that the structure is properly
aligned.
llvm-svn: 311574
The C++ ABI requires that the exception object is double-word aligned.
The alignment attribute was applied to the `_Unwind_Exception` type
which is used on non-EHABI targets. On EHABI, the exception object type
is `_Unwind_Control_Block`. Apply the explicit maximal alignment on the
type to ensure that the allocation has the correct alignment.
Resolves PR33858!
llvm-svn: 311562
UnwindCursor<A, R>::getInfoFromEHABISection assumes the last
entry in the index table never corresponds to a real function.
Indeed, GNU ld always inserts an EXIDX_CANTUNWIND entry,
containing the end of the .text section. However, the EHABI specification
(http://infocenter.arm.com/help/topic/com.arm.doc.ihi0038b/IHI0038B_ehabi.pdf)
does not seem to contain text that requires the presence of a sentinel entry.
In that sense the libunwind implementation isn't compliant with the specification.
This patch makes getInfoFromEHABISection examine the last entry in the index
table if upper_bound returns the end iterator.
Fixes https://bugs.llvm.org/show_bug.cgi?id=31091
Differential revision: https://reviews.llvm.org/D35265
llvm-svn: 308871
This matches the behavior of libc++abi and libc++ and ensures that
we get a working toolchain when building libunwind as part of LLVM.
Differential Revision: https://reviews.llvm.org/D34375
llvm-svn: 308380
This is going to be used by the runtime build in the multi-target
setup to allow using different install prefix for each target.
Differential Revision: https://reviews.llvm.org/D33760
llvm-svn: 307606
Mostly cargo-culted from libcxxabi, since the unwinder was forked from there in
the first place. There may still be cruft that's only applicable to libcxxabi,
but that can be addressed in-tree.
https://reviews.llvm.org/D35038
llvm-svn: 307266
It's useful to be able to disable visibility annotations entirely; for
example, if we're building libunwind static to include in another library,
and we don't want any libunwind functions getting exported out of that
library.
https://reviews.llvm.org/D34637
Patch from Thomas Anderson <thomasanderson@chromium.org>!
llvm-svn: 306442
CMake has the problem with the single dash variant because of the
space, so use the double dash with equal sign version. The compile
flag handling had a typo which caused these flag not to be properly
include. We also don't have to pass the target triple when checking
for compiler-rt since that flag is already included in compile flags
now.
Differential Revision: https://reviews.llvm.org/D32071
llvm-svn: 300419
This is a reland of commit r299796.
Turned out that we need gcc_s or compiler-rt on ARM when checking
the support for -funwind-tables which creates a dependency on
__aeabi_unwind_cpp_pr0 symbol that's provided by the compiler
runtime.
Differential Revision: https://reviews.llvm.org/D31858
llvm-svn: 300020
Since libunwind is built with -nodefaultlibs, we should be using this
option even for CMake checks to avoid any inconsistency and also to
avoid dependency on a working C++ standard library just for the setting
up the build itself. The implementation is largely similar to the one
used by libc++.
Differential Revision: https://reviews.llvm.org/D31640
llvm-svn: 299796
On certain versions of android x86, the main module `app_process` is not
built as PIE. When accessing the PT_GNU_EH_FRAME_HDR in such a
scenario, the `dlpi_addr` is 0, but the virtual address is not
relocated. Manually rebase the address to avoid an invalid memory
access.
llvm-svn: 299575
In debug mode, we have assertions that the values do not exceed the
limits of the type holding them. In order to account for the type being
derived from the AddressSpace and thus a typedef, we use
`std::numeric_limits`. Include the appropriate header.
Thanks to Marshal Clow for pointing out the missing include!
llvm-svn: 297744
The AddressSpace.hpp header declares two classes: LocalAddressSpace and
RemoteAddressSpace. These classes are only used in a very small number
of source files, but passed in as template arguments to many other
classes.
Let's go ahead and only include AddressSpace.hpp in source files where
at least one of these two classes is mentioned. This gets rid of a
cyclic header dependency that was already present, but only caused
breakage on macOS until recently.
Reported by: Marshall Clow
llvm-svn: 297364
Other source files in the source tree tend to include this header file
unconditionally. It also parses perfectly fine on ARM EHABI systems.
llvm-svn: 297175
All of the access to __exidx_*, dl_iterate_phdr(), etc. is specific to
the findUnwindSections() function. Right now all of the includes and
declarations related to them are scattered throughout the source file.
For example, for <link.h>, we have a full list of operating systems
guarding the #include, even though the code that uses dl_iterate_phdr()
miraculously doesn't use the same list.
Change the code so that findUnwindSections() is preceded by a block of
#ifdefs that share the same structure as the function itself. First
comes all of the macOS specific bits, followed by bare-metal ARM,
followed by ELF EHABI + DWARF.
This actually allows us to build a copy of libunwind without any
specific ifdefs for NetBSD, CloudABI, etc. It likely also unbreaks the
build of libunwind on FreeBSD/armv6, though I can't confirm.
Reviewed by: compnerd
Differential Revision: https://reviews.llvm.org/D30696
llvm-svn: 297174
Exception section data that we extract for DWARF gets stored as the
offset and the number of bytes. For ARM exception info, we seem to
deviate from this by storing the number of entries. Attempt to make this
more consistent.
By storing the number of bytes, we can get rid of the EHTEntry structure
declared in AddressSpace.hpp. In UnwindCursor.hpp we already have
another structure declared for the same purpose.
Reviewed by: Keith Walker
Differential Revision: https://reviews.llvm.org/D30681
llvm-svn: 297149
While porting libunwind over to CloudABI for ARMv6, I observed that this
source file doesn't build, as it depends on dl_unwind_find_exidx(),
which CloudABI's C library was lacking. After I added that function, I
still needed to patch up libunwind to define _Unwind_Ptr.
Taking a step back, I wonder why we need to make use of this function
anyway. The unwinder already has some nice code to use dl_iterate_phdr()
to scan for a PT_GNU_EH_FRAME header. The dl_unwind_find_exidx() does
the same thing, except matching PT_ARM_EXIDX instead. We could also do
that ourselves.
This change gets rid of the dl_unwind_find_exidx() call and extends the
dl_iterate_phdr() loop. This approach has the advantage of getting rid
of some of those OS-specific #ifdefs. This now means that if an
operating system only provides dl_iterate_phdr(), it gets support for
unwinding on all architectures. There is no need to add more stuff, just
to get ARMv6 support.
This change is identical to r295944, except that it now adds the
necessary code to do bounds checking on PT_LOAD. The previous version of
this change lacked this, which didn't cause any problems on CloudABI,
but did break the Linux build bots. This is why I reverted it in
r295948.
Differential Revision: https://reviews.llvm.org/D30306
llvm-svn: 296991
When libunwind was spinned off libcxxabi, most file were copied from
libcxxabi to libunwind. However, libc++abi's toplevel LICENSE.TXT was
forgotten in the copying. It's considered a good practice to have the
license file at the root of the project, and making linunwind a separate
project was not supposed to change its licensing. Besides, several
header files refer to the LICENSE.TXT, so copy the one from libc++abi.
llvm-svn: 296358
We've been having issues with using libcxxabi and libunwind for baremetal
targets because fprintf is dependent on io functions, this patch disables calls
to fprintf when building for baremetal in release mode.
Differential Revision: https://reviews.llvm.org/D30340
llvm-svn: 296135
Even though the change works perfectly fine on CloudABI, it fails to
work on the libcxx-libcxxabi-libunwind-arm-linux-noexceptions build bot.
Looking at the code, this may be attributed to the fact that the code
doesn't take the PT_LOAD addresses into consideration.
I will rework this change to fix that and send out an updated version
for review in the nearby future.
llvm-svn: 295948
While porting libunwind over to CloudABI for ARMv6, I observed that this
source file doesn't build, as it depends on dl_unwind_find_exidx(),
which CloudABI's C library was lacking. After I added that function,
I still needed to patch up libunwind to define _Unwind_Ptr.
Taking a step back, I wonder why we need to make use of this function
anyway. The unwinder already has some nice code to use dl_iterate_phdr()
to scan for a PT_GNU_EH_FRAME header. The dl_unwind_find_exidx() does
the same thing, except matching PT_ARM_EXIDX instead. We could also do
that ourselves.
This change gets rid of the dl_unwind_find_exidx() call and extends the
dl_iterate_phdr() loop. In addition to making the code a bit shorter, it
has the advantage of getting rid of some of those OS-specific #ifdefs.
This now means that if an operating system only provides
dl_iterate_phdr(), it gets support for unwinding on all architectures.
There is no need to add more stuff, just to get ARMv6 support.
Differential Revision: https://reviews.llvm.org/D28082
llvm-svn: 295944
libunwind depends on C++ library headers. When building libunwind
as part of LLVM and libc++ is available, use its headers.
Differential Revision: https://reviews.llvm.org/D29997
llvm-svn: 295285
This causing build failure on sanitizer bots because of the unused
argument '-nostdinc++' during linking of libunwind.
This reverts commit 0e14fd1a1d37b9c6d55a2d3bc7649e5b39ce74d3.
llvm-svn: 295202
libunwind depends on C++ library headers. When building libunwind
as part of LLVM and libc++ is available, use its headers.
Differential Revision: https://reviews.llvm.org/D29800
llvm-svn: 295153
libunwind depends on C++ library headers. When building libunwind
as part of LLVM and libc++ is available, use its headers.
Differential Revision: https://reviews.llvm.org/D29573
llvm-svn: 294554
Sterling Augustine