This should fix a test failure in TestScriptedFrameProvider.py:
https://lab.llvm.org/buildbot/#/builders/18/builds/23398/steps/6/logs/stdio
This is a happening because on 32bit system, addresses don't have the
leading zeroes. This patch removes them to satisfy the checks.
Signed-off-by: Med Ismail Bennani <ismail@bennani.ma>
Majority of the time users are less interested on the memory address of
a type. It is mostly useful for pointer types (the memory address is
shown).
It makes the view more bloated without adding useful information.
can always fall back to the debug console or watch pane to view the
information if necessary.
This patch refactors the way we check for the windows version in the
`@skipIfWindows` decorator.
The new logic reuses the `expectedCompilerVersion` method logic for the
parsing and comparison of the version.
This patch is a follow-up to 96c733e to fix a missing space in the
frame.pc format entity. This space was intended to be prepended to the
module format entity scope but if the module is not valid, which is
often the case for python pc-less scripted frames, the space between the
pc and the function name is missing.
Signed-off-by: Med Ismail Bennani <ismail@bennani.ma>
A test failure on green dragon shows the ivars with unexpected values.
This makes the test us an explicit `+new` instead of `+alloc` (which is
missing an `-init` call).
Because PDB doesn't know about templates, we need to get to `T` of
`std::atomic<T>` differently. The type includes the `value_type`
typedef, which is always equal to `T`. The native PDB plugin includes
this since #169248.
Then we can run the `std::atomic` test with (native) PDB.
Since we updated our buildbot setup, these have been failing. Ignore
them until we have time to find the real problem, which is something to
do with failing to backtrace, or missing debug info when we do.
Implementation files using the Intel syntax typically explicitly specify it.
Do the same for the few files using AT&T syntax.
This enables building LLVM with `-mllvm -x86-asm-syntax=intel` in one's Clang config files
(i.e. a global preference for Intel syntax).
This reverts commit 13b4eb9452d37106b1143723e658010a9b58d344 since it
doesn't fix the test failure.
Signed-off-by: Med Ismail Bennani <ismail@bennani.ma>
This is necessary to get the function name in the test, following
20a6c59d8311d92bd8553b22b82a3874e0016edb.
Signed-off-by: Med Ismail Bennani <ismail@bennani.ma>
This patch implements a workaround for a VSCode bug that causes it to
send disassemble requests with empty memory reference. You can find more
detailed description
[here](https://github.com/microsoft/vscode/pull/270361). I propose to
allow empty memory reference and return invalid instructions when this
occurs.
Error log example:
```
1759923554.517830610 (stdio) --> {"command":"disassemble","arguments":{"memoryReference":"","offset":0,"instructionOffset":-50,"instructionCount":50,"resolveSymbols":true},"type":"request","seq":3}
1759923554.518007517 (stdio) queued (command=disassemble seq=3)
1759923554.518254757 (stdio) <-- {"body":{"error":{"format":"invalid arguments for request 'disassemble': malformed memory reference at arguments.memoryReference\n{\n \"instructionCount\": 50,\n \"instructionOffset\": -50,\n \"memoryReference\": /* error: malformed memory reference */ \"\",\n \"offset\": 0,\n \"resolveSymbols\": true\n}","id":3,"showUser":true}},"command":"disassemble","request_seq":3,"seq":0,"success":false,"type":"response"}
```
I am not sure that we should add workaround here when bug on VSCode
side, but I think this bug affects our users. WDYT?
This is a follow-on to:
https://github.com/llvm/llvm-project/pull/168611
which added a bunch of tests for detecting dynamic types of C++ result
variables. I don't actually know how well dynamic type detection works
on Windows if at all. It would require Windows support, since the
Linux/Darwin version is specific to the Itanium ABI.
When you run an expression and the result has a dynamic type that is
different from the expression's static result type, we print the result
variable using the dynamic type, but at present when you use the result
variable in an expression later on, we only give you access to the
static type. For instance:
```
(lldb) expr MakeADerivedReportABase()
(Derived *) $0 = 0x00000001007e93e0
(lldb) expr $0->method_from_derived()
^
error: no member named 'method_from_derived' in 'Base'
(lldb)
```
The static return type of that function is `Base *`, but we printed that
the result was a `Derived *` and then only used the `Base *` part of it
in subsequent expressions. That's not very helpful, and forces you to
guess and then cast the result types to their dynamic type in order to
be able to access the full type you were returned, which is
inconvenient.
This patch makes lldb retain the dynamic type of the result variable
(and ditto for persistent result variables).
It also adds more testing of expression result variables with various
types of dynamic values, to ensure we can access both the ivars and
methods of the type we print the result as.
Split off from #171489. This only adds the lookup of the active type for
a `std::variant` based on the head type (since PDB doesn't have template
info).
This commit leaves "b" aliased to the old _regexp-break for now. The two
variants are identical except that `_regexp-break` allows you to say:
`(lldb) b <unrecognized_input>
`
which gets translated to:
`break set <unrecognized_input>
`
So switching people to `_regexp-break-add` would be a surprising
behavior change. It would be wrong for `_regexp_break-add` have one
branch that call `break set`, so to avoid surprise, I'll add the command
and let people who are playing with `break add` instead of `break set`
can set the alias to the new one by hand for now.
This seems the standard way to get the path to such tools within LLVM.
Calling findBuiltClang() has some annoying behavior like falling back to
CC when it cannot find anything else, which might point to anything or
not even be set.
We noticed this with our internal build system as the lli binary is not
in the same path as the clang binary.
This patch converts the `jit-loader_rtdyld_elf.test` test from a Shell
test to an API test.
This test is timing out in CI on Windows and the hang cannot be
reproduced at desk. Converting it to an API test would allow us to
instrument it better in order to trace the failure.
We follow LLVM's style guide for diagnostics, which instructs to start
the first sentence with a lowercase letter, and finish the last sentence
without a period, if it would end in one otherwise.
Remove `CommandReturnObject::AppendRawError` and replace its two uses
with `AppendError`, which correctly prefixes the message with `error:`.
The comment for the method is outdated and the prefixing is clearly
desired in both situations.
Similar to the other PRs, this runs the `std::optional` test with PDB.
Since we don't know that variables use typedefs, we check for the full
name when testing PDB.
Since PDB doesn't have template information, we need to get the element
type from somewhere else. I'm using the type of `_Myval` in a list node,
which holds the element type.
Runs the `std::shared/unique_ptr` tests with PDB with two changes:
- PDB uses the "full" name, so `std::string` is `std::basic_string<char,
std::char_traits<char>, std::allocator<char>>`
- The type of the pointer inside the shared/unique_ptr isn't the
`element_type` typedef
Basic API tests to check how template aliases are rendered by LLDB
(using both `DW_TAG_template_alias` and `DW_TAG_typedef`, with and
without `-gsimple-template-names`).
We used to set it to `true` up until recently, see
[here](https://github.com/llvm/llvm-project/pull/170802). That's
incorrect because `SuppressInlineNamespace` is actually an enum. What
probably happened is that `SuppressInlineNamespace` used to be a boolean
but got turned into an enum. But the assignment in LLDB wasn't updated.
But because the bitfield is an `unsigned`, the compiler never
complained.
This meant that ever since `SuppressInlineNamespace` became an enum,
we've been setting it to `SuppressInlineNamespaceMode::Redundant`. Which
means we would only omit the inline namespace when displaying typenames
if Clang deemed it unambiguous. This is probably a rare situtation but
the attached test-case is one such scenario. Here, `target var t1`
followed by `target var t2` would print the inline namespace for `t2`,
because in that context, the type is otherwise ambiguous. But because
LLDB's context is lazily constructed, evaluating `t2` first would omit
the inline namespace, because `t1` isn't in the context yet to make it
ambiguous.
This patch sets the `SuppressInlineNamespace` to
`SuppressInlineNamespaceMode::All`, which is most likely what was
intended in the first place, and also removes the above-mentioned
non-determinism from our typename printing.
lldbutil.run_to_line_breakpoint had usages that set column breakpoints,
so I thought there was coverage of that on the command-line, but
actually all the `run_to` utilities use the SB API's, and there weren't
any tests of setting file line & column breakpoint through
`run_break_set`. So I missed that I had typed the column option `c` -
that's taken by `--command`.
This patch fixes that typo and adds a CLI test for file + line + column.
Scripted frames that materialize Python functions or other non-native
code are PC-less by design, meaning they don't have valid program
counter values. Previously, these frames would display invalid addresses
(`0xffffffffffffffff`) in backtrace output.
This patch updates `FormatEntity` to detect and suppress invalid address
display for PC-less frames, adds fallback to frame methods when symbol
context is unavailable, and modifies `StackFrame::GetSymbolContext` to
skip PC-based symbol resolution for invalid addresses.
The changes enable PC-less frames to display cleanly with proper
function names, file paths, and line numbers, and allow for source
display of foreign sources (like Python). Includes comprehensive test
coverage demonstrating frames pointing to Python source files.
Signed-off-by: Med Ismail Bennani <ismail@bennani.ma>
Someone came up with a clever idea for a new breakpoint type, but we
couldn't figure out how to add it in an ergonomic way because
`breakpoint set` has used up all the short-option characters. And even
if they did find a way to add it, the help for `break set` is so
confusing - because of the way it is implemented - that very few people
would detect the addition.
The basic problem is that `break set` distinguishes amongst the
fundamental breakpoint types it offers by which "required option" you
provide. If you pass a `-a` you are setting an address breakpoint, if
`-n`, `-F`, etc. a symbol name based breakpoint. And so forth. That is
however pretty hard to discern from the option grouping printing from
`help break set`. `break set` also suffers from the problem that it uses
common options in different ways depending on which "required" option is
present, which makes documenting the various behaviors difficult. And as
we run out of single letters it makes extending it difficult to
impossible.
This PR fixes that situation by adding a new command for adding
breakpoints - `break add`. The new command specifies the "breakpoint
types" as subcommands of `break add` rather than distinguishing them by
their being one of the "required" options the way `break set` does. That
both makes it much clearer what the breakpoint types actually are, and
means that the option set can be dedicated to that particular breakpoint
type, and so the help for each is less cluttered, and can be documented
properly for each usage.
Instead of trying to parse the meaning of:
```
(lldb) help break set
Sets a breakpoint or set of breakpoints in the executable.
Syntax: breakpoint set <cmd-options>
Command Options Usage:
breakpoint set [-DHd] -l <linenum> [-G <boolean>] [-C <command>] [-c <expr>] [-Y <source-language>] [-i <count>] [-o <boolean>] [-q <queue-name>] [-t <thread-id>] [-x <thread-index>] [-T <thread-name>] [-R <address>] [-N <breakpoint-name>] [-u <column>] [-f <filename>] [-m <boolean>] [-s <shlib-name>] [-K <boolean>]
breakpoint set [-DHd] -a <address-expression> [-G <boolean>] [-C <command>] [-c <expr>] [-Y <source-language>] [-i <count>] [-o <boolean>] [-q <queue-name>] [-t <thread-id>] [-x <thread-index>] [-T <thread-name>] [-N <breakpoint-name>] [-s <shlib-name>]
breakpoint set [-DHd] -n <function-name> [-G <boolean>] [-C <command>] [-c <expr>] [-Y <source-language>] [-i <count>] [-o <boolean>] [-q <queue-name>] [-t <thread-id>] [-x <thread-index>] [-T <thread-name>] [-R <address>] [-N <breakpoint-name>] [-f <filename>] [-L <source-language>] [-s <shlib-name>] [-K <boolean>]
breakpoint set [-DHd] -F <fullname> [-G <boolean>] [-C <command>] [-c <expr>] [-Y <source-language>] [-i <count>] [-o <boolean>] [-q <queue-name>] [-t <thread-id>] [-x <thread-index>] [-T <thread-name>] [-R <address>] [-N <breakpoint-name>] [-f <filename>] [-L <source-language>] [-s <shlib-name>] [-K <boolean>]
breakpoint set [-DHd] -S <selector> [-G <boolean>] [-C <command>] [-c <expr>] [-Y <source-language>] [-i <count>] [-o <boolean>] [-q <queue-name>] [-t <thread-id>] [-x <thread-index>] [-T <thread-name>] [-R <address>] [-N <breakpoint-name>] [-f <filename>] [-L <source-language>] [-s <shlib-name>] [-K <boolean>]
breakpoint set [-DHd] -M <method> [-G <boolean>] [-C <command>] [-c <expr>] [-Y <source-language>] [-i <count>] [-o <boolean>] [-q <queue-name>] [-t <thread-id>] [-x <thread-index>] [-T <thread-name>] [-R <address>] [-N <breakpoint-name>] [-f <filename>] [-L <source-language>] [-s <shlib-name>] [-K <boolean>]
breakpoint set [-DHd] -r <regular-expression> [-G <boolean>] [-C <command>] [-c <expr>] [-Y <source-language>] [-i <count>] [-o <boolean>] [-q <queue-name>] [-t <thread-id>] [-x <thread-index>] [-T <thread-name>] [-R <address>] [-N <breakpoint-name>] [-f <filename>] [-L <source-language>] [-s <shlib-name>] [-K <boolean>]
breakpoint set [-DHd] -b <function-name> [-G <boolean>] [-C <command>] [-c <expr>] [-Y <source-language>] [-i <count>] [-o <boolean>] [-q <queue-name>] [-t <thread-id>] [-x <thread-index>] [-T <thread-name>] [-R <address>] [-N <breakpoint-name>] [-f <filename>] [-L <source-language>] [-s <shlib-name>] [-K <boolean>]
breakpoint set [-ADHd] -p <regular-expression> [-G <boolean>] [-C <command>] [-c <expr>] [-Y <source-language>] [-i <count>] [-o <boolean>] [-q <queue-name>] [-t <thread-id>] [-x <thread-index>] [-T <thread-name>] [-N <breakpoint-name>] [-f <filename>] [-m <boolean>] [-s <shlib-name>] [-X <function-name>]
breakpoint set [-DHd] -E <source-language> [-G <boolean>] [-C <command>] [-c <expr>] [-Y <source-language>] [-i <count>] [-o <boolean>] [-q <queue-name>] [-t <thread-id>] [-x <thread-index>] [-T <thread-name>] [-N <breakpoint-name>] [-h <boolean>] [-w <boolean>]
breakpoint set [-DHd] -P <python-class> [-k <none>] [-v <none>] [-G <boolean>] [-C <command>] [-c <expr>] [-Y <source-language>] [-i <count>] [-o <boolean>] [-q <queue-name>] [-t <thread-id>] [-x <thread-index>] [-T <thread-name>] [-N <breakpoint-name>] [-f <filename>] [-s <shlib-name>]
breakpoint set [-DHd] -y <linespec> [-G <boolean>] [-C <command>] [-c <expr>] [-Y <source-language>] [-i <count>] [-o <boolean>] [-q <queue-name>] [-t <thread-id>] [-x <thread-index>] [-T <thread-name>] [-R <address>] [-N <breakpoint-name>] [-m <boolean>] [-s <shlib-name>] [-K <boolean>]
```
We instead offer:
```
(lldb) help break add
Commands to add breakpoints of various types
Syntax: breakpoint add
Access the breakpoint search kernels built into lldb. Along with specifying the
search kernel, each breakpoint add operation can specify a common set of
"reaction" options for each breakpoint. The reaction options can also be
modified after breakpoint creation using the "breakpoint modify" command.
The following subcommands are supported:
address -- Add breakpoints by raw address
exception -- Add breakpoints on language exceptions. If no language is specified, break on exceptions for all supported languages
file -- Add breakpoints on lines in specified source files
name -- Add breakpoints matching function or symbol names
pattern -- Add breakpoints matching patterns in the source text Expects 'raw' input (see 'help raw-input'.)
scripted -- Add breakpoints using a scripted breakpoint resolver.
For more help on any particular subcommand, type 'help <command> <subcommand>'.
```
The individual subcommand helps are also easier to read. They are still
a little too verbose because they all repeat the options for the
`reactions`. A general fix for our help system would be when a command
incorporates an OptionGroup whole into the command options, help would
show the option group name - which you could separately look up. But
even without that:
```
(lldb) help br a a
Add breakpoints by raw address
Syntax: breakpoint add address <cmd-options> <address> [<address> [...]]
Command Options Usage:
breakpoint add address [-DHde] [-G <boolean>] [-C <command>] [-c <expr>] [-Y <source-language>] [-i <count>] [-o <boolean>] [-q <queue-name>] [-t <thread-id>] [-x <thread-index>] [-T <thread-name>] [-N <breakpoint-name>] [-s <shlib-name>] <address> [<address> [...]]
-C <command> ( --command <command> )
A command to run when the breakpoint is hit, can be provided more than once, the commands will be run in left-to-right order.
-D ( --dummy-breakpoints )
Act on Dummy breakpoints - i.e. breakpoints set before a file is provided, which prime new targets.
-G <boolean> ( --auto-continue <boolean> )
The breakpoint will auto-continue after running its commands.
-H ( --hardware )
Require the breakpoint to use hardware breakpoints.
-N <breakpoint-name> ( --breakpoint-name <breakpoint-name> )
Adds this name to the list of names for this breakpoint. Can be specified more than once.
-T <thread-name> ( --thread-name <thread-name> )
The breakpoint stops only for the thread whose thread name matches this argument.
-Y <source-language> ( --condition-language <source-language> )
Specifies the Language to use when executing the breakpoint's condition expression.
-c <expr> ( --condition <expr> )
The breakpoint stops only if this condition expression evaluates to true.
-d ( --disable )
Disable the breakpoint.
-e ( --enable )
Enable the breakpoint.
-i <count> ( --ignore-count <count> )
Set the number of times this breakpoint is skipped before stopping.
-o <boolean> ( --one-shot <boolean> )
The breakpoint is deleted the first time it stop causes a stop.
-q <queue-name> ( --queue-name <queue-name> )
The breakpoint stops only for threads in the queue whose name is given by this argument.
-s <shlib-name> ( --shlib <shlib-name> )
Set the breakpoint at an address relative to sections in this shared library.
-t <thread-id> ( --thread-id <thread-id> )
The breakpoint stops only for the thread whose TID matches this argument. The token 'current' resolves to the current thread's ID.
-x <thread-index> ( --thread-index <thread-index> )
The breakpoint stops only for the thread whose index matches this argument.
This command takes options and free-form arguments. If your arguments resemble option specifiers (i.e., they start with a - or --), you must use ' --
' between the end of the command options and the beginning of the arguments.
```
is pretty readable.
Scripted frames that materialize Python functions are PC-less by design,
meaning they don't have valid address ranges. Previously,
LineEntry::IsValid()
required both a valid address range and a line number, preventing
scripted
frames from creating valid line entries for these synthetic stack
frames.
Relaxing this requirement is necessary since
`SBSymbolContext::SetLineEntry`
will first check if the LineEntry is valid and discard it otherwise.
This change introduces an `synthetic` flag that gets set when LineEntry
objects are created or modified through the SBAPI (specifically via
SetLine).
When this flag is set, IsValid() no longer requires a valid address
range,
only a valid line number.
This is risk-free because the flag is only set for LineEntry objects
created
through the SBAPI, which are primarily used by scripted processes and
frames.
Regular debug information-derived line entries continue to require valid
address ranges.
Signed-off-by: Med Ismail Bennani <ismail@bennani.ma>
Some months ago, the LookupInfo constructor logic was refactored to not
depend on language specific logic, and use languages plugins instead. In
this refactor, when the language type is unknown, a single LookupInfo
object will handle multiple languages. This doesn't work well, as
multiple languages might want to configure the LookupInfo object in
different ways. For example, different languages might want to set the
m_lookup_name differently from each other, but the previous
implementation would pick the first name a language provided, and
effectively ignored every other language. Other fields of the LookupInfo
object are also configured in incompatible ways.
This approach doesn't seem to be a problem upstream, since only the
C++/Objective-C language plugins are available, but it broke downstream
on the Swift fork, as adding Swift to the list of default languages when
the language type is unknown breaks C++ tests.
This patch makes it so instead of building a single LookupInfo object
for multiple languages, one LookupInfo object is built per language
instead.
rdar://159531216
It looks like the providers don't get loaded on arm32 bots:
https://github.com/llvm/llvm-project/issues/170412
Skipping for now since I don't have access to a machine to investigate
it.
Signed-off-by: Med Ismail Bennani <ismail@bennani.ma>
This patch disables TestFrameProviderCircularDependency.py on Windows
since the scripted frame provider uses SBTarget.FindFunctions which
doesn't seem to be working (according to TestTargetAPI.test_find_functions).
Signed-off-by: Med Ismail Bennani <ismail@bennani.ma>
On ARM32, FixCodeAddress unconditionally clears bit 0 (the Thumb bit)
from all code addresses, including synthetic frame PCs. This causes
test failures where synthetic PCs like 0xFFFF and 0xDEADBEEF become
0xFFFE and 0xDEADBEEE respectively.
This adjusts the tests to expect the modified PC values on ARM32.
Signed-off-by: Med Ismail Bennani <ismail@bennani.ma>
Vector registers have synthetic values for display purposes. This causes
SBValue::GetExpressionPath to dispatch
to ValueObjectSynthetic instead of ValueObjectRegister, producing
incorrect results.
Fixes#147144
This patch tries to upstream code landed downstream in
https://github.com/swiftlang/llvm-project/pull/11835.
This patch implements an instrumentation plugin for the
`-fbounds-safety` soft trap mode first implemented in
https://github.com/swiftlang/llvm-project/pull/11645 (rdar://158088757).
That functionality isn't supported in upstream Clang yet, however the
instrumented plugin can be compiled without issue so this patch tries to
upstream it. The included tests are all disabled when the clang used for
testing doesn't support `-fbounds-safety`. This means the tests will be
skipped. However, it's fairly easy to point LLDB at a clang that does
support `-fbounds-safety. I've done this and confirmed the tests pass.
To use a custom clang the following can be done:
* For API tests set the `LLDB_TEST_COMPILER` CMake cache variable to
point to appropriate compiler.
* For shell tests applying a patch like this can be used to set the
appropriate compiler:
```
--- a/lldb/test/Shell/helper/toolchain.py
+++ b/lldb/test/Shell/helper/toolchain.py
@@ -271,6 +271,7 @@ def use_support_substitutions(config):
if config.lldb_lit_tools_dir:
additional_tool_dirs.append(config.lldb_lit_tools_dir)
+ config.environment['CLANG'] = '/path/to/clang'
llvm_config.use_clang(
```
The current implementation of -fbounds-safety traps works by emitting
calls to runtime functions intended to log the occurrence of a soft
trap.
While the user could just set a breakpoint of these functions the
instrumentation plugin sets it automatically and provides several
additional features:
When debug info is available:
* It adjusts the stop reason to be the reason for trapping. This is
extracted from the artificial frame in the debug info (similar to
-fbounds-safety hard traps).
* It adjusts the selected frame to be the frame where the soft trap
occurred.
When debug info is not available:
* For the `call-with-str` soft trap mode the soft trap reason is
read from the first argument register.
* For the `call-minimal` soft trap mode the stop reason is adjusted
to note its a bounds check failure but does not give further
information because none is available.
* In this situation the selected frame is not adjusted because in
this mode the user will be looking at assembly and adjusting the
frame makes things confusing.
This patch includes shell and api tests. The shell tests seemed like the
best way to test behavior when debug info is missing because those tests
make it easy to disable building with debug info completely.
rdar://163230807
Currently, disabling the statusline with `settings set show-statusline
false` leaves LLDB in a broken state. The same is true when trying to
toggle the setting again.
The issue was that setting the scroll window to 0 is apparently not
identical to setting it to the correct number of rows, even though some
documentation online incorrectly claims so.
Fixes#166608
This patch adds support for `dataBreakpointInfoBytes` capability from
DAP. You can test this feature in VSCode (`Add data breakpoint at
address` button in breakpoints tab).
