Before, in the absence of any configured REPLs, LLDB would act as if there were
multiple possible REPL options, whereas actually no REPL language is supported.
Now we make a better error.
llvm-svn: 250931
Summary:
Along with this, support for an optional argument to the "num_children"
method of a Python synthetic child provider has also been added. These have
been added with the following use case in mind:
Synthetic child providers currently have a method "has_children" and
"num_children". While the former is good enough to know if there are
children, it does not give any insight into how many children there are.
Though the latter serves this purpose, calculating the number for children
of a data structure could be an O(N) operation if the data structure has N
children. The new method added in this change provide a middle ground.
One can call GetNumChildren(K) to know if a child exists at an index K
which can be as large as the callers tolerance can be. If the caller wants
to know about children beyond K, it can make an other call with 2K. If the
synthetic child provider maintains state about it counting till K
previosly, then the next call is only an O(K) operation. Infact, all
calls made progressively with steps of K will be O(K) operations.
Reviewers: vharron, clayborg, granata.enrico
Subscribers: labath, lldb-commits
Differential Revision: http://reviews.llvm.org/D13778
llvm-svn: 250930
Summary:
Loop detection code is being called before every element access. Although it tries to cache some
of the data by remembering the loop-free initial segment, every time it needs to increase this
segment, it will start from scratch. For the typical usage pattern, where one accesses the
elements in order, the loop detection will need to be run after every access, resulting in
quadratic behavior. This behavior is noticable even for the default 255 element limit.
In this commit, I rewrite the algorithm to be truly incremental -- it maintains the state of its
loop-detection runners between calls, and reuses them when it needs to check another segment.
This way, each part of the list is scanned only once, resulting in linear behavior.
Also note that I have changed the operator== of ListEntry to do the comparison based on the
value() function (instead of relying on ValueObjectSP equality). In my experiments, I kept
getting different ValueObjectSPs when going through the same element twice.
Reviewers: granata.enrico
Subscribers: lldb-commits, sivachandra
Differential Revision: http://reviews.llvm.org/D13902
llvm-svn: 250890
Patch adds command 'language renderscript allocation save' to store the contents of an allocation in a binary file.
And 'language renderscript allocation load' to restore an allocation with the saved data from a binary file.
Binary file format contains a header FileHeader with meta information preceding the raw data.
Reviewed by: jingham, clayborg
Subscribers: lldb-commits, domipheus
Differential Revision: http://reviews.llvm.org/D13903
llvm-svn: 250886
To allow that, I've added a SetREPL call to the Target, which allows a REPL
that just created a target to install itself as the go-to REPL for the
corresponding language.
llvm-svn: 250870
Revert it bacuse it introduces several race condition detected by
the build bots.
This reverts commit 5107a5ebdb7c4571a30a7098b40bf8098b678447.
llvm-svn: 250832
Loading the debug info from a large application is the slowest task
LLDB do. This CL makes most of the dwarf parsing code multi-threaded.
As a result the speed of "attach; backtrace; exit;" when the inferior
is an LLDB with full debug info increased by a factor of 2 (on my machine).
Differential revision: http://reviews.llvm.org/D13662
llvm-svn: 250821
The purpose of the class is to make it easy to execute tasks in parallel
Basic design goals:
* Have a very lightweight and easy to use interface where a list of
lambdas can be executed in parallel
* Use a global thread pool to limit the number of threads used
(std::async don't do it on Linux) and to eliminate the thread creation
overhead
* Destroy the thread currently not in use to avoid the confusion caused
by them during debugging LLDB
Possible future improvements:
* Possibility to cancel already added, but not yet started tasks
* Parallel for_each implementation
* Optimizations in the thread creation destroyation code
Differential revision: http://reviews.llvm.org/D13727
llvm-svn: 250820
Summary:
ADB packets have a maximum size of 4k. This means the size of memory reads does not affect speed
too much (as long as it fits in one packet). Therefore, I am increasing the default memory read
size for android to 2k. This value is used only if the user has not modified the default
memory-cache-line-size setting.
Reviewers: clayborg, tberghammer
Subscribers: tberghammer, danalbert, srhines, lldb-commits
Differential Revision: http://reviews.llvm.org/D13812
llvm-svn: 250814
"expr -r" does this. It also returns to a REPL if the LLDB command interpreter
is neseted inside it, for example in cases where a REPL command resulted in a
breakpoint being hit or a crash.
llvm-svn: 250780
This makes LLDB launch and create a REPL, specifying no target so that the REPL
can create one for itself. Also added the "--repl-language" option, which
specifies the language to use. Plumbed the relevant arguments and errors
through the REPL creation mechanism.
llvm-svn: 250773
m_directory = "/tmp"
m_filename = "."
To look like:
m_directory = "/tmp/."
m_filename = "foo.txt"
if "foo.txt" was appended to it. With this fix it will be:
m_directory = "/tmp"
m_filename = "foo.txt"
llvm-svn: 250770
A REPL takes over the command line and typically treats input as source code.
REPLs can also do code completion. The REPL class allows its subclasses to
implement the language-specific functionality without having to know about the
IOHandler-specific internals.
Also added a PluginManager-based way of getting to a REPL given a language and
a target.
Also brought in some utility code and expression options that are useful for
REPLs, such as line offsets for expressions, ANSI terminal coloring of errors,
and a few IOHandler convenience functions.
llvm-svn: 250753
This allows open source MacOSX clients to not have to build debugserver and the current LLDB can find debugserver inside the selected Xcode.app on your system.
<rdar://problem/23167253>
llvm-svn: 250735
This patch corrects the number of bytes of debug register resources which are written while installing or removing a breakpoint using ptrace interface on arm64 targets.
Differential revision: http://reviews.llvm.org/D12522
llvm-svn: 250700
For o32 applications on mips we were getting segmentation fault while launching lldb-server because of overwritting stack when using elf_gregset_t in DoWriteRegisterValue.
We are now using the GPR_mips_linux buffer in DoWriteRegisterValue as done in DoReadRegisterValue also, which solves the above issue.
llvm-svn: 250696
There were a number of const qualifiers being cast away which caused warnings.
This cluttered the output hiding real errors. Silence them by explicit casting.
NFC.
llvm-svn: 250662
ValueObjectPrinter can now mask out pointer values during a printout; also, it supports helper functions to print declarations in different formats if needed
Practically speaking however, this change is NFC as nothing yet uses it in the codebase
llvm-svn: 250599
Using the Python native C API is non-portable across Python versions,
so this patch changes them to use the `PythonFile` class which hides
the version specific differences behind a single interface.
llvm-svn: 250525
Summary:
On linux, the environment variables for temp directories that lldb checks for are generally not
defined, and the temp directory computation failed. This caused expression evaluation to fall
back to creating "/tmp/lldb-*.expr" debugging files instead of the usual
"$TMP/lldb/pid/lldb-*.expr". Crucially, these files were not cleaned up on lldb exit, which
caused clutter in the /tmp folder, especially on long-running machines (e.g. builtbots). This
commit fixes lldb to use llvm::sys::path::system_temp_directory, which does the same environment
variable dance, but (!) also falls back to the P_tmpdir macro, which is how the temp directory is
defined on linux.
Since the linux temp path computation now succeeds, I needed to also modify Android path
computation to check for actual directory existence, rather then checking whether the operation
failed.
Reviewers: clayborg, tberghammer
Subscribers: tberghammer, lldb-commits, danalbert, srhines, emaste
Differential Revision: http://reviews.llvm.org/D13772
llvm-svn: 250502
Previous commit r250281 broke TestDataFormatterSmartArray.py
Resolved in in this patch by adding the new enum eFormatVectorOfFloat16 to FormatManager.
Differential Revision: http://reviews.llvm.org/D13730
llvm-svn: 250499
Summary: r249597 introduced a usage of GetTypeSummary in lldb-mi.
That function used to only be available when python is enabled.
However, there is no reason for that anymore since that is now
dealt with at a different abstraction layer.
Reviewers: ki.stfu, evgeny777, clayborg, granata.enrico
Subscribers: elehcim, brucem, lldb-commits
Differential Revision: http://reviews.llvm.org/D13577
llvm-svn: 250494
Summary: This function reads m_frames, so it needs to lock the
mutex like all other function that access m_frames.
Found using AddressSanitizer.
Reviewers: jingham, clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13638
llvm-svn: 250493
Python file handling got an overhaul in Python 3, and it affects
the way we have to interact with files. Notably:
1) `PyFile_FromFile` no longer exists, and instead we have to use
`PyFile_FromFd`. This means having a way to get an fd from
a FILE*. For this we reuse the lldb_private::File class to
convert between FILE*s and fds, since there are some subtleties
regarding ownership rules when FILE*s and fds refer to the same
file.
2) PyFile is no longer a builtin type, so there is no such thing as
`PyFile_Check`. Instead, files in Python 3 are just instances
of `io.IOBase`. So the logic for checking if something is a file
in Python 3 is to check if it is a subclass of that module.
Additionally, some unit tests are added to verify that `PythonFile`
works as expected on Python 2 and Python 3, and
`ScriptInterpreterPython` is updated to use `PythonFile` instead of
manual calls to the various `PyFile_XXX` methods.
llvm-svn: 250444
On linux, the environment variables for temp directories that lldb checks for are generally not
defined, and the temp directory computation failed. This caused expression evaluation to fall
back to creating "/tmp/lldb-*.expr" debugging files instead of the usual
"$TMP/lldb/pid/lldb-*.expr". Crucially, these files were not cleaned up on lldb exit, which
caused clutter in the /tmp folder, especially on long-running machines (e.g. builtbots). This
commit fixes lldb to use llvm::sys::path::system_temp_directory, which does the same environment
variable dance, but (!) also falls back to the P_tmpdir macro, which is how the temp directory is
defined on linux.
Since the linux temp path computation now succeeds, I needed to also modify Android path
computation to check for actual directory existence, rather then checking whether the operation
failed.
llvm-svn: 250409
disabled the use of the jThreadGetExtendedInfo packet which is used
to retrieve additional information about a thread, such as the QoS
setting for that thread on darwin systems.
Re-enable the use of the jThreadGetExtendedInfo packet, and add
some quick tests to the TestQueues mac test case which will verify
that we can retrieve the QoS names for these test threads.
<rdar://problem/22925096>
llvm-svn: 250364
Most platforms have "/dev/null". Windows has "nul". Instead of
hardcoding the string /dev/null at various places, make a constant
that contains the correct value depending on the platform, and use
that everywhere instead.
llvm-svn: 250331
There were a couple of issues related to string handling that
needed to be fixed. In particular, we cannot get away with
converting `PyUnicode` objects to `PyBytes` objects and storing
the `PyBytes` regardless of Python version. Instead we have to
store a `PyUnicode` on Python 3 and a `PyString` on Python 2.
The reason for this is that if you call `PyObject_Str` on a
`PyBytes` in Python 3, it will return you a string that actually
contains the string value wrappedin the characters b''. So if we
create a `PythonString` with the value "test", and we call Str()
on it, we will get back the string "b'test'", which breaks string
equality. The only way to fix this is to store a native
`PyUnicode` object under Python 3.
With this CL, ScriptInterpreterPythonTests unit tests pass 100%
under Python 2 and Python 3.
llvm-svn: 250327
Python 3 reverses the order in which you must call Py_InitializeEx
and PyEval_InitThreads. Since that log is in itself already a
little nuanced, it is refactored into a function so that the reversal
is more clear. At the same time, there's a lot of logic during
Python initialization to save off a bunch of state and then restore
it after initialization is complete. To express this more cleanly,
it is refactored to an RAII-style pattern where state is saved off
on acquisition and restored on release.
llvm-svn: 250306
Added a constructor that takes list_size for `PythonList`.
Made all single-argument constructors explicit.
Re-ordered constructors to be consistent with other classes.
llvm-svn: 250304
Adding the following flag to a cmake line:
-DLLDB_EXPORT_ALL_SYMBOLS=TRUE
will cause all symbols to be exported from liblldb. This enables the llvm
backtrace mechanism to see and report backtrace symbols properly when using
(lldb) log enable --stack ...
Prior to this change, only the SB API symbols would show up on Linux and other
systems that use a public-symbols-based backtrace lookup mechanism.
log enable --stack ... is a very handy, quick way to understand the flow
of how some log lines are getting hit within lldb without having to hook
up a top-level debugger over your current debug session.
llvm-svn: 250299
Summary:
This commit adds support for binary memory reads ($x) to lldb-server. It also removes the "0x"
prefix from the $x client packet, to make it more compatible with the old $m packet. This allows
us to use almost the same code for handling both packet types. I have verified that debugserver
correctly handles $x packets even without the leading "0x". I have added a test which verifies
that the stub returns the same memory contents for both kinds of memory reads ($x and $m).
Reviewers: tberghammer, jasonmolenda
Subscribers: iancottrell, lldb-commits
Differential Revision: http://reviews.llvm.org/D13695
llvm-svn: 250295
Previously ConstString had a single mutex guarding the global string
pool for each access what become a bottleneck when using it with a
large number of threads.
This CL distributes the strings to 256 individual string pools based on
a simple hash function to eliminate the bottleneck and speed up the
multi-thread access.
The goal of the change is to prepare to multi-threaded symbol parsing code
to speed up the symbol parsing speed.
Differential revision: http://reviews.llvm.org/D13652
llvm-svn: 250289
It is a new function added to the llvm::Triple class to simplify the
checking if we are targeting android to clean up the confusion between
android being an OS or an environment.
llvm-svn: 250286
This patch adds the command 'language renderscript allocation dump <ID>' for printing the contents of a RS allocation.
Displaying the coordinate of each element as well as its formatted value
e.g (lldb) language renderscript allocation dump 1
Data (X, Y, Z):
(0, 0, 0) = {0 1}
(1, 0, 0) = {2 3}
(2, 0, 0) = {4 5}
A --file <filename> option is also included, since for large allocations it may be more helpful to view this text as a file.
Reviewed by: jingham, clayborg
Subscribers: lldb-commits, ADodds, domipheus, brucem
Differential Revision: http://reviews.llvm.org/D13699
llvm-svn: 250281
* ArchSpec::MergeFrom() would erroneously promote an unspecified
unknown to a specified unknown when both the ArchSpec and the merged
in ArchSpec were both unspecified unknowns. This no longer happens,
which fixes issues with global module cache lookup in some
situations.
* Added ArchSpec::DumpTriple(Stream&) that now properly prints
unspecified unknowns as '*' and specified unknows as 'unknown'.
This makes it trivial to tell the difference between the two.
Converted printing code over ot using DumpTriple() rather than
building from scratch.
* Fixed up a couple places that were not guaranteeing that an
unspecified unknown was recorded as such.
llvm-svn: 250253
PythonObjects were being incorrectly ref-counted. This problem was
pervasive throughout the codebase, leading to an unknown number of memory
leaks and potentially use-after-free.
The issue stems from the fact that Python native methods can either return
"borrowed" references or "owned" references. For the former category, you
*must* incref it prior to decrefing it. And for the latter category, you
should not incref it before decrefing it. This is mostly an issue when a
Python C API method returns a `PyObject` to you, but it can also happen with
a method accepts a `PyObject`. Notably, this happens in `PyList_SetItem`,
which is documented to "steal" the reference that you give it. So if you
pass something to `PyList_SetItem`, you cannot hold onto it unless you
incref it first. But since this is one of only two exceptions in the
entire API, it's confusing and difficult to remember.
Our `PythonObject` class was indiscriminantely increfing every object it
received, which means that if you passed it an owned reference, you now
have a dangling reference since owned references should not be increfed.
We were doing this in quite a few places.
There was also a fair amount of manual increfing and decrefing prevalent
throughout the codebase, which is easy to get wrong.
This patch solves the problem by making any construction of a
`PythonObject` from a `PyObject` take a flag which indicates whether it is
an owned reference or a borrowed reference. There is no way to construct a
`PythonObject` without this flag, and it does not offer a default value,
forcing the user to make an explicit decision every time.
All manual uses of `PyObject` have been cleaned up throughout the codebase
and replaced with `PythonObject` in order to make RAII the predominant
pattern when dealing with native Python objects.
Differential Revision: http://reviews.llvm.org/D13617
Reviewed By: Greg Clayton
llvm-svn: 250195
Summary:
/usr/lib/debug doesn't exist on Windows so there's no point even
attempting to look for symbol files in there.
Reviewers: zturner, clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13636
llvm-svn: 250175
Summary:
This adds platform code without the cmake/gmake glue to the existing infrastructure.
The missing and incompatibility ptrace(2) bits (existing in FreeBSD) are under active research and development and will be submitted once verified to work.
This code was tested to build and run on NetBSD-current/amd64.
Proper build scripts will be integrated separately as a new commit.
Reviewers: joerg
Subscribers: tfiala, brucem, labath, emaste, lldb-commits
Differential Revision: http://reviews.llvm.org/D13334
llvm-svn: 250146
set to true, but all plans run by RunThreadPlan need to have this set to false so they will
return control to RunThreadPlan without consulting plans higher on the stack.
Since this seems like a common error, I also modified RunThreadPlan to enforce this behavior.
<rdar://problem/22543166>
llvm-svn: 250084
strictly necessary because RunToAddress is always used as a subsidiary plan, so
it's ShouldStop seldom matters. But get it right anyway.
llvm-svn: 250083
Summary:
- Changed from 16 bits to 8 bits for Intel Architecture
-- FXSAVE structure now conforms with the layout of FXSAVE
area specified by IA Architecture Software Developer Manual
- Modified Linux and FreeBSD specific files to support this change
-- MacOSX already uses 8 bits for ftag register
- Modified TestRegisters.py and a.cpp:
-- Change allows 8 bit comparison of ftag values
-- Change resolves Bug 24733:
Removed XFAIL for Clang as the test works and passes for
Clang compiler as well
-- Change provides a Generic/Better way of testing Bug 24457
and Bug 25050 by using 'int3' inline assembly in inferior
Signed-off-by: Abhishek Aggarwal <abhishek.a.aggarwal@intel.com>
Reviewers: ovyalov, jingham, clayborg
Subscribers: tfiala, emaste
Differential Revision: http://reviews.llvm.org/D13587
llvm-svn: 250022
The underlying raw_string_stream buffer was not being flushed
after asking llvm to collect the backtrace. This worked fine
on OS X but was failing to print anything on Linux.
llvm-svn: 249930
With this change, liblldb is 95% of the way towards being able
to work under both Python 2.x and Python 3.x. This should
introduce no functional change for Python 2.x, but for Python
3.x there are some important changes. Primarily, these are:
1) PyString doesn't exist in Python 3. Everything is a PyUnicode.
To account for this, PythonString now stores a PyBytes instead
of a PyString. In Python 2, this is equivalent to a PyUnicode,
and in Python 3, we do a conversion from PyUnicode to PyBytes
and store the PyBytes.
2) PyInt doesn't exist in Python 3. Everything is a PyLong. To
account for this, PythonInteger stores a PyLong instead of a
PyInt. In Python 2.x, this requires doing a conversion to
PyLong when creating a PythonInteger from a PyInt. In 3.x,
there is no PyInt anyway, so we can assume everything is a
PyLong.
3) PyFile_FromFile doesn't exist in Python 3. Instead there is a
PyFile_FromFd. This is not addressed in this patch because it
will require quite a large change to plumb fd's all the way
through the system into the ScriptInterpreter. This is the only
remaining piece of the puzzle to get LLDB supporting Python 3.x.
Being able to run the test suite is not addressed in this patch.
After the extension module can compile and you can enter an embedded
3.x interpreter, the test suite will be addressed in a followup.
llvm-svn: 249886
Instead check what languages are supported for expressions; use C if available,
but otherwise pick one of the supported languages.
This can be overridden using the target settings.
<rdar://problem/22290878>
llvm-svn: 249864
Summary:
EnablePluginLogging, GetPluginCommandHelp and ExecutePluginCommand aren't
implemented or used anywhere, so remove them from the Hexagon and POSIX
Dynamic Loaders as well as the FreeBSD process.
Reviewers: clayborg, labath, emaste
Subscribers: lldb-commits, emaste
Differential Revision: http://reviews.llvm.org/D13581
llvm-svn: 249840
Summary:
This is Darwin only.
The symbol defined by ${LLDB_VERS_GENERATED_FILE} is used by
source/lldb.cpp, so anything that uses lldb.cpp (which is in
lldbBase) should also have the generated symbol. This means
that the entire process can be centralized within source/CMakeLists.txt
where lldbBase is constructed.
Additionally, the custom command should have dependencies on the
project file as well as the generation script so that if either
changes, the version file is correctly re-generated and everything
is re-linked appropriately.
* cmake/LLDBDependencies.cmake: Remove everything related to
the generated version file from here.
* source/CMakeLists.txt: On Darwin, add the generated version
file to the sources that make up lldbBase. Also, create a
custom target and make lldbBase depend on it to re-generate
the generated file as needed.
* source/API/CMakeLists.txt: Don't need to build the generated
version file here or use it to control linking against swig_wrapper.
* tools/lldb-server/CMakeLists.txt: Likewise.
Reviewers: dawn, sas, clayborg, zturner
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13552
llvm-svn: 249806
when they introduced android testsuite regressions. Pavel has run the
testsuite against the updated patch and it completes cleanly now.
The original commit message:
Fixing a subtle issue on Mac OS X systems with dSYMs (possibly
introduced by r235737 but I didn't look into it too closely).
A dSYM can have a per-UUID plist in it which tells lldb where
to find an executable binary for the dSYM (DBGSymbolRichExecutable)
- other information can be included in this plist, like how to
remap the source file paths from their build pathnames to their
long-term storage pathnames.
This per-UUID plist is a unusual; it is used probably exclusively
inside apple with our build system. It is not created by default
in normal dSYMs.
The problem was like this:
1. lldb wants to find an executable, given only a UUID
(this happens when lldb is doing cross-host debugging
and doesn't have a copy of the target system's binaries)
2. It eventually calls LocateMacOSXFilesUsingDebugSymbols
which does a spotlight search for the dSYM on the local
system, and failing that, tries the DBGShellCommands
command to find the dSYM.
3. It gets a dSYM. It reads the per-UUID plist in the dSYM.
The dSYM has a DBGSymbolRichExecutable kv pair pointing to
the binary on a network filesystem.
4. Using the binary on the network filesystem, lldb now goes
to find the dSYM.
5. It starts by looking for a dSYM next to the binary it found.
6. lldb is now reading the dSYM over a network filesystem,
ignoring the one it found on its local filesystem earlier.
Everything still *works* but it's much slower.
This would be a tricky one to write up in a testsuite case;
you really need the binary to not exist on the local system.
And LocateMacOSXFilesUsingDebugSymbols will only compile on
Mac OS X - even if I found a way to write up a test case, it
would not run anywhere but on a mac.
One change Greg wanted while I was touching this code was to
have LocateMacOSXFilesUsingDebugSymbols (which could be asked
to find a binary OR find a dSYM) to instead return a ModuleSpec
with the sum total of everything it could find. This
change of passing around a ModuleSpec instead of a FileSpec
was percolated up into ModuleList::GetSharedModule.
The changes to LocateMacOSXFilesUsingDebugSymbols look larger
than they really are - there's a lot of simple whitespace changes
in there.
I ran the testsuites on mac, no new regressions introduced
<rdar://problem/21993813>
llvm-svn: 249755
This involved changing the TypeSystem::CreateInstance to take a module or a target. This allows type systems to create an AST for modules (no expression support needed) or targets (expression support is needed) and return the correct class instance for both cases.
llvm-svn: 249747
Summary:
In bug 24074, the type information is not shown
correctly. This commit includes the following -
-> Changes for displaying correct type based on
current lexical scope for the command "image
lookup -t"
-> The corresponding testcase.
-> This patch was reverted due to segfaults in
FreeBSD and Mac, I fixed the problems for both now.
Reviewers: emaste, granata.enrico, jingham, clayborg
Differential Revision: http://reviews.llvm.org/D13290
llvm-svn: 249673
Summary: This change fixes pr24916. As associated test has been added.
Reviewers: clayborg
Subscribers: zturner, lldb-commits
Differential Revision: http://reviews.llvm.org/D13224
llvm-svn: 249629
When the target settings are consulted to decide the expression language
is decided in CommandObjectExpression, this doesn't help if you're running
SBFrame::EvaluateExpression(). Moving the logic into UserExpression fixes
this.
Based on patch from scallanan@apple.com
Reviewed by: dawn
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13267
llvm-svn: 249624
If a string contained characters outside the ASCII range, lldb-mi would
print them as hexadecimal codes. This patch fixes this behaviour by
converting to UTF-8 instead, by having lldb-mi use registered type
summary providers, when they are available. This patch also fixes
incorrect evaluation of some composite types, like std::string, by
having them use a type registered type summary provider.
Based on patch from evgeny.leviant@gmail.com
Reviewed by: ki.stfu, granata.enrico, clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13058
llvm-svn: 249597
Introduce the notion of Language-based formatter prefix/suffix
This is meant for languages that share certain data types but present them in syntatically different ways, such that LLDB can now have language-based awareness of which of the syntax variations it has to present to the user when formatting those values
This is goodness for new languages and interoperability, but is NFC for existing languages. As such, existing tests cover this
llvm-svn: 249587
Summary:
This removes all uses of virtual on functions
where override could be used, including on destructors.
It also adds override where virtual was previously
missing.
Reviewers: clayborg, labath
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13503
llvm-svn: 249564
Summary:
The StackUsesFrames and FunctionCallsChangeCFA virtual functions
aren't used anywhere and aren't overridden by anything.
They were introduced when the ABISysV_ppc* code was added and weren't
used at the time. The review for the commit that added them can be
found at http://reviews.llvm.org/D5988
The commit comment notes that backtraces don't yet work:
Backtraces don't work. This is due to PowerPC ABI using a
backchain pointer in memory, instead of a dedicated frame
pointer register for the backchain.
So there is a possibility these were added with the intent of using
them in the future.
Reviewers: clayborg, jingham, jhibbits, emaste
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13506
llvm-svn: 249563
Summary: This was deprecated and removed.
Reviewers: clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13463
llvm-svn: 249452
Summary: We were missing the symbol for the version number.
Reviewers: clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13271
llvm-svn: 249434
Summary:
When `module_spec.GetFileSpec().GetDirectory().AsCString()` returned a `nullptr` this line caused a segmentation fault:
`std::string module_directory = module_spec.GetFileSpec().GetDirectory().AsCString()`
Some context:
I was remote debugging an executable built with Clang in an Ubuntu VM on my Windows machine using lldb-mi. I copied the executable and nothing else from the Ubuntu VM to the Windows machine.
Then started lldb-server in the Ubuntu VM:
```
./bin/lldb-server gdbserver *:8888 -- /home/enlight/Projects/dbgmits/build/Debug/data_tests_target
```
And ran `lldb-mi --interpreter` on Windows with the following commands:
```
-file-exec-and-symbols C:\Projects\data_tests_target
-target-select remote 192.168.56.101:8888
-exec-continue
```
After which the segmentation fault occurred at the aforementioned line. Inside this method `module_spec.GetFileSpec()` returns an empty `FileSpec` (no dir, no filename), while `module_spec.GetSymbolFileSpec().GetFilename()` returns `"libc-2.19.so"`.
Patch thanks to Vadim Macagon.
Reviewers: brucem, zturner, clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13201
llvm-svn: 249387
SUMMARY:
This patch includes:
1. Emulation of prologue/epilogue and branch instructions for microMIPS.
2. Setting up alternate disassembler (to be used for microMIPS).
So there will be two disassembler instances, one for microMIPS and other for MIPS.
Appropriate disassembler will be used based on the address class of instruction address.
3. Some of the branch instructions does not have fixed sized delay slot, that means delay slot instruction can be of 2-byte or 4-byte.
For this "m_next_inst_size" has been introduced which stores the size of next instruction (i.e size of delay slot instruction in case of branch).
This can be used wherever the size of next instruction is required.
4. A minor change to use mips32 register names instead of mips64 names.
Reviewers: clayborg, tberghammer
Subscribers: mohit.bhakkad, sagar, jaydeep, nitesh.jain, lldb-commits
Differential Revision: http://reviews.llvm.org/D13282
llvm-svn: 249381
This patch adds a new command 'language renderscript allocation list' for printing the details of all loaded RS allocations.
In order to work out this information lldb JITs the runtime for the data it wants.
This has a penalty of a couple seconds latency, so is only done once for each allocation and the results cached.
If the user later wants to recalculate this information however, they can force lldb to do so with the --refresh flag.
Reviewed by: jingham, clayborg
Subscribers: lldb-commits, ADodds, domipheus, dean, tberghammer, danalbert, srhines
Differential Revision: http://reviews.llvm.org/D13247
llvm-svn: 249380
Added the ability to specify if an attach by name should be synchronous or not in SBAttachInfo and ProcessAttachInfo.
<rdar://problem/22821480>
llvm-svn: 249361
The concept here is that languages may have different ways of communicating
results. In particular, languages may have different names for their result
variables and in fact may have multiple types of result variables (e.g.,
error results). Materializer was tied to one specific model of result handling.
Instead, now UserExpressions can register their own handlers for the result
variables they inject. This allows language-specific code in Materializer to
be moved into the expression parser plug-in, and it simplifies Materializer.
These delegates are subclasses of PersistentVariableDelegate.
PersistentVariableDelegate can provide the name of the result variable, and is
notified when the result variable is populated. It can also be used to touch
persistent variables if need be, updating language-specific state. The
UserExpression owns the delegate and can decide on its result based on
consulting all of its (potentially multiple) delegates.
The user expression itself now makes the determination of what the final result
of the expression is, rather than relying on the Materializer, and I've added a
virtual function to UserExpression to allow this.
llvm-svn: 249233
This is meant to support languages that can do some sort of bridging from<-->to these ObjC types via types that statically vend themselves as Cocoa types, but dynamically have an implementation that does not match any of our well-known types, but where an introspecting formatter can be vended by the bridged language
llvm-svn: 249185
* Use .ARM.exidx as a fallback unwind plan for non-call site when the
instruction emulation based unwind failed.
* Work around an old compiler issue where the compiler isn't sort the
entries in .ARM.exidx based on their address.
* Fix unwind info parsing when the virtual file address >= 0x80000000
* Fix bug in unwind info parsing when neither lr nor pc is explicitly
restored.
Differential revision: http://reviews.llvm.org/D13380
llvm-svn: 249119
The ClangExpressionVariable::CreateVariableInList functions looked cute, but
caused more confusion than they solved. I removed them, and instead made sure
that there are adequate facilities for easily adding newly-constructed
ExpressionVariables to lists.
I also made some of the constructors that are common be generic, so that it's
possible to construct expression variables from generic places (like the ABI and
ValueObject) without having to know the specifics about the class.
llvm-svn: 249095
Currently, it only supports Objective-C - C++ types can be looked up through debug info via 'image lookup -t', whereas ObjC types via this command are looked up by runtime introspection
This behavior is in line with type lookup's behavior in Xcode 7, but I am definitely open to feedback as to what makes the most sense here
llvm-svn: 249047
Also added some target-level search functions so that persistent variables and
symbols can be searched for without hand-iterating across the map of
TypeSystems.
llvm-svn: 249027
GP registers for o32 applications were always giving zero value because SetType() on the RegisterValue was causing the accessor functions to pickup the value from m_scalar of RegisterValue which is zero.
In this patch byte size and byte order of register value is set at the time of setting the value of the register.
llvm-svn: 249020
Run the getprop command with AdbClient::Shell instead of
Platform::RunShellCommand because getting the output from getprop
with Platform::RunShellCommand have some (currently unknown) issues.
llvm-svn: 249014
Change the way we detect if we have to place a thumb breakpoint instead
of an arm breakpoint in the case when no symbol table or mapping symbols
are available. Detect it based on the LSB of the FileAddress instead of
the LSB of the LoadAddress because the LSB of the LoadAddress is already
masked out.
llvm-svn: 249013
The hack is there to work around an incorrect load address reported
by the android linker on API 21 and 22 devices. This CL restricts the
hack to those android API levels.
Differential revision: http://reviews.llvm.org/D13288
llvm-svn: 249012
introduced by r235737 but I didn't look into it too closely).
A dSYM can have a per-UUID plist in it which tells lldb where
to find an executable binary for the dSYM (DBGSymbolRichExecutable)
- other information can be included in this plist, like how to
remap the source file paths from their build pathnames to their
long-term storage pathnames.
This per-UUID plist is a unusual; it is used probably exclusively
inside apple with our build system. It is not created by default
in normal dSYMs.
The problem was like this:
1. lldb wants to find an executable, given only a UUID
(this happens when lldb is doing cross-host debugging
and doesn't have a copy of the target system's binaries)
2. It eventually calls LocateMacOSXFilesUsingDebugSymbols
which does a spotlight search for the dSYM on the local
system, and failing that, tries the DBGShellCommands
command to find the dSYM.
3. It gets a dSYM. It reads the per-UUID plist in the dSYM.
The dSYM has a DBGSymbolRichExecutable kv pair pointing to
the binary on a network filesystem.
4. Using the binary on the network filesystem, lldb now goes
to find the dSYM.
5. It starts by looking for a dSYM next to the binary it found.
6. lldb is now reading the dSYM over a network filesystem,
ignoring the one it found on its local filesystem earlier.
Everything still *works* but it's much slower.
This would be a tricky one to write up in a testsuite case;
you really need the binary to not exist on the local system.
And LocateMacOSXFilesUsingDebugSymbols will only compile on
Mac OS X - even if I found a way to write up a test case, it
would not run anywhere but on a mac.
One change Greg wanted while I was touching this code was to
have LocateMacOSXFilesUsingDebugSymbols (which could be asked
to find a binary OR find a dSYM) to instead return a ModuleSpec
with the sum total of everything it could find. This
change of passing around a ModuleSpec instead of a FileSpec
was percolated up into ModuleList::GetSharedModule.
The changes to LocateMacOSXFilesUsingDebugSymbols look larger
than they really are - there's a lot of simple whitespace changes
in there.
I ran the testsuites on mac, no new regressions introduced
<rdar://problem/21993813>
llvm-svn: 248985
This is meant to support languages that have a scripting mode with top-level code that acts as global
For now, this flag only controls whether 'frame variable' will attempt to treat globals as locals when within such a function
llvm-svn: 248960
the corresponding TypeSystem. This makes sense because what kind of data there
is -- and how it can be looked up -- depends on the language.
Functionality that is common to all type systems is factored out into
PersistentExpressionState.
llvm-svn: 248934
.ARM.exidx/.ARM.extab sections contain unwind information used on ARM
architecture from unwinding from an exception.
Differential revision: http://reviews.llvm.org/D13245
llvm-svn: 248903
On android when debugging an apk we run lldb-server as application user
because the sell user (on non-rooted device) can't attach to an
application. The problem is that "adb pull" will run as a shell user
what can't access to files created by lldb-server because they will be
owned by the application user. This CL changes the oat symbolization
code to run "oatdump --symbolize" to generate an output what is owned
by the shell user.
Differential revision: http://reviews.llvm.org/D13162
llvm-svn: 248788
There are still a bunch of dependencies on the plug-in, but this helps to
identify them.
There are also a few more bits we need to move (and abstract, for example the
ClangPersistentVariables).
llvm-svn: 248612
See:
https://llvm.org/bugs/show_bug.cgi?id=24926
for details. On OS X, when LLDB.framework is
not part of the lldb.dylib path, the supporting
executable path is resolved to be the bin directory
sitting next to the lib directory with the dylib
lives. Not a perfect solution, but we also can't
base it on the executable path since both Python
and the lldb driver can be the executable.
llvm-svn: 248545
Summary: This is no longer needed as this file no longer calls backtrace().
Reviewers: clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13049
llvm-svn: 248457
And remove the switch default, so that the -Wcovered-switch-default
warning will catch new types next time they're added.
Differential Revision: http://reviews.llvm.org/D13096
llvm-svn: 248414
Summary:
With this change DWARFASTParserClang::CompleteTypeFromDWARF returns false if
DWARFASTParserClang::ParseChildMembers returns false. Similarly, it returns
false if any base class is of an incomplete type. This helps in cases like
these:
class Foo
{
public:
std::string str;
};
...
Foo f;
If a file with the above code is compiled with a modern clang but without
the -fno-limit-debug-info (or similar) option, then the DWARF has only
a forward declration for std::string. In which case, the type for
"class Foo" cannot be completed. If LLDB does not detect that a child
member has incomplete type, then it wrongly conveys to clang (the LLDB
compiler) that "class Foo" is complete, and consequently crashes due to
an assertion failure in clang when running commands like "p f" or
"frame var f".
Reviewers: clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13066
llvm-svn: 248401
Summary:
The following situation occured in TestAttachResume:
The inferior was stoped at a breakpoint and we did a continue, immediately followed by a detach.
Since there was a trap instruction under the IP, the continue did a step-over-breakpoint before
resuming the inferior for real. In some cases, the detach command was executed between these two
events (after the step-over stop, but before continue). Here, public state was running, but
private state was stopped. This caused a problem because HaltForDestroyOrDetach was checking the
public state to see whether it needs to stop the process (call Halt()), but Halt() was checking
the private state and concluded that there is nothing for it to do.
Solution: Instead of Halt() call SendAsyncInterrupt(), which will then cause Halt() to be
executed in the context of the private state thread. I also rename HaltForDestroyOrDetach to
reflect it does not call halt directly.
Reviewers: jingham, clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13056
llvm-svn: 248371
Summary:
In bug 24074, the type information is not shown
correctly. This commit includes the following -
-> Changes for displaying correct type based on
current lexical scope for the command "image
lookup -t"
-> The corresponding testcase.
Reviewers: jingham, ovyalov, spyffe, richard.mitton, clayborg
Differential Revision: http://reviews.llvm.org/D12404
llvm-svn: 248366
Summary:
The default case doesn't need to be here as the switch covers
all possible values. If there's a new "lazy bool" value added
in the future, the compiler would start to warn about the new
case not being covered.
Reviewers: granata.enrico, clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13084
llvm-svn: 248365
Different type system may have different notions of attributes of a type that do not matter for data formatters matching purposes
For instance, in the case of clang types, we remove some qualifiers (e.g. "volatile") as it doesn't make much sense to differentiate volatile T from T in the data formatters
This new API allows each type system to generate, if needed, a type that does not have those unwanted attributes that the data formatters can then consume to generate matches
llvm-svn: 248359
on iOS devices; fallout from Vince's cleanups made
in r237218 back in May. iOS native lldbs will call
StartDebugserverProcess() with a random port #
(see ProcessGDBRemote::LaunchAndConnectToDebugserver)
and neither side of this conditional expression should
be followed in that case.
I added an "if (in_port == 0) { ..." check around the
entire if/then/else and indented the block of code so
the diff looks larger than it really is.
<rdar://problem/21712643>
llvm-svn: 248343
The argdumper-based launching is more friendly to System Integrity Protection, and will work on older releases of OS X as well
Leave non-Apple builds alone
llvm-svn: 248338
This is meant to cover cases such as the obvious
Base *base = new Derived();
where GetDynamicTypeAndAddress(base) would return the type "Derived", not "Derived *"
llvm-svn: 248315
Summary:
This is no longer related to Clang and is just an opaque pointer
to data for a compiler type.
Reviewers: clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13039
llvm-svn: 248288
Summary:
pthread_setname_np() is a nonstandard GNU extension and isn't available
in every C library. Check before it's usage that GLIBC is available or
that we are targeting Android.
Reviewers: clayborg, ovyalov
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13019
llvm-svn: 248280
After the std::move operation the unique pointer is null.
So this statement always returns a null pointer.
Also remove unnecessary call to Module::ParseAllDebugSymbols(),
which spews errors due to how it incorrectly tries to parse DWARF DIE types.
llvm-svn: 248274
Summary:
Normally, these macros are defined in fnctl.h. However, GLIBC exposes their
definition through <sys/file.h> too. This change allows us to compile
LLDB with non-GLIBC C libraries.
Reviewers: clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D13022
llvm-svn: 248255
This patch adds some of the groundwork required for tracking the lifetime of scripts and allocations and collecting data associated with them during execution.
Committed on behalf of Aidan Dodds.
Authored by: ADodds
Reviewed by: clayborg, jingham
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D12936
llvm-svn: 248149
Both GNU AS and LLVM emits language type DW_LANG_Mips_Assembler for
all assembly code.
Differential revision: http://reviews.llvm.org/D12962
llvm-svn: 248146
We use the symbolic link to resolver to find the target of the LLDB shlib
symlink if there is a symlink. This allows us to find shlib-relative resources
even when running under the testsuite, where _lldb.so is a symlink in the Python
resource directory.
Also changed a comment to be slightly more clear about what resolve_path in the
constructor for FileSpec means, since if we were actually using realpath() this
code wouldn't have been necessary.
http://reviews.llvm.org/D12984
llvm-svn: 248048
Summary:
With the recent changes to separate clang from the core structures
of LLDB, many inclusions of clang headers can be removed.
Reviewers: clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D12954
llvm-svn: 248004
For C++ and ObjC, dynamic values are always (at least somewhat) pointer-like in nature, so a ValueType of scalar is actually good enough that it could originally be hardcoded as the right choice
Other languages, might have broader notions of things that are dynamic (e.g. a language where a value type can be dynamic). In those cases, it might actually be the case that a dynamic value is a pointer-to the data, or even a host address if dynamic expression results entirely in host space are being talked about
This patch enables the language runtime to make that decision, and makes ValueObjectDynamicValue comply with it
llvm-svn: 247957
This cleans up type systems to be more pluggable. Prior to this we had issues:
- Module, SymbolFile, and many others has "ClangASTContext &GetClangASTContext()" functions. All have been switched over to use "TypeSystem *GetTypeSystemForLanguage()"
- Cleaned up any places that were using the GetClangASTContext() functions to use TypeSystem
- Cleaned up Module so that it no longer has dedicated type system member variables:
lldb::ClangASTContextUP m_ast; ///< The Clang AST context for this module.
lldb::GoASTContextUP m_go_ast; ///< The Go AST context for this module.
Now we have a type system map:
typedef std::map<lldb::LanguageType, lldb::TypeSystemSP> TypeSystemMap;
TypeSystemMap m_type_system_map; ///< A map of any type systems associated with this module
- Many places in code were using ClangASTContext static functions to place with CompilerType objects and add modifiers (const, volatile, restrict) and to make typedefs, L and R value references and more. These have been made into CompilerType functions that are abstract:
class CompilerType
{
...
//----------------------------------------------------------------------
// Return a new CompilerType that is a L value reference to this type if
// this type is valid and the type system supports L value references,
// else return an invalid type.
//----------------------------------------------------------------------
CompilerType
GetLValueReferenceType () const;
//----------------------------------------------------------------------
// Return a new CompilerType that is a R value reference to this type if
// this type is valid and the type system supports R value references,
// else return an invalid type.
//----------------------------------------------------------------------
CompilerType
GetRValueReferenceType () const;
//----------------------------------------------------------------------
// Return a new CompilerType adds a const modifier to this type if
// this type is valid and the type system supports const modifiers,
// else return an invalid type.
//----------------------------------------------------------------------
CompilerType
AddConstModifier () const;
//----------------------------------------------------------------------
// Return a new CompilerType adds a volatile modifier to this type if
// this type is valid and the type system supports volatile modifiers,
// else return an invalid type.
//----------------------------------------------------------------------
CompilerType
AddVolatileModifier () const;
//----------------------------------------------------------------------
// Return a new CompilerType adds a restrict modifier to this type if
// this type is valid and the type system supports restrict modifiers,
// else return an invalid type.
//----------------------------------------------------------------------
CompilerType
AddRestrictModifier () const;
//----------------------------------------------------------------------
// Create a typedef to this type using "name" as the name of the typedef
// this type is valid and the type system supports typedefs, else return
// an invalid type.
//----------------------------------------------------------------------
CompilerType
CreateTypedef (const char *name, const CompilerDeclContext &decl_ctx) const;
};
Other changes include:
- Removed "CompilerType TypeSystem::GetIntTypeFromBitSize(...)" and CompilerType TypeSystem::GetFloatTypeFromBitSize(...) and replaced it with "CompilerType TypeSystem::GetBuiltinTypeForEncodingAndBitSize(lldb::Encoding encoding, size_t bit_size);"
- Fixed code in Type.h to not request the full type for a type for no good reason, just request the forward type and let the type expand as needed
llvm-svn: 247953
If a breakpoint was hit in the inferior after shutdown had
started but before it was complete, it would cause an unclean
terminate of the inferior, leading to various problems the most
visible of which is that handles to the inferior executable would
remain locked, and the test suite would fail to run subsequent
tests because it could not recompile the inferior.
This fixes a major source of flakiness in the test suite.
llvm-svn: 247929
Character with ASCII code 0 is incorrectly treated by LLDB as the end of
RSP packet. The left of the debugger server output is silently ignored.
Patch from evgeny.leviant@gmail.com
Reviewed by: clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D12523
llvm-svn: 247908
The Go runtime schedules user level threads (goroutines) across real threads.
This adds an OS plugin to create memory threads for goroutines.
It supports the 1.4 and 1.5 go runtime.
Differential Revision: http://reviews.llvm.org/D5871
llvm-svn: 247852
The implications of this bug where that "log disable windows" would
not actually disable the log, and worse it would lock the file handle
making it impossible to delete the file until lldb was shut down.
This was then causing the test suite to fail, because the test suite
tries to delete log files in certain situations.
llvm-svn: 247841
Summary: Supports the parsing of the "using namespace XXX" and "using XXX::XXX" directives. Added ambiguity errors when it two decls with the same name are encountered (see comments in TestCppNsImport). Fixes using directives being duplicated for anonymous namespaces. Fixes GetDeclForUID for specification DIEs.
Reviewers: sivachandra, chaoren, clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D12897
llvm-svn: 247836
Split-dwarf uses a different header format to specify the address range
for the elements of the location lists.
Differential revision: http://reviews.llvm.org/D12880
llvm-svn: 247789
GCC don't use the is_prologue_end flag to mark the first instruction
after the prologue. Instead of it it is issuing a line table entry for
the first instruction of the prologue and one for the first instruction
after the prologue. If the size of the prologue is 0 instruction then
the 2 line entry will have the same file address.
We remove these duplicates entries as they are violating the dwarf spec
and can cause confusion in the debugger. To prevent the lost of
information about the end of prologue we should set the prologue end
flag for the line entries what are representing more then 1 entry.
Differential revision: http://reviews.llvm.org/D12757
llvm-svn: 247788
Use Breakpoint::AddName to mark all RenderScript kernel breakpoints with the name 'RenderScriptKernel'.
Also update logging channels to include LIBLLDB_LOG_BREAKPOINT where appropriate.
llvm-svn: 247782
SUMMARY:
Refer to http://lists.llvm.org/pipermail/lldb-dev/2015-August/008024.html for discussion
on this topic. Bare-iron target like YAMON gdb-stub does not support qProcessInfo, qC,
qfThreadInfo, Hg and Hc packets. Reply from ? packet is as simple as S05. There is no
packet which gives us process or threads information. In such cases, assume pid=tid=1.
Reviewers: clayborg
Subscribers: nitesh.jain, mohit.bhakkad, sagar, bhushan and lldb-commits
Differential Revision: http://reviews.llvm.org/D12876
llvm-svn: 247773
Summary: SymbolFileDWARF now creates VarDecl and BlockDecl and adds them to the Decl tree. Then, in ClangExpressionDeclMap it uses the Decl tree to search for a variable. This fixes lots of variable scoping problems.
Reviewers: sivachandra, chaoren, spyffe, clayborg
Subscribers: tberghammer, jingham, lldb-commits
Differential Revision: http://reviews.llvm.org/D12658
llvm-svn: 247746
"gcc" register numbers are now correctly referred to as "ehframe"
register numbers. In almost all cases, ehframe and dwarf register
numbers are identical (the one exception is i386 darwin where ehframe
regnums were incorrect).
The old "gdb" register numbers, which I incorrectly thought were
stabs register numbers, are now referred to as "Process Plugin"
register numbers. This is the register numbering scheme that the
remote process controller stub (lldb-server, gdbserver, core file
support, kdp server, remote jtag devices, etc) uses to refer to the
registers. The process plugin register numbers may not be contiguous
- there are remote jtag devices that have gaps in their register
numbering schemes.
I removed all of the enums for "gdb" register numbers that we had
in lldb - these were meaningless - and I put LLDB_INVALID_REGNUM
in all of the register tables for the Process Plugin regnum slot.
This change is almost entirely mechnical; the one actual change in
here is to ProcessGDBRemote.cpp's ParseRegisters() which parses the
qXfer:features:read:target.xml response. As it parses register
definitions from the xml, it will assign sequential numbers as the
eRegisterKindLLDB numbers (the lldb register numberings must be
sequential, without any gaps) and if the xml file specifies register
numbers, those will be used as the eRegisterKindProcessPlugin
register numbers (and those may have gaps). A J-Link jtag device's
target.xml does contain a gap in register numbers, and it only
specifies the register numbers for the registers after that gap.
The device supports many different ARM boards and probably selects
different part of its register file as appropriate.
http://reviews.llvm.org/D12791
<rdar://problem/22623262>
llvm-svn: 247741
Before we had:
ClangFunction
ClangUtilityFunction
ClangUserExpression
and code all over in lldb that explicitly made Clang-based expressions. This patch adds an Expression
base class, and three pure virtual implementations for the Expression kinds:
FunctionCaller
UtilityFunction
UserExpression
You can request one of these expression types from the Target using the Get<ExpressionType>ForLanguage.
The Target will then consult all the registered TypeSystem plugins, and if the type system that matches
the language can make an expression of that kind, it will do so and return it.
Because all of the real expression types need to communicate with their ExpressionParser in a uniform way,
I also added a ExpressionTypeSystemHelper class that expressions generically can vend, and a ClangExpressionHelper
that encapsulates the operations that the ClangExpressionParser needs to perform on the ClangExpression types.
Then each of the Clang* expression kinds constructs the appropriate helper to do what it needs.
The patch also fixes a wart in the UtilityFunction that to use it you had to create a parallel FunctionCaller
to actually call the function made by the UtilityFunction. Now the UtilityFunction can be asked to vend a
FunctionCaller that will run its function. This cleaned up a lot of boiler plate code using UtilityFunctions.
Note, in this patch all the expression types explicitly depend on the LLVM JIT and IR, and all the common
JIT running code is in the FunctionCaller etc base classes. At some point we could also abstract that dependency
but I don't see us adding another back end in the near term, so I'll leave that exercise till it is actually necessary.
llvm-svn: 247720
Eugene Zelenko