| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
| |
union containing either a member named __cc, or either of __cc and __nc (const vs. non-const). This level of wrapping is quite useless for LLDB to show to people, so try to detect it, and filter it out
llvm-svn: 217651
|
| |
|
|
|
|
|
|
| |
ProcessStructReader instead of reading the structure out of libdispatch
by hand. I will convert more of the struct readers in SystemRuntimeMacOSX
over to use this class eventually.
llvm-svn: 217649
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* Sends a SIGSTOP to the process.
* Fixes busted SIGSTOP handling. Now builds a list of non-stopped
that we wait for the PTRACE group-stop for. When the final must-stop
tid gets its group stop, we propagate the process state change.
Only the signal receiving the notification of the pending SIGSTOP
is marked with the SIGSTOP signal. All the rest, if they weren't
already stopped, are marked as stopped with signal 0.
* Fixes a few broken tests.
* Marks the Linux test I added earlier as expect-pass (no longer XFAIL).
Implements fix for http://llvm.org/bugs/show_bug.cgi?id=20908.
llvm-svn: 217647
|
| |
|
|
|
|
|
|
|
|
| |
The purpose of a ProcessStructReader is to allow intelligent reading of data from the underlying process
Traditionally, the way this has been handled is to have a load_address and shuffle it around, and use Process::ReadMemory()-kind APIs
With a ProcessStructReader one can define a clang type that matches exactly the definition of the thing they are trying to ingest from the inferior process, and then just point LLDB at where the data is
Since this work is done in terms of clang types, one can honor packed/aligned attributes, sizes of types on the inferior architecture, and similar tricky caveats without having to hardcode them
llvm-svn: 217644
|
| |
|
|
|
|
| |
summary is a good candidate to ask. While in theory one could want one-liner printing with a non-one-liner summary, I don't see LLDB as the best place to solve such inner conflicts
llvm-svn: 217641
|
| |
|
|
| |
llvm-svn: 217634
|
| |
|
|
| |
llvm-svn: 217632
|
| |
|
|
|
|
|
|
| |
* ssize_t isn't defined by default on Windows.
* New public API files need to be defined in a different file for
Windows.
llvm-svn: 217624
|
| |
|
|
|
|
|
|
|
|
| |
app is interrupted.
The Linux version is marked XFAIL for the moment, fixing next.
Related to http://llvm.org/bugs/show_bug.cgi?id=20908.
llvm-svn: 217613
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
1. DW_FORM_strp and DW_FORM_sec_offset are 64bits for DWARF64 / 32bits for DWARF32
They are different from DW_FORM_addr, whose size is specified in .debug_info
2. Bump DWARF version support form [2,3] to [2,4] in DWARFDebugLine.cpp
3. Fix DWARFDebugLine to support DWARF64
See http://reviews.llvm.org/D5307 for more details.
Reviewed by Greg Clayton and Jason Molenda.
Change by Tong Shen.
llvm-svn: 217607
|
| |
|
|
|
|
|
|
| |
locations with new ones if possible.
Next up some test cases...
llvm-svn: 217551
|
| |
|
|
|
|
|
|
|
|
|
| |
This change implements this ticket:
http://llvm.org/bugs/show_bug.cgi?id=20899
Adds the qThreadStopInfo RSP command for llgs and includes a test that
verifies both debugserver and llgs respond with something reasonable
on a multithreaded app.
llvm-svn: 217549
|
| |
|
|
|
|
|
|
| |
SetName is only used in LLDB to set a thead's own name. Move it there
to match OS X and Windows and slightly reduce the effort in any future
HostThread/ThisThread name refactoring.
llvm-svn: 217521
|
| |
|
|
|
|
|
| |
More work on the GetName/SetName arguments (thread_t vs tid_t) is needed
but this change should restore the build and basic operation.
llvm-svn: 217502
|
| |
|
|
| |
llvm-svn: 217464
|
| |
|
|
| |
llvm-svn: 217462
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
This patch moves creates a thread abstraction that represents a
thread running inside the LLDB process. This is a replacement for
otherwise using lldb::thread_t, and provides a platform agnostic
interface to managing these threads.
Differential Revision: http://reviews.llvm.org/D5198
Reviewed by: Jim Ingham
llvm-svn: 217460
|
| |
|
|
|
|
| |
variable.
llvm-svn: 217456
|
| |
|
|
|
|
| |
when compiling optimized.
llvm-svn: 217429
|
| |
|
|
|
|
| |
with -Wunused-function as far as I can tell.
llvm-svn: 217428
|
| |
|
|
| |
llvm-svn: 217426
|
| |
|
|
| |
llvm-svn: 217425
|
| |
|
|
| |
llvm-svn: 217424
|
| |
|
|
| |
llvm-svn: 217423
|
| |
|
|
| |
llvm-svn: 217422
|
| |
|
|
| |
llvm-svn: 217421
|
| |
|
|
| |
llvm-svn: 217420
|
| |
|
|
| |
llvm-svn: 217419
|
| |
|
|
|
|
|
| |
as part of an attempt to get lldb building inside llvm with -Werror
enabled.
llvm-svn: 217418
|
| |
|
|
| |
llvm-svn: 217415
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
This makes sure that nothing that requires Python is being built
when the LLDB_DISABLE_PYTHON flag is being passed in.
It also changes a use of CPPFLAGS to CPP.Flags since the former is overridden
when external flags are passed in while the later is not. I'm not sure exactly
why LLDB_DISABLE_PYTHON is in CXXFLAGS rather than CPPFLAGS,
but cleaning that up is for another commit.
Differential Revision: http://reviews.llvm.org/D4918
llvm-svn: 217414
|
| |
|
|
|
|
|
| |
The last of the missing symbols to correct the make based build to restore the
FreeBSD buildbot!
llvm-svn: 217394
|
| |
|
|
|
|
|
|
| |
test/functionalities/connect_remote/TestConnectRemote.py when port 12345 is already in use
Reviewed at http://reviews.llvm.org/D5221
llvm-svn: 217388
|
| |
|
|
|
|
| |
during test run.
llvm-svn: 217387
|
| |
|
|
|
|
|
|
| |
See http://reviews.llvm.org/D5135 for more details.
Change by Zephyr Zhao.
llvm-svn: 217382
|
| |
|
|
|
|
|
|
|
| |
This cleans up a couple of warnings [-Wcovered-switch-default] from the build by
removing the default case from a couple of switches which are fully covered.
This is generally better as it will help identify when a new item is added to
the enumeration but the use sites are not updated.
llvm-svn: 217376
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
This adds a definition for the TypeValidatorImpl_CXX destructor. Because the
destructor is first virtual method, and declared out-of-line, it also serves as
the key function. Since no definition was present, no virtual table for
TypeValidatorImpl_CXX was emitted, which results in link failures due to
references to undefined symbols.
Also add a definition for a TypeValidatorImpl contructor which was declared
out-of-line and referenced in a constructor for TypeValidatorImpl_CXX.
llvm-svn: 217375
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Linking Release+Asserts executable lldb-gdbserver (without symbols)
liblldb.so: undefined reference to `lldb_private::MemoryHistoryASan::Initialize()'
liblldb.so: undefined reference to `lldb_private::MemoryHistoryASan::Terminate()'
liblldb.so: undefined reference to `vtable for lldb_private::TypeValidatorImpl_CXX'
liblldb.so: undefined reference to `lldb_private::TypeValidatorImpl::TypeValidatorImpl(lldb_private::TypeValidatorImpl::Flags const&)'
liblldb.so underlinked to lldbPluginMemoryHistoryASan.a when building with the
make based build system (as opposed to CMake).
llvm-svn: 217360
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We would previously simply assume that the write would always succeed. However,
write(2) may return -1 for error as well as fail to perform a complete write (in
which case the returned number of bytes will be less than the requested bytes).
Explicitly check if an error condition is encountered. This would previously
not be caught as we default initialized success to true. Add an assertion that
we always perform a complete write (a continuous retry could be added to ensure
that we finish writing completely).
This was caught by GCC's signed comparison warning and manual inspection.
llvm-svn: 217355
|
| |
|
|
|
|
|
|
|
|
| |
'#import' is an Objective-C construct; avoid using it in C++. NFC.
Addresses PR20867.
Patch by Kevin Avila!
llvm-svn: 217340
|
| |
|
|
|
|
| |
Patch by Remco Verhoef.
llvm-svn: 217312
|
| |
|
|
|
|
| |
type validators. The jury is still out on what the user experience of type validators should be, so for now gate it on a specific flag. The mode I am using is prefix variables that fail to validate with a bang, and then emitting the actual validation error on a separate line. Of course, given the total absence of validators, this should never actually happen to you
llvm-svn: 217303
|
| |
|
|
|
|
| |
PR20864
llvm-svn: 217301
|
| |
|
|
|
|
|
|
|
| |
Reviewed at
http://reviews.llvm.org/D5219
and
http://lists.cs.uiuc.edu/pipermail/lldb-commits/Week-of-Mon-20140901/012809.html
llvm-svn: 217300
|
| |
|
|
|
|
| |
API. To keep it simple, do not expose the pair, but just return a NULL string for success, and a non-NULL string for error; If we were to decide to expose the pair, we would need an SBTypeValidatorResult, which is fine, but it should come as part of exposing type validators through the SB API rather than as a one-off thing. So, KISS for now
llvm-svn: 217299
|
| |
|
|
| |
llvm-svn: 217297
|
| |
|
|
|
|
|
|
|
| |
Reviewed at
http://reviews.llvm.org/D5218
and
http://lists.cs.uiuc.edu/pipermail/lldb-commits/Week-of-Mon-20140901/012828.html
llvm-svn: 217296
|
| |
|
|
|
|
| |
address)
llvm-svn: 217287
|
| |
|
|
|
|
|
|
|
| |
handling.
Covers more of the behavior of rendezvous breakpoint handling and other
dynamic loader aspects, all on the 'enable log lldb dyld' log channel.
llvm-svn: 217283
|
| |
|
|
|
|
| |
ValueObject to have a validator, to update it from the FormatManager, and to retrieve (and cache) the result of the validation
llvm-svn: 217282
|