Commit Graph

9538 Commits

Author SHA1 Message Date
Todd Fiala 4dcb6aded2 gtest: modified Xcode integration so gtest assert/expect failure now shows error content in the error marker at the line of failure.
This change modifies the python test runner to combine lines starting with a
failure report including up to 3 more lines, terminated by the next
"[  FAILED  ]" line.  These are all emitted on the same line as the file:line
indication, which allows Xcode's failure marker code to pick it up and display
it along with the error badge in the Xcode editor window.  Makes for a nice
gtest development experience.

llvm-svn: 218518
2014-09-26 12:58:25 +00:00
Todd Fiala f3a3fc5120 gtest: remove recursive make, add python driver + Xcode error hook-up.
This change does the following:
* Removes the gtest/Makefile recursive-make-based calling strategy
  for gtest execution.
* Adds the gtest/do-gtest.py call script.
  - This handles finding and calling the Makefiles that really
    run tests.
  - This script also transforms the test output into something
    that Xcode can place a failure marker on when a test fails.
* Modifies the Xcode external build command target for gtest.
  It now calls the gtest/do-gtest.py script.

There is still room for improvement on Xcode integration of
do-gtest.py.  Essentially the next several lines of error reporting
from the gtest output should be coalesced into a single line so that
Xcode can tell more about the error directly in the editor.  Right now
it just puts a red mark and says "failure" but doesn't give any
details.

llvm-svn: 218470
2014-09-25 22:12:33 +00:00
Todd Fiala b3185e710e Fixup gtest layout, add Linux ThreadStateCoordinator.cpp.
This change does the following:
* Remove test/c++/...
* Add gtest.
* Add gtest/unittest directory for unittesting individual classes.
* Add an initial Plugins/Process?linux/ThreadStateCoordinatorTest.cpp.
  - currently failing a test (intentional).
  - added a bare-bones ThreadStateCoordinator.cpp to Plugins/Process/Linux,
    more soon. Just enough to prove out running gtest on Ubuntu and MacOSX.
* Added recursive make machinery so that doing a 'make' in gtest/ is
  sufficient to kick off the existing test several directories down.
  - Caveat - I currently short circuit from gtest/unittest/Makefile directly to
    the one and only gtest/unittest/Plugins/Process/Linux directory.  We'll need
    to add the intervening layers.  I haven't done this yet since to fix the
    Xcode test failure correspondence, I may need to add a python layer which
    might just handle the directory crawling.
* Added an Xcode project to the lldb workspace for gtest.
  - Runs the recursive make system in gtest/Makefile.
  - Default target is 'test'.  test and clean are supported.
  - Currently does not support test failure file/line correspondence.
    Requires a bit of text transformation to hook that up.

llvm-svn: 218460
2014-09-25 19:25:07 +00:00
Todd Fiala e15fcaa2e5 Moved PlatformDarwin::GetEnvironment() to PlatformPOSIX to fix Linux/llgs issue.
See http://reviews.llvm.org/D5486 for more details.

I was tracking a problem where llgs on linux would not pick up any environment variables. On OSX there is a virtual function PlatformDarwin::GetEnvironment() which correctly sets up the list of environment variables. On linux llgs it defaults to a base class default implementation which clears the list.

I moved the OSX implementation down to PlatformPOSIX. This fixes my problem on linux still works properly on OSX.

Change by Shawn Best.
Slight tweak to convert 'virtual' to 'override' in PlatformDarwin.h virtual method override by Todd.

Tested:
Ubuntu 14.04 x86_64, cmake/ninja build + tests.
MacOSX 10.9.5 x86_64, Xcode 6.1 Beta build + tests.

llvm-svn: 218424
2014-09-24 23:10:30 +00:00
Todd Fiala f502d72d4e The beginnings of a gtest-based test framework.
Makes use of LLVM gtest support.  This does *not* run as part of
the lldb test suite.

I'm using it to start testing some components that
I'll be adding to the inner guts of NativeThreadLinux to more
maintainably handle thread states and deferred thread state notification.

Runs with default Makefile target "test" using gmake within a given
test directory (currently only test/c++/native_process/thread_state_coordinator).

The Makefile.rules currently assume it is using the LLVM gtest.  It works on
a canonical MacOSX dir structture (i.e. lldb, lldb/llvm, lldb/llvm/tools/clang).
It also works on Ubuntu assuming the standard dir layout of llvm, llvm/tools/clang,
llvm/tools/lldb.  In this case, it expects a directory called build-debug parallel
to the llvm source dir.  All directory assumptions can be overridden with
environment variables.  See test/c++/make/Makefile.rules for details.

We'll want to make this smarter in the future, particularly around finding the LLVM build
output dir.

llvm-svn: 218422
2014-09-24 22:57:33 +00:00
Enrico Granata 5e35cf9dcc Add an explicit cast to silence compiler warning here
llvm-svn: 218410
2014-09-24 20:52:58 +00:00
Todd Fiala ad6eee6399 Change inline-breakpoint-strategy setting to default to eInlineBreakpointsAlways strategy.
See thread started here for motivation:
http://lists.cs.uiuc.edu/pipermail/lldb-dev/2014-September/005225.html

This change enables the ability to set breakpoints in ccache-based and executables that
make use of preprocessed source files.  This ability existed in lldb before, but was off
by default.

Change by Doug Snyder.

llvm-svn: 218405
2014-09-24 19:59:13 +00:00
Enrico Granata a540496e6f These APIs were defined in the .h file, but never actually implemented. Just get rid of them
llvm-svn: 218396
2014-09-24 17:57:11 +00:00
Todd Fiala d5635cdd7f Adjust to LLVM JIT API change
Those wrapper functions seems not used by lldb... Removed

Reference for

replacing JITMemoryManager with SectionMemoryManager
replacing "llvm::JITMemoryManager::CreateDefaultMemManager()" with "new llvm::SectionMemoryManager()"
http://llvm.org/viewvc/llvm-project/llvm/trunk/tools/llvm-jitlistener/llvm-jitlistener.cpp?r1=218316&r2=218315&pathrev=218316

Change by Tong Shen.

Tested:
Ubuntu 14.04 x86_64
MacOSX 10.9.5 with Xcode Version 6.1 (6A1030) (Beta)

llvm-svn: 218383
2014-09-24 15:55:47 +00:00
Zachary Turner acee96ae52 Fix up the HostThread interface, making the interface simpler.
Reviewed by: Greg Clayton
Differential Revision: http://reviews.llvm.org/D5417

llvm-svn: 218325
2014-09-23 18:32:09 +00:00
Kuba Brecka 9d28cf5609 Add ASan history threads into process_sp->GetExtendedThreadList, so they don't get freed too early
Reviewed at http://reviews.llvm.org/D5452

llvm-svn: 218323
2014-09-23 18:20:24 +00:00
Todd Fiala 5bc56a4b34 android: add a command script to help display mixed Java/native backtraces.
This script supports displaying developer-focused backtraces when working
with mixed Java and C/C++ stack frames within lldb.  On Android, this represents
just about every app, since all apps start in Java code.

The script currently supports the Art JVM when run on host-side x86_64 and x86,
but does require a patch not yet accepted in AOSP:
AOSP patch: https://android-review.googlesource.com/#/c/106523/

The backtraces will hide Art VM machinery for interpreted and AOT code
and display the Java file/line numbers for Java code, while displaying
native backtrace info for native frames.  Effectively the developer will
get an app-centric view of the call stack.

This script is not yet tested on device-side Art nor is it tested on
any architecture other than x86_64 or x86 32-bit.  Several changes were
needed on the AOSP side to enable it to work properly for x86_64 and x86, 
so it is quite likely we'll need to do something similar for other cpu
architectures as well.

Change by Tong Shen

llvm-svn: 218315
2014-09-23 16:28:01 +00:00
Jason Molenda f6a4258020 ConnectionFileDescriptor::Connect()'s handling of the "fd://" method
was broken in r214984 by the addition of an unconditional error
return at the start of the code block handling this method.  Remove
the errant lines.
<rdar://problem/18416691> 

llvm-svn: 218291
2014-09-23 02:43:35 +00:00
Greg Clayton 994ba64756 Played around with TK UI a bit this weekend.
If you "command script import" this file, then you will have two new commands:

(lldb) tk-variables
(lldb) tk-process

Not sure how this will work on all other systems, but on MacOSX, you will get a window with a tree view that allows you to inspect your local variables by expanding variables to see the child values.

The "tk-process" allows you to inspect the currently selected process by expanding the process to see the threads, the threads to see the frames, and the frames to see the variables. Very handy if you want to view variables for all frames simultaneously.

llvm-svn: 218279
2014-09-22 22:06:41 +00:00
Ed Maste da7c5b253f Enable libcxx map test on FreeBSD again
The test used to trigger an assertion failure "Cannot get layout of
forward declarations!", but it no longer fails when built with
Clang 3.4.1 (system compiler) or 3.5 from SVN on FreeBSD.

llvm.org/pr17231

llvm-svn: 218196
2014-09-21 02:24:01 +00:00
Todd Fiala 21130fbc04 Fix lldb-gdbserver build.
Build break change by Paul Osmialowski.

Minor changes to argument passing (converted unintentional pass-by-value to pass-by-ref) by Todd.

llvm-svn: 218186
2014-09-20 17:34:48 +00:00
Jason Molenda e59b0d2c48 Have CommandObject::CheckRequirements() report the largest missing
requirement for a command instead of the smallest.  e.g. if a command
requires a Target, Process, Thread, and Frame, and none of those
are available, report the largest -- Target -- as being missing
instead of the smallest -- Frame.

Patch by Paul Osmialowski.

llvm-svn: 218181
2014-09-20 09:14:31 +00:00
Jim Ingham 16ebdd6c1b Fix a thinko in how the RemoteiOS Platform looked up files in the SDK & other
platform locations.  We didn't always do an exhaustive search through all the 
platform locations, so we would have to read some files out of memory even though
they existed in the exploded shared cache or SDK.

<rdar://problem/18385947>

llvm-svn: 218157
2014-09-19 21:58:45 +00:00
Jim Ingham 03d1730aad We had to squirrel away the dyld module before doing ResolveExecutableModule, since
that would clear the module list, and then put it back by hand.  But we forgot to 
also put its sections back in the target SectionList, so we would jettison it as
unloaded when we finished handling the first real load event.  Add its sections.

<rdar://problem/18385947>

llvm-svn: 218156
2014-09-19 21:56:45 +00:00
Zachary Turner 4171da5cfe Make sure to #include <atomic> when using std::atomic.
llvm-svn: 218148
2014-09-19 20:12:32 +00:00
Carlo Kok b77aba7708 Fix for 218140 for SBTarget.i, the added functions were in the wrong class definition
llvm-svn: 218147
2014-09-19 20:12:24 +00:00
Greg Clayton 615eb7e609 Test suite runs better again after recent fixes that would select a platform if a "file a.out" auto selected a different platform than the selected one.
Changes include:
- fix it so you can select the "host" platform using "platform select host"
- change all callbacks that create platforms to returns shared pointers
- fix TestImageListMultiArchitecture.py to restore the "host" platform by running "platform select host"
- Add a new "PlatformSP Platform::Find(const ConstString &name)" method to get a cached platform
- cache platforms that are created and re-use them instead of always creating a new one

llvm-svn: 218145
2014-09-19 20:11:50 +00:00
Carlo Kok 0fd6fd4fd4 Adds two new functions to SBTarget FindGlobalVariables and FindGlobalFunctions that lets you search by name, by regular expression and by starts with.
llvm-svn: 218140
2014-09-19 19:38:19 +00:00
Todd Fiala 850f9a25a5 Fix typo in Linux ASLR logging.
llvm-svn: 218133
2014-09-19 18:27:45 +00:00
Enrico Granata 47caf9a956 Extend the member function discovery APIs to also support Objective-C as well as C++
For the Objective-C case, we do not have a "function type" notion, so we actually end up wrapping the clang ObjCMethodDecl in the Impl object, and ask function-y questions of it
In general, you can always ask for return type, number of arguments, and type of each argument using the TypeMemberFunction layer - but in the C++ case, you can also acquire a Type object for the function itself, which instead you can't do in the Objective-C case

llvm-svn: 218132
2014-09-19 18:21:05 +00:00
Greg Clayton ee3be91995 Add some tests to be skipped when run remotely and also fixed a test to be more reliable when it comes to stopping on a specific line.
llvm-svn: 218079
2014-09-18 21:32:05 +00:00
Todd Fiala 87bac59adc llgs: removed some wait-for-stop code in inferior process launch pipeline.
The $A handler was unnecessarily waiting for the launched app to hit a stop
before returning.  Removed this code.

Renamed the llgs inferior launching code to LaunchProcessForDebugging ()
to prevent it from possibly being mistaken as code that lldb-platform uses
to launch a debugserver process.  We probably want to look at breaking out
llgs-specific and lldb-platform-specific code into separate derived classes,
with common code in a shared base class.

llvm-svn: 218075
2014-09-18 21:02:03 +00:00
Greg Clayton 109534cb59 Fix the ability of "lldb-platform" to upload files.
The issue was GDBRemoteCommunication::CheckForPacket() already fixes up any prefixed bytes (0x7d followed by value that is XOR'ed with 0x20). If we do this again, we cause binary packets to lose bytes.

This allows lldb-platform to be able to upload binaries and debug them remotely.

llvm-svn: 218002
2014-09-18 00:20:51 +00:00
Greg Clayton 44272a40dc Hex encode the triple values in case they contain special characters.
llvm-svn: 218001
2014-09-18 00:18:32 +00:00
Greg Clayton 7e2443258f Add better logging for the "$vFile:pwrite:" packet so we can show binary data instead of nothing or unprintable characters. This can easily be extended for other packets that have binary data.
llvm-svn: 218000
2014-09-18 00:17:36 +00:00
Greg Clayton 90d9a35ff1 Listen to the return value of the Platform::WriteFile() call within PlatformPOSIX::PutFile() in case we write less than we wanted to. Also adjust the input stream's offset in such cases.
llvm-svn: 217999
2014-09-18 00:16:13 +00:00
Greg Clayton 56bf559253 Retry remote connection one time in case of timeout and return an error that specifies why debugging failed.
llvm-svn: 217998
2014-09-18 00:14:38 +00:00
Greg Clayton 95bbdf6428 If a new platform is created/selected for an architecture, make it the selected platform so you can do things like:
% lldb ios-executable
(lldb) platform connect connect://localhost:11111

Prior to this fix, the host platform would be selected even though the target was using the ios-remote platform.

llvm-svn: 217963
2014-09-17 16:42:50 +00:00
Ed Maste def0dae250 Build lldb-mi on all platforms with Makefile builds
CMake builds already do so.

llvm-svn: 217931
2014-09-17 00:28:31 +00:00
Enrico Granata 6389a7da08 More categories could be added, and having an explicit list in this test case is a fairly bullet-proof recipe for disaster. Use the * enabling feature to workaround this. Fixes rdar://18358649
llvm-svn: 217927
2014-09-16 22:53:30 +00:00
Greg Clayton e37df2e64b Fixed an issue where the last N load commands in the mach-o core file would not be read in where N was the number of LC_THREAD load commands.
I now properly increment the ncmds for each LC_THREAD and now core files are saved correctly.

<rdar://problem/18312703> 

llvm-svn: 217905
2014-09-16 20:50:29 +00:00
Enrico Granata ba4b788ae7 Unused functions break the -Werror build. Revert for now.
llvm-svn: 217900
2014-09-16 20:28:26 +00:00
Enrico Granata 438aba6fdd Add a convenience function to FormatManager to setup an empty filter (one that suppresses all children, that is)
llvm-svn: 217891
2014-09-16 17:41:54 +00:00
Enrico Granata 2267ad442a Change the ClangASTMap implementation to use a thread-safe wrapper over llvm::DenseMap. This helps avoid a certain class of spins per <rdar://problem/18160764>
llvm-svn: 217888
2014-09-16 17:28:40 +00:00
David Majnemer 8faf9370fa Clean-up warnings on Linux/GCC
llvm-svn: 217862
2014-09-16 06:34:29 +00:00
Greg Clayton 48e17578ed Continuation broken for Python scripts when using non-interactive input (Xcode for example).
The problem was the read_func we were supplying to the interactive interpreter wasn't stripping the newline from the end of the string. Now it does and multi-line python scripts can be typed in Xcode.

<rdar://problem/17696438>

llvm-svn: 217843
2014-09-15 22:46:25 +00:00
Todd Fiala fbd703add5 ELF: store the thumbness of a function in symbol flags.
This allows us to fixup the address of the symbol as soon as we parse it
so that lldb is not confused thinking there are two different symbols in
the binary (one with the thumb bit, one without). Also, differentiating
between THUMB and ARM symbols allows the debugger to place the right
type of breakpoint.

Change by Stephane Sezer.

llvm-svn: 217841
2014-09-15 22:33:39 +00:00
Enrico Granata 6c42cb11d6 Change SBType.GetMemberFunctionAtIndex() to return an object describing the member function in more detail. A type was really quite vague. This now has function name, kind, as well as function type
llvm-svn: 217828
2014-09-15 21:30:36 +00:00
Todd Fiala 7b0917a0c5 use std::atomic<> to protect variables being accessed by multiple threads
There are several places where multiple threads are accessing the same variables simultaneously without any kind of protection. I propose using std::atomic<> to make it safer. I did a special build of lldb, using the google tool 'thread sanitizer' which identified many cases of multiple threads accessing the same memory. std::atomic is low overhead and does not use any locks for simple types such as int/bool.

See http://reviews.llvm.org/D5302 for more details.

Change by Shawn Best.

llvm-svn: 217818
2014-09-15 20:07:33 +00:00
Todd Fiala 49bc2d9dd3 Fix JITLoaderGDB for 64-bit host and 32-bit target
There are target pointer members in struct jit_code_entry and jit_descriptor.

Data layout of those structures should be decided by target, not host.

This fixes JITLoaderGDB for 64-bit host and 32-bit target.

See http://reviews.llvm.org/D5339 for more details.

Change by Tong Shen.

llvm-svn: 217816
2014-09-15 19:55:27 +00:00
Enrico Granata 735152e3b0 Add a --help (-h) option to "command script add" that enables users to define a one-liner short help for their command
Also, in case they don't define any, change the default from "Run Python function <blah>" into "For more information run help <blah>"

The core issue here is that Python only allows one docstring per function, so we can't really attach both a short and a long help to the same command easily
There are alternatives but this is not a pressing enough concern to go through the motions quite yet

Fixes rdar://18322737

llvm-svn: 217795
2014-09-15 17:52:44 +00:00
Todd Fiala 0fceef8065 lldb fix ARM64 register access - llgs side
See http://reviews.llvm.org/D5341 for more details.

Change by Paul Osmialowski.

llvm-svn: 217788
2014-09-15 17:09:23 +00:00
Todd Fiala 1a08866aab Handle ARM ELF symbols properly: skip $t* and $a* symbols in ObjectFileELF.
ELF objects contain marker symbols to differentiate between ARM and
THUMB functions. Instead of storing them internally and having garbage
show up when symbols are searched for by the user, we can just skip them
and not store them at all, as we never actually need them.

Change by Stephane Sezer.

Tested:
Ubuntu 14.04 x86_64
MacOSX 10.9.4 x86_64

llvm-svn: 217782
2014-09-15 16:27:44 +00:00
Todd Fiala 3daa176986 Properly decode architecture type in GDBRemoteCommunicationClient::GetCurrentProcessInfo.
Instead of forcing the remote arch type to MachO all the time, we
inspect the OS/vendor that the remote debug server reports and use it to
set the arch type to MachO, ELF or COFF accordingly.

See thread here for more context:
http://lists.cs.uiuc.edu/pipermail/lldb-commits/Week-of-Mon-20140915/012968.html

Change by Stephane Sezer.

Tested:
MacOSX 10.9.4 x86_64
Ubuntu 14.04 x86_64

llvm-svn: 217779
2014-09-15 16:01:29 +00:00
Todd Fiala 5c9d5bf81e Check for byte order correctness in GDBRemoteCommunicationClient::GetCurrentProcessInfo.
This is useful for checking inconsistencies between what the remote debug server thinks we are debugging and we think we are debugging. This follows the check for pointer byte size done just above.

Change by Stephane Sezer.

Tested:
Ubuntu 14.04 x86_64, llvm-3.5-built lldb
MacOSX 10.9.4, Xcode-Beta(2014-09-09)-built lldb.

llvm-svn: 217773
2014-09-15 15:31:11 +00:00