Summary:
This patch fixes issues with a stack realignment.
MSVC maintains two frame pointers (`ebx` and `ebp`) for a realigned stack - one
is used for access to function parameters, while another is used for access to
locals. To support this the patch:
- adds an alternative frame pointer (`ebx`);
- considers stack realignment instructions (e.g. `and esp, -32`);
- along with CFA (Canonical Frame Address) which point to the position next to
the saved return address (or to the first parameter on the stack) introduces
AFA (Aligned Frame Address) which points to the position of the stack pointer
right after realignment. AFA is used for access to registers saved after the
realignment (see the test);
Here is an example of the code with the realignment:
```
struct __declspec(align(256)) OverAligned {
char c;
};
void foo(int foo_arg) {
OverAligned oa_foo = { 1 };
auto aaa_foo = 1234;
}
void bar(int bar_arg) {
OverAligned oa_bar = { 2 };
auto aaa_bar = 5678;
foo(1111);
}
int main() {
bar(2222);
return 0;
}
```
and here is the `bar` disassembly:
```
push ebx
mov ebx, esp
sub esp, 8
and esp, -100h
add esp, 4
push ebp
mov ebp, [ebx+4]
mov [esp+4], ebp
mov ebp, esp
sub esp, 200h
mov byte ptr [ebp-200h], 2
mov dword ptr [ebp-4], 5678
push 1111 ; foo_arg
call j_?foo@@YAXH@Z ; foo(int)
add esp, 4
mov esp, ebp
pop ebp
mov esp, ebx
pop ebx
retn
```
Reviewers: labath, zturner, jasonmolenda, stella.stamenova
Reviewed By: jasonmolenda
Subscribers: abidh, lldb-commits
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D53435
llvm-svn: 345577
We currently had a 2-step process where we had to call
SetBaseClassesForType and DeleteBaseClasses. Every single caller
followed this exact 2-step process, and there was manual memory
management going on with raw pointers. We can do better than this
by storing a vector of unique_ptrs and passing this around.
This makes for a cleaner API, and we only need to call one method
so there is no possibility of a user forgetting to call
DeleteBaseClassSpecifiers.
In addition to this, it also makes for a *simpler* API. Part of
why I wanted to do this is because when I was implementing the native
PDB interface I had to spend some time understanding exactly what I
was deleting and why. ClangAST has significant mental overhead
associated with it, and reducing the API surface can go along
way to making it simpler for people to understand.
Differential Revision: https://reviews.llvm.org/D53590
llvm-svn: 345312
This fixes a bug PlatformDarwin::SDKSupportsModule introduced by
https://reviews.llvm.org/D47889. VersionTuple::tryParse() can deal
with an optional third (micro) component, but the parse will fail when
there are extra characters after the version number (e.g.: trying to
parse the substring "12.0.sdk" out of "iPhoneSimulator12.0.sdk" fails
after that patch). Fixed here by stripping the ".sdk" suffix first.
(Part of) rdar://problem/45041492
Differential Revision https://reviews.llvm.org/D53677
llvm-svn: 345274
Summary: A RichManglingContext constructed with an invalid demangled name or with a demangled function name without any context will have an empty context. This triggers an assertion in RichManglingContext::GetBufferRef() when debugging a native Windows process on x86 when it shouldn't. Remove the assertion.
Reviewers: aleksandr.urakov, zturner, lldb-commits
Reviewed By: zturner
Subscribers: erik.pilkington
Differential Revision: https://reviews.llvm.org/D52626
llvm-svn: 343292
NativeProcessProtocol::ReadMemoryWithoutTrap had a bug, where it failed
to properly remove inserted breakpoint opcodes if the memory read
partially overlapped the trap opcode. This could not happen on x86
because it has a one-byte breakpoint instruction, but it could happen on
arm, which has a 4-byte breakpoint instruction (in arm mode).
Since triggerring this condition would only be possible on an arm
machine (and even then it would be a bit tricky). I test this using a
NativeProcessProtocol unit test.
llvm-svn: 343076
Summary:
NativeProcessProtocol is an abstract class, but it still contains a
significant amount of code. Some of that code is tested via tests of
specific derived classes, but these tests don't run everywhere, as they
are OS and arch-specific. They are also relatively high-level, which
means some functionalities (particularly the failure cases) are
hard/impossible to test.
In this approach, I replace the abstract methods with mocks, which
allows me to inject failures into the lowest levels of breakpoint
setting code and test the class behavior in this situation.
Reviewers: zturner, teemperor
Subscribers: mgorny, lldb-commits
Differential Revision: https://reviews.llvm.org/D52152
llvm-svn: 342875
Summary:
This patch adds a framework for adding descriptions to the command completions we provide.
It also adds descriptions for completed top-level commands so that we can test this code.
Completions are in general supposed to be displayed alongside the completion itself. The descriptions
can be used to provide additional information about the completion to the user. Examples for descriptions
are function signatures when completing function calls in the expression command or the binary name
when providing completion for a symbol.
There is still some boilerplate code from the old completion API left in LLDB (mostly because the respective
APIs are reused for non-completion related purposes, so the CompletionRequest doesn't make sense to be
used), so that's why I still had to change some function signatures. Also, as the old API only passes around a
list of matches, and the descriptions are for these functions just another list, I had to add some code that
essentially just ensures that both lists are always the same side (e.g. all the manual calls to
`descriptions->AddString(X)` below a `matches->AddString(Y)` call).
The initial command descriptions that come with this patch are just reusing the existing
short help that is already added in LLDB.
An example completion with descriptions looks like this:
```
(lldb) pl
Available completions:
platform -- Commands to manage and create platforms.
plugin -- Commands for managing LLDB plugins.
```
Reviewers: #lldb, jingham
Reviewed By: #lldb, jingham
Subscribers: jingham, JDevlieghere, lldb-commits
Differential Revision: https://reviews.llvm.org/D51175
llvm-svn: 342181
tests and don't mark this as a failure. This happens when we've
linked against an llvm without the ARM target built in. Davide
added some cmake conditionals to avoid building this test when the
target was absent from llvm's build, but we're still finding some
bots that manage to get in this situation.
<rdar://problem/44270082>
llvm-svn: 342072
Summary:
This commit fixes following problems after rL341782:
- Broken SymbolFilePDBTests
- Warning on comparison of integers of different signs
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D51162
llvm-svn: 341942
The warning is
comparison of integers of different signs: 'const int' and 'const unsigned long'
and triggered by
EXPECT_EQ (num_of_instructions, inst_list.GetSize());
as num_of_instructions is an int in this comparison (and the RHS is size_t).
llvm-svn: 341931
code. This will enable disassembly of the optional subset of
neon that some Cortex cores support. Add a unit test to check
that a few of these instructions disassemble as expected.
<rdar://problem/26674303>
llvm-svn: 341623
TestCompletion was failing quite frequently on our Linux bots. Some tracing
revealed that when we are iterating BaseDir we are not getting all the entries.
More specifically, we are sometimes missing the entry corresponding to the
TestCompletion directory that the first test in DirCompletionAbsolute is
looking for. BaseDir is the directory where lit is creating all the temporary
files. The semantics of opendir/readdir are unclear when it comes to iterating
over a directory that changes contents, but it seems like on Linux you might
fail to list an entry even if it was there before opendir and is still present
throughout the iteration. Changing the test to only look inside of the test-
specific directory seems to fix the instability.
This commit also removes some assertions that were added to try to track down
this issue.
llvm-svn: 341425
If you tried to complete somwthing like ~/., lldb would come up with a lot
of non-existent filenames by concatenating every exisitng file in the directory
with an initial '.'.
This was due to a workaround for an llvm::fs::path::filename behavior that
was not applied selectively enough.
llvm-svn: 341268
Summary:
This class was initially in Host because its implementation used to be
very OS-specific. However, with C++11, it has become a very simple
std::condition_variable wrapper, with no host-specific code.
It is also a general purpose utility class, so it makes sense for it to
live in a place where it can be used by everyone.
This has no effect on the layering right now, but it enables me to later
move the Listener+Broadcaster+Event combo to a lower layer, which is
important, as these are used in a lot of places (notably for launching a
process in Host code).
Reviewers: jingham, zturner, teemperor
Reviewed By: zturner
Subscribers: xiaobai, mgorny, lldb-commits
Differential Revision: https://reviews.llvm.org/D50384
llvm-svn: 341089
Summary:
The syntax highlighting feature so far is mutually exclusive with the lldb feature
that marks the current column in the line by underlining it via an ANSI color code.
Meaning that if you enable one, the other is automatically disabled by LLDB.
This was caused by the fact that both features inserted color codes into the the
source code and were likely to interfere with each other (which would result
in a broken source code printout to the user).
This patch moves the cursor code into the highlighting framework, which provides
the same feature to the user in normal non-C source code. For any source code
that is highlighted by Clang, we now also have cursor marking for the whole token
that is under the current source location. E.g., before we underlined only the '!' in the
expression '1 != 2', but now the whole token '!=' is underlined. The same for function
calls and so on. Below you can see two examples where we before only underlined
the first character of the token, but now underline the whole token.
{F7075400}
{F7075414}
It also simplifies the DisplaySourceLines method in the SourceManager as most of
the code in there was essentially just for getting this column marker to work as
a FormatEntity.
Reviewers: aprantl
Reviewed By: aprantl
Subscribers: lldb-commits
Differential Revision: https://reviews.llvm.org/D51466
llvm-svn: 341003
As we only use PATH_MAX for an assert in a unit test that is supposed
to catch the random failures on the Swift CI bots, we might as well
just ifdef this assert out on Windows.
llvm-svn: 340652
Summary:
The DirCompletionAbsolute is still randomly failing on the nodes even after D50722, so this patch adds more asserts
that verify certain properties on which the actual completion implementation relies on.
The first assert checks that the directory we complete on actually exists. If the directory doesn't exist on the
next CI failure, this assert should catch it and we know that the 0 matches come from a missing base directory.
The second assert is just checking that we are below the PATH_MAX limit that the completion checks against.
This check could randomly fail if the temporary directories we generate are sometimes longer than PATH_MAX,
and the assert can tell us that this is the reason we failed (instead of the mysterious '0 matches').
(As a sidenote: We shouldn't be checking against PATH_MAX anyway in the code (as this is just wrong). Also
the disk completion API really needs a better error mechanism than returning 0 on both error or no-results.)
Reviewers: aprantl, friss
Reviewed By: aprantl
Subscribers: abidh
Differential Revision: https://reviews.llvm.org/D51111
llvm-svn: 340589
Summary:
There is currently a way to skip the debugserver build. See how the CMake
variables SKIP_DEBUGSERVER and LLDB_CODESIGN_IDENTITY are used if you're
interested in that.
This allows us to skip building lldb-server as well. There is another
debug server called ds2 that can be used with LLDB. If you choose to use
ds2, this flag is very useful because it can cut down the build time of LLDB.
Differential Revision: https://reviews.llvm.org/D49282
llvm-svn: 340560
Summary:
CompletionTest.DirCompletionAbsolute had a random failure on a CI node
(in the failure, the completion count was 0, while we expected it to be 1),
but there seems no good reason for it to fail. The sanitizers don't complain
about the test when it's run, so I think we don't have some uninitialized
memory that we access here.
My best bet is that the unique directory selection randomly failed on the CI
node because maybe the FS there doesn't actually guarantee the atomic fopen
assumptions we make in the LLVM code (or some other funny race condition).
In this case a different test run could get the same directory and clean its contents
which would lead to 0 results.
The other possible explanation is that someone changed the CI configuration
on the node and changed the working dir to something very long, which would
make our PATH_MAX test fail (which also leads to 0 results), but I think that case
is unlikely.
This patch is just a stab in the dark that (hopefully) fixes this random failure by
giving each test a (more) unique working directory by appending the unique
test name to the temp-dir prefix. Also adds one more ASSERT_NO_ERROR to
one of our chdir calls just in case that is the reason for failing.
The good thing is that this refactor gets rid of most of the static variables
and files that we previously had as shared state between the different tests.
Potentially fixes rdar://problem/43150260
Reviewers: aprantl
Reviewed By: aprantl
Subscribers: jfb, lldb-commits
Differential Revision: https://reviews.llvm.org/D50722
llvm-svn: 339715
Summary:
This issue came up because it caused problems in our unit tests. The StringPool did connect counterparts only once and silently ignored the values passed in subsequent calls.
The simplest solution for the unit tests would be silent overwrite. In practice, however, it seems useful to assert that we never overwrite a different mangled counterpart.
If we ever have mangled counterparts for other languages than C++, this makes it more likely to notice collisions.
I added an assertion that allows the following cases:
* inserting a new value
* overwriting the empty string
* overwriting with an identical value
I fixed the unit tests, which used "random" strings and thus produced collisions.
It would be even better if there was a way to reset or isolate the StringPool, but that's a different story.
Reviewers: jingham, friss, labath
Subscribers: lldb-commits
Differential Revision: https://reviews.llvm.org/D50536
llvm-svn: 339669
Summary:
We can optimize and refactor some of the classes in RangeMap.h, but first
we should have some tests for all the data structures in there. This adds a first
batch of tests for the Range class itself.
There are some unexpected results happening when mixing invalid and valid ranges, so
I added some FIXME's for that in the tests.
Reviewers: vsk
Reviewed By: vsk
Subscribers: mgorny, lldb-commits
Differential Revision: https://reviews.llvm.org/D50620
llvm-svn: 339611
Summary:
Removing FastDemangle will greatly reduce maintenance efforts. This patch replaces the last point of use in LLDB. Semantics should be kept intact.
Once this is agreed upon, we can:
* Remove the FastDemangle sources
* Add more features e.g. substitutions in template parameters, considering all variations, etc.
Depends on LLVM patch https://reviews.llvm.org/D50586
Reviewers: erik.pilkington, friss, jingham, JDevlieghere
Subscribers: kristof.beyls, chrib, lldb-commits
Differential Revision: https://reviews.llvm.org/D50587
llvm-svn: 339583
Summary: It was not immediately clear to me whether or not non-null-terminated StringRef's are supported in ConstString and/or the counterpart mechanism. From this test it seems to be fine. Maybe useful to keep?
Reviewers: labath
Subscribers: lldb-commits
Differential Revision: https://reviews.llvm.org/D50334
llvm-svn: 339292
Summary:
I set up a new review, because not all the code I touched was marked as a change in old one anymore.
In preparation for this review, there were two earlier ones:
* https://reviews.llvm.org/D49612 introduced the ItaniumPartialDemangler to LLDB demangling without conceptual changes
* https://reviews.llvm.org/D49909 added a unit test that covers all relevant code paths in the InitNameIndexes() function
Primary goals for this patch are:
(1) Use ItaniumPartialDemangler's rich mangling info for building LLDB's name index.
(2) Provide a uniform interface.
(3) Improve indexing performance.
The central implementation in this patch is our new function for explicit demangling:
```
const RichManglingInfo *
Mangled::DemangleWithRichManglingInfo(RichManglingContext &, SkipMangledNameFn *)
```
It takes a context object and a filter function and provides read-only access to the rich mangling info on success, or otherwise returns null. The two new classes are:
* `RichManglingInfo` offers a uniform interface to query symbol properties like `getFunctionDeclContextName()` or `isCtorOrDtor()` that are forwarded to the respective provider internally (`llvm::ItaniumPartialDemangler` or `lldb_private::CPlusPlusLanguage::MethodName`).
* `RichManglingContext` works a bit like `LLVMContext`, it the actual `RichManglingInfo` returned from `DemangleWithRichManglingInfo()` and handles lifetime and configuration. It is likely stack-allocated and can be reused for multiple queries during batch processing.
The idea here is that `DemangleWithRichManglingInfo()` acts like a gate keeper. It only provides access to `RichManglingInfo` on success, which in turn avoids the need to handle a `NoInfo` state in every single one of its getters. Having it stored within the context, avoids extra heap allocations and aids (3). As instantiations of the IPD the are considered expensive, the context is the ideal place to store it too. An efficient filtering function `SkipMangledNameFn` is another piece in the performance puzzle and it helps to mimic the original behavior of `InitNameIndexes`.
Future potential:
* `DemangleWithRichManglingInfo()` is thread-safe, IFF using different contexts in different threads. This may be exploited in the future. (It's another thing that it has in common with `LLVMContext`.)
* The old implementation only parsed and indexed Itanium mangled names. The new `RichManglingInfo` can be extended for various mangling schemes and languages.
One problem with the implementation of RichManglingInfo is the inaccessibility of class `CPlusPlusLanguage::MethodName` (defined in source/Plugins/Language/..), from within any header in the Core components of LLDB. The rather hacky solution is to store a type erased reference and cast it to the correct type on access in the cpp - see `RichManglingInfo::get<ParserT>()`. At the moment there seems to be no better way to do it. IMHO `CPlusPlusLanguage::MethodName` should be a top-level class in order to enable forward delcarations (but that is a rather big change I guess).
First simple profiling shows a good speedup. `target create clang` now takes 0.64s on average. Before the change I observed runtimes between 0.76s an 1.01s. This is still no bulletproof data (I only ran it on one machine!), but it's a promising indicator I think.
Reviewers: labath, jingham, JDevlieghere, erik.pilkington
Subscribers: zturner, clayborg, mgorny, lldb-commits
Differential Revision: https://reviews.llvm.org/D50071
llvm-svn: 339291
These three classes have no external dependencies, but they are used
from various low-level APIs. Moving them down to Utility improves
overall code layering (although it still does not break any particular
dependency completely).
The XCode project will need to be updated after this change.
Differential Revision: https://reviews.llvm.org/D49740
llvm-svn: 339127
Summary: `IsEmpty()` and `operator bool() == false` have equal semantics. Usage in Mangled::GetDemangledName() was incorrect. What it actually wants is a check for null-string. Split this off of D50071 and added a test to clarify usage.
Reviewers: labath, jingham
Subscribers: erik.pilkington, lldb-commits
Differential Revision: https://reviews.llvm.org/D50327
llvm-svn: 339014
Summary:
This patch allows LLDB's Stream class to count the bytes it has written to so far.
There are two major motivations for this patch:
The first one is that this will allow us to get rid of all the handwritten byte counting code
we have in LLDB so far. Examples for this are pretty much all functions in LLDB that
take a Stream to write to and return a size_t, which usually represents the bytes written.
By moving to this centralized byte counting mechanism, we hopefully can avoid some
tricky errors that happen when some code forgets to count the written bytes while
writing something to a stream.
The second motivation is that this is needed for the migration away from LLDB's `Stream`
and towards LLVM's `raw_ostream`. My current plan is to start offering a fake raw_ostream
class that just forwards to a LLDB Stream.
However, for this raw_ostream wrapper we need to fulfill the raw_ostream interface with
LLDB's Stream, which currently lacks the ability to count the bytes written so far (which
raw_ostream exposes by it's `tell()` method). By adding this functionality it is trivial to start
rolling out our raw_ostream wrapper (and then eventually completely move to raw_ostream).
Also, once this fake raw_ostream is available, we can start replacing our own code writing
to LLDB's Stream by LLVM code writing to raw_ostream. The best example for this is the
LEB128 encoding we currently ship, which can be replaced with by LLVM's version which
accepts an raw_ostream.
From the point of view of the pure source changes this test does, we essentially just renamed
the Write implementation in Stream to `WriteImpl` while the `Write` method everyone is using
to write its raw bytes is now just forwarding and counting the written bytes.
Reviewers: labath, davide
Reviewed By: labath
Subscribers: JDevlieghere, lldb-commits
Differential Revision: https://reviews.llvm.org/D50159
llvm-svn: 338733
Summary: In order to exploit the potential of LLVM's new ItaniumPartialDemangler for indexing in LLDB, we expect conceptual changes in the implementation of the InitNameIndexes function. Here is a unit test that aims at covering all relevant code paths in that function.
Reviewers: labath, jingham, JDevlieghere
Reviewed By: JDevlieghere
Subscribers: friss, teemperor, davide, clayborg, erik.pilkington, lldb-commits, mgorny
Differential Revision: https://reviews.llvm.org/D49909
llvm-svn: 338695
Summary:
This patch adds syntax highlighting support to LLDB. When enabled (and lldb is allowed
to use colors), printed source code is annotated with the ANSI color escape sequences.
So far we have only one highlighter which is based on Clang and is responsible for all
languages that are supported by Clang. It essentially just runs the raw lexer over the input
and then surrounds the specific tokens with the configured escape sequences.
Reviewers: zturner, davide
Reviewed By: davide
Subscribers: labath, teemperor, llvm-commits, mgorny, lldb-commits
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D49334
llvm-svn: 338662
Summary:
When I added the Stream unit test (r338488), the build bots failed due to an out-of-
bound reads when passing an empty string to the PutCStringAsRawHex8 method.
In r338491 I removed the test case to fix the bots.
This patch fixes this in PutCStringAsRawHex8 by always checking for the terminating
null character in the given string (instead of skipping it the first time). It also re-adds the
test case I removed.
Reviewers: vsk
Reviewed By: vsk
Subscribers: vsk, lldb-commits
Differential Revision: https://reviews.llvm.org/D50149
llvm-svn: 338637
The suspicious behavior is obviously because this method reads
OOB memory, so I'll remove it for now and re-add the test alongside
the fix later.
llvm-svn: 338491
Summary:
This adds an initial small unit test for LLDB's Stream class, which should at least cover
most of the functions in the Stream class. StreamString is always in big endian
mode, so that's the only stream byte order path this test covers as of now. Also,
the binary mode still needs to be tested for all print methods.
Also adds some FIXMEs for wrong/strange result values of the Stream class that we hit
while testing those functions.
Reviewers: labath
Reviewed By: labath
Subscribers: probinson, labath, mgorny, lldb-commits
Differential Revision: https://reviews.llvm.org/D50027
llvm-svn: 338488
Summary:
We currently allow any completion handler to read and manipulate the list of matches we
calculated so far. This leads to a few problems:
Firstly, a completion handler's logic can now depend on previously calculated results
by another handlers. No completion handler should have such an implicit dependency,
but the current API makes it likely that this could happen (or already happens). Especially
the fact that some completion handler deleted all previously calculated results can mess
things up right now.
Secondly, all completion handlers have knowledge about our internal data structures with
this API. This makes refactoring this internal data structure much harder than it should be.
Especially planned changes like the support of descriptions for completions are currently
giant patches because we have to refactor every single completion handler.
This patch narrows the contract the CompletionRequest has with the different handlers to:
1. A handler can suggest a completion.
2. A handler can ask how many suggestions we already have.
Point 2 obviously means we still have a dependency left between the different handlers, but
getting rid of this is too large to just append it to this patch.
Otherwise this patch just completely hides the internal StringList to the different handlers.
The CompletionRequest API now also ensures that the list of completions is unique and we
don't suggest the same value multiple times to the user. This property has been so far only
been ensured by the `Option` handler, but is now applied globally. This is part of this patch
as the OptionHandler is no longer able to implement this functionality itself.
Reviewers: jingham, davide, labath
Reviewed By: davide
Subscribers: lldb-commits
Differential Revision: https://reviews.llvm.org/D49322
llvm-svn: 338151
Summary:
Replace the existing combination of FastDemangle and the fallback to llvm::itaniumDemangle() with LLVM's new ItaniumPartialDemangler. It slightly reduces complexity and slightly improves performance, but doesn't introduce conceptual changes. This patch is preparing for more fundamental improvements on LLDB's demangling approach.
Reviewers: friss, jingham, erik.pilkington, labath, clayborg, mgorny, davide, JDevlieghere
Reviewed By: JDevlieghere
Subscribers: teemperor, JDevlieghere, labath, clayborg, davide, lldb-commits, mgorny, erik.pilkington
Differential Revision: https://reviews.llvm.org/D49612
llvm-svn: 337931
Aleksandr Urakov