Normally variables that have a named struct as a type will get a
typedef entry as their type, sometimes however the Go linker will
decide to use the DW_TAG_structure_type entry instead.
For consistency always wrap a struct type into a typedef when we are
creating a new variables (see comment in newVariable for exceptions).
This fixes a bug where it would be impossible to call methods on a
global variable.
Evaluates var.method expressions into a variable holding the
corresponding method with the receiver variable as a child, in
preparation for extending CallFunction so that it can call methods.
Changes (*Variable).setValue so that it can be used in CallFunction to
set up the argument frame for the function call, adding the ability to:
- nil nillable types
- set strings to the empty string
- copy from one structure to another (including strings and slices)
- convert any interface type to interface{}
- convert pointer shaped types (map, chan, pointers, and structs
consisting of a single pointer field) to interface{}
This covers all cases where an assignment statement can be evaluated
without allocating memory or calling functions in the target process.
Adds -defer flag to the stack command that decorates the stack traces
by associating each stack frame with its deferred calls.
Reworks proc.next to use this feature instead of using proc.DeferPC,
laying the groundwork to implement #1240.
Implements the function call injection protocol introduced in go 1.11
by https://go-review.googlesource.com/c/go/+/109699.
This is only the basic support, see TODO comments in pkg/proc/fncall.go
for a list of missing features.
Updates #119
This pull request makes several changes to delve to allow headless
instancess that are started with the --accept-multiclient flag to
keep running even if there is no connected client. Specifically:
1. Makes a headless instance started with --accept-multiclient quit
after one of the clients sends a Detach request (previously they
would never ever quit, which was a bug).
2. Changes proc/gdbserial and proc/native so that they mark the
Process as exited after they detach, even if they did not kill the
process during detach. This prevents bugs such as #1231 where we
attempt to manipulate a target process after we detached from it.
3. On non --accept-multiclient instances do not kill the target
process unless we started it or the client specifically requests
it (previously if the client did not Detach before closing the
connection we would kill the target process unconditionally)
4. Add a -c option to the quit command that detaches from the
headless server after restarting the target.
5. Change terminal so that, when attached to --accept-multiclient,
pressing ^C will prompt the user to either disconnect from the
server or pause the target process. Also extend the exit prompt to
ask if the user wants to keep the headless server running.
Implements #245, #952, #1159, #1231
Implements structured logging via Logrus. This gives us a logger per
boundry that we care about, allowing for easier parsing of logs if users
have more than one log option enabled. Also, cleans up a lot of
conditionals in the code by simply silencing the logger at creation as
opposed to conditionally logging everywhere.
If the application being debugged imports two packages with the same
name (but different paths) there was no way to disambiguate the two,
since the character '/' can not appear inside a go identifier.
By allowing users to use a string literal as the package name a package
path can be specified.
Displays the return values of the current function when we step out of
it after executing a step, next or stepout command.
Implementation of this feature is tricky: when the function has
returned the return variables are not in scope anymore. Implementing
this feature requires evaluating variables that are out of scope, using
a stack frame that doesn't exist anymore.
We can't calculate the address of these variables when the
next/step/stepout command is initiated either, because between that
point and the time where the stepout breakpoint is actually hit the
goroutine stack could grow and be moved to a different memory address.
Change the linux verison of proc/native and proc/gdbserial (with
debugserver) so that they let the target process use the terminal when
delve is launched in headless mode.
Windows already worked, proc/gdbserial (with rr) already worked.
I couldn't find a way to make proc/gdbserial (with lldb-server) work.
No tests are added because I can't think of a way to test for
foregroundness of a process.
Fixes#65
debug_info entries can use DW_AT_abstract_origin to inherit the
attributes of another entry, supporting this attribute is necessary to
support DW_TAG_inlined_subroutine.
Go, starting with 1.10, emits DW_TAG_inlined_subroutine entries when
inlining is enabled.
Much like the bug in issue #1031 and commit
f6f6f0bf13e4c708cb501202b83a6327a0f00e31 pointers can also escape to
the heap and then have a zero address (and no children) when we
autodereference.
1. Mark autodereferenced escaped variables with a 0 address as
unreadable.
2. Add guards to the pretty printers for unsafe.Pointer and pointers.
Fixes#1075
Replace the unsafe.Pointer type of the buf field of channels with the
appropriate array type, allow expressions accessing member field of the
channel struct.
Fixes#962
* string to []rune
* string to []byte
* []rune to string
* []byte to string
* any pointer to uintptr
The string, []rune, []byte conversion pairs aligns this to the go
language.
The pointer -> uintptr conversion pair is symmetric to the uintptr ->
pointer that we already have.
Also lets the user specify any size for byte array types instead of
just the ones already used by the program, this can be used to read
arbitrary memory.
Fixes#548, #867
Before go1.9 embedded struct fields had name == "" in runtime and ==
type name in DWARF. After go1.9 both runtime and DWARF use a simplified
version of the type as name.
Embedded structs are distinguished from normal fields by setting a flag
in the runtime.structfield, for runtime, and by adding a custom
attribute in DWARF.
If the user tries to list the contents of a function pointer but
forgets the '*' operator the location lookup will fail and result in a
unhelpful "location not found" error.
Instead if the location lookup fails we should try interpreting the
locspec as if it was preceded by '*'.
If one of the expressions that are automatically evaluated when a
breakpoint is hit can't be evaluated breakpoint information collection
should continue and the error should be returned for that specific
expression instead of the whole command.
The concrete value of an interface is always stored as a pointer inside
an interface variable. So far we have followed the memory layout and
reported the type of the 'data' attribute of interfaces as a pointer,
however this makes it impossible to distinguish interfaces with
concrete value of type 'A' from interfaces of concrete value of type
'*A'.
With this changeset when we autodereference pointers when the concrete
type of an interface is not a pointer.
* Fix various issues detected by megacheck
I've ran honnef.co/go/tools/cmd/megacheck and fixed a few of the
things that came up there.
* Cleanup using Gogland
A next/step/stepout command could hit a normal breakpoint, decorated
with a list of variables to evaluate, if that happens the variable
should be evaluated just as if the breakpoint was hit by a continue.
- moved target.Interface into proc as proc.Process
- rename proc.IThread to proc.Thread
- replaced interfaces DisassembleInfo, Continuable and
EvalScopeConvertible with Process.
- removed superfluous Gdbserver prefix from types in the gdbserial
backend.
- removed superfluous Core prefix from types in the core backend.
According to https://golang.org/cmd/go/#hdr-Test_packages
service_test is more appropriate becuase this directory contains
no non-test code and the intention is to compile these *_test.go
files as a separate package and link/run with the main test package.
* proc: Refactor stackIterator to use memoryReadWriter and BinaryInfo
* proc: refactor EvalScope to use memoryReadWriter and BinaryInfo
* proc: refactor Disassemble to use memoryReadWriter and BinaryInfo
On Windows we can sometimes encounter threads stopped in locations for
which we do not have entries in debug_frame.
These cases seem to be due to calls to Windows API in the go runtime,
we can still produce a (partial) stack trace in this circumstance by
following frame pointers (starting with BP).
We still prefer debug_frame entries when available since go functions
do not have frame pointers before go1.8.
We are already doing this in GoroutinesInfo we should be doing it for
GetG. The main consequence of not doing this is that the CurrentLoc of
DebuggerState.SelectedGoroutine is out of date compared to the location
of the thread running it.
* dwarf/line: bugfix: not all values of the state machine can be used
According to DWARF Version 3 Section 6.2 "Line Number Information" not
all the values transversed by the line numbers state machine are valid
instructions, only the ones after a "special opcode", after the
standard opcode DW_LNS_copy and the extended opcode
DW_LINE_end_sequence.
DWARF3 describes this by specifying that only the opcodes listed above
"append a row to the matrix".
Additionally the implementation of DW_LNS_const_add_pc was wrong.
Fixes#664
* dwarf/line: fixed test failing with go1.8
* service/test: fix prologue detection tests
The conditions about which function prologue is emitted by the compiler
changed in go1.8, changed the test program so that callme2 will still
have a prologue under go1.8.
* service/test: fix step test
compilation units are linked in a different order under go1.8 so the
code of 'fmt' is no longer located after 'main' in the executable,
changed the tests so that they don't rely on this assumption anymore.
* proc: change runtime.Breakpoint support for go1.8
Before 1.8 it was sufficient to step twice to exit a
runtime.Breakpoint(), but go 1.8 added frame pointer tracking to small
functions making runtime.Breakpoint longer.
This changes runtime.Breakpoint handling in Continue to single step as
many times as are needed to exit runtime.Breakpoint.
* proc/test: fix TestIssue561 for go1.8
* service/debugger: Restore breakpoints using file:line on restart
Restoring by address can cause the breakpoint to be inserted in the
middle of an instruction if the executable file has changed.
* terminal: Warn of stale executable when printing source
* proc: Add `wd` to Launch
This change adds the `wd` arg which specify working directory of the
program.
Fixes#295
* service/debugger: Add `Wd` field to debugger.Config
This change adds the `Wd` field which specify working directory of the
program launched by debugger.
Fixes#295
* service: Add `Wd` to service.Config
This change adds the `Wd` field which specify working directory of the
program debugger will launch.
Fixes#295
* cmd/dlv: Add `Wd` flag
This change adds `Wd` flag which specify working directory of the
program which launched by debugger.
Fixes#295
* only set the Linux working directory if it is set,
stub out param in darwin and windows
* set working directory for Windows
https://godoc.org/golang.org/x/sys/windows#CreateProcesshttps://msdn.microsoft.com/en-us/library/windows/desktop/ms682425(v=vs.85).aspx
* Windows workingDir must be an *uint16
* attempt to chdir on darwin via @yuntan
* proc/exec_darwin.c: fix working directory for darwin
* Add tests to check if working directory works.
* Fix darwin implementation of fork/exec, which paniced if
child fork returned.
* cmd, service: rename Wd to WorkingDir
Generate names of the concrete types stored inside interface variables
by fully parsing their runtime._type instead of simply using the str
field.
This allows delve to read the contents of an interface variable when
the program imports multiple packages that have the same name. It also
allows delve to correctly interpret some complex anonymous types.
Fixes#455
If the location specification matches the name of a function exactly
return that function as a match event if the expression matches other
functions as well.
Without this some functions, like math/rand.Intn are unmatchable.
This version preserves the order of requests, allows the
client to switch between API versions and introduces a
way to send notifications to the client (see TODO item at:
proc/proc_linux.go:325).
Fixes#523, #571
* tests: update to cope with go1.7 SSA compiler
* de-vendored golang.org/x/debug/dwarf
We need our own tweaked version
* dwarf/debug/dwarf: always use the entry's name attribute
Using the name attribute leads to better type names as well as fixes
inconsistencies between 1.5, 1.6 and 1.7.
* proc: Updated loadInterface to work with go1.7
go1.7 changed the internal representation of types, removing the string
field from runtime._type.
Updated loadInterface to use the new str field.
* service/api: Removed unused fields of service/api.Function
* proc/eval: Set return variable name to input expression
* all: fine-grained control of loadValue for better variable printing
Makes proc.(*Variable).loadValue loading parameters configurable
through one extra argument of type LoadConfig.
This interface is also exposed through the API so clients can control
how much of a variable delve should read.
New API version with better backwards compatibility plus mechanism to
select the API version that a headless instance should use.
Adds service/test/cmd/typecheckrpc.go to type check the RPC interface.
proc.(*Process) methods are not thread safe, multiple clients
connecting simultaneously to a delve server (Issue #383) or a even
a single over-eager client (Issue #408) can easily crash it.
Additionally (Issue #419) calls to client.(*RPCClient).Halt can
crash the server because they can result in calling the function
debug/dwarf.(*Data).Type simultaneously in multiple threads which
will cause it to return incompletely parsed dwarf.Type values.
Fixes#408, #419 (partial)
- Unlike FunctionEntryToFirstLine can skip the prologue on functions
that are defined on a single line, either because they weren't
formatted or because they were autogenerated
- Can skip the prologue on most functions when setting a breakpoint
with the filename:line syntax
Fixes#396
Past the maximum recursion depth maps shouldn't be loaded at all,
adding map children and not loading them breaks assumptions in
the prettyprinter.
Fixes#406
Typedefs that resolve to slices are not recorded in DWARF as typedefs
but instead as structs in a way that there is no way to know they
are really slices using debug/dwarf.
Using golang.org/x/debug/dwarf instead this problem is solved and
as a bonus some types are printed with a nicer names: (struct string
→ string, struct []int → []int, etc)
Fixes#356 and #293
Location specifiers starting with '*' can be followed by any
expression supported by the evaluator.
The expression should evaluate to either an integer (which will be
interpreted as an address) or to a function pointer (which will be
dereferenced to get the function's entry point).
The concrete type of an interface only contains the abbreviated
package name, we must construct a map from package names to package
paths to be able to resolve the concrete type of an interface.
packagename.SomeFunction should match
github.com/someuser/packagename.SomeFunction since the former is
the familiar syntax.
To disambiguate between io.SomeFunction and
github.com/someuser/somepackage/io.SomeFunction specify one extra
slash at the start of the location specifier: /io.SomeFunction.
Fixes Issue #296
Supported operators:
- All (binary and unary) operators between basic types except <-,
++ and -- (includes & to take the address of an expression)
- Comparison operators between supported compound types
- Typecast of integer constants into pointer types
- struct members
- indexing of arrays, slices and strings
- slicing of arrays, slices and strings
- pointer dereferencing
- true, false and nil constants
Implements #116, #117 and #251
Instead of trying to be clever and make an 'educated guess' as to where
the flow of control may go next, simple do the more naive, yet correct,
approach of setting a breakpoint everywhere we can in the function and
seeing where we end up. On top of this we were already setting a
breakpoint at the return address and deferred functions, so that remains
the same.
This removes a lot of gnarly, hard to maintain code and takes all the
guesswork out of this command.
Fixes#281
Breakpoints, tracepoints, etc.. take a location spec as input. This
patch improves the expressiveness of that API. It allows:
* Breakpoint at line
* Breakpoint at function (handling package / receiver smoothing)
* Breakpoint at address
* Breakpoint at file:line
* Setting breakpoint based off regexp
the entry point of a function is the beginning of the prologue, which can be run multiple times for each invocation of a function if the stack needs to be expanded or the scheduler needs to be run.
Instead of maintaining two separate client / server implementations,
maintain only the more lightweight JSON-RPC service. The reasoning
behind the merging of the original HTTP service was ease of tooling, in
other words low barrier of entry for external clients (editor
integrations, etc...).
I believe the JSON-RPC solution still satisfies that constraint while
have the advantage of being a more lightweight solution. HTTP, while
highly supported in most modern languages, carries with it too many
features we would never take advantage of. The RPC architecture seems
a more natural approach.
The infrastructure set up during the initial HTTP service implementation
was leveraged in the JSON-RPC implementation, so if any of those
original authors are reading this commit message: thank you for that
work, it was not in vain even if though the original HTTP service is not
being removed.
