Commit Graph

2270 Commits

Author SHA1 Message Date
Enrico Granata 1ac6296376 <rdar://problem/12055586>
Enable data formatters to see-through pointers/references to typedefs

For instance, if Foo is a typedef to Bar, and there is a formatter for any/all of Bar*, Bar&, Bar&&, then Foo*, Foo&, and Foo&& should pick these up if Foo-specific formatters don't exist

llvm-svn: 205939
2014-04-10 00:14:07 +00:00
Ed Maste 6562983234 Do not rely on invalid pthread API use in thread test
Calling mutex_lock from one thread and then mutex_unlock from another is
not permitted.  Replace the awkward mutex usage with a mutex and
condition variable.

llvm.org/pr18061                                                            

llvm-svn: 205900
2014-04-09 14:48:25 +00:00
Ed Maste 8b2c281a75 Correct typo: waychpoint -> watchpoint
llvm-svn: 205719
2014-04-07 16:33:53 +00:00
Ed Maste e63b5125e0 Add decorator for FreeBSD DynamicValueTestCase failure
Not yet root-caused, but I presume it is the same issue that affects
Linux.

llvm-svn: 205712
2014-04-07 13:42:05 +00:00
Ed Maste 99ed6dfa00 Add failure decorators
These tests started failing on FreeBSD after r205497 "Make the fail
messages"

llvm.org/pr19347

llvm-svn: 205652
2014-04-04 21:16:39 +00:00
Jim Ingham 6c9ed91cca Make the fail messages
llvm-svn: 205497
2014-04-03 01:26:14 +00:00
Jim Ingham 6cc0d2f61d Print the new eStopReasonExec in stop_reason_to_str.
llvm-svn: 205496
2014-04-03 01:25:28 +00:00
Jim Ingham 24e84d2c87 I removed SetCallback from the Python API's since it wasn't actually useful, and added SetScriptCallbackFunction,
and SetScriptCallbackBody.  So add these to the default constructor test case.

llvm-svn: 205494
2014-04-03 00:50:56 +00:00
Ed Maste abd482a650 Add decorator for pr19311 on FreeBSD
A breakpoint at the resume address for "continue" is skipped.  This is
presumably the root cause for the Linux failure as well.

llvm-svn: 205482
2014-04-02 23:01:04 +00:00
Enrico Granata c1fb7bd33e rdar://16424649
Clean up the DynamicValueTestCase

Namely, I split off the part of the test that validates child counts to a separate test

llvm-svn: 205450
2014-04-02 18:55:29 +00:00
Ed Maste e7f59bb04d Add annotation for failure from new Attach/Resume test case on FreeBSD
llvm-svn: 205427
2014-04-02 14:45:46 +00:00
Andrew MacPherson 82aae0d835 Use getpgid() with waitpid() in case the process pgid is not equal to its pid, as is the case with a forked subprocess. Also a couple of fixes for unit test failures from Todd Fiala.
llvm-svn: 205405
2014-04-02 06:57:45 +00:00
Jim Ingham 7d8f86ceae Test case for the previous commit.
llvm-svn: 205381
2014-04-02 01:05:27 +00:00
Enrico Granata a4d5fc3841 Use continue instead of next
llvm-svn: 205354
2014-04-01 19:37:14 +00:00
Ed Maste c97323ea8b Use libc++ on FreeBSD as on Darwin
llvm-svn: 205345
2014-04-01 18:47:58 +00:00
Ed Maste 1087f2980d Add errno import missed in r205246
llvm-svn: 205306
2014-04-01 12:49:21 +00:00
Ed Maste ce5ac7d776 Workaround test trace output dir creation race condition
Since dosep.ty started invoking multiple tests in parallel, the FreeBSD
buildbot occasionally has a failure due to os.mkdir returning EEXIST.
Silently ignore that exception, but reraise any other.

llvm-svn: 205246
2014-03-31 20:36:38 +00:00
Jason Molenda a332978b2a lldb arm64 import.
These changes were written by Greg Clayton, Jim Ingham, Jason Molenda.

It builds cleanly against TOT llvm with xcodebuild.  I updated the
cmake files by visual inspection but did not try a build.  I haven't
built these sources on any non-Mac platforms - I don't think this
patch adds any code that requires darwin, but please let me know if
I missed something.

In debugserver, MachProcess.cpp and MachTask.cpp were renamed to
MachProcess.mm and MachTask.mm as they picked up some new Objective-C
code needed to launch processes when running on iOS.

llvm-svn: 205113
2014-03-29 18:54:20 +00:00
Jim Ingham 914f4e7092 Add the ability from the SB API's to set the "one thread" timeout
for expression evaluations that try one and then all threads.

<rdar://problem/15598528>

llvm-svn: 205060
2014-03-28 21:58:28 +00:00
Ed Maste fdb85d8dc3 Remove annotation for llvm.org/pr19241
The issue has been fixed by r204745 and r204750

llvm-svn: 204779
2014-03-26 02:45:44 +00:00
Enrico Granata dd75b8556d Add a test case for the previous commit
llvm-svn: 204763
2014-03-25 22:09:24 +00:00
Enrico Granata f6601d0498 Make sure this test has a looser dependency on the exact class generated here.. it is going to be some sort of NS-provided String, but let's not bet on the details
llvm-svn: 204761
2014-03-25 21:59:40 +00:00
Todd Fiala e373b68977 Fixed up intermittently failing tests to skip on Linux.
Also added 'import sys' on some tests that are using non-standard
unittest2.skipUnless blocks with code that is intended to do things
that we have more specializes @* attributes for.  These skip
conditions were failing to execute due to missing import, causing
darwin-only tests to run on Linux regardless.  Will file a bug for
that separately.

llvm-svn: 204747
2014-03-25 18:55:48 +00:00
Ed Maste 03c9207c34 Update 'running threaded' messages after option parsing change
"Running multithreaded with n threads (from LLDB_TEST_THREADS)" is
incorrect in the common case, because the thread count is now obtained
from a --threads option, or a default of multiprocessing.cpu_count().

I left the messages in for now as there are some intermittent issues
that happen while running tests multithreaded and it may help in
triaging those.  We may wish to just remove them later on.

llvm-svn: 204725
2014-03-25 15:17:23 +00:00
Ed Maste 8c3e9b2515 Add expected failure annotation for llvm.org/pr19241
llvm-svn: 204718
2014-03-25 13:15:21 +00:00
Greg Clayton 2256d0dced Add a "--threads N" option instead of having to use an environment variable. It also now defaults to running with the number of CPUs on the machine.
llvm-svn: 204681
2014-03-24 23:01:57 +00:00
Ed Maste b04fb08c8a Add annotation for test failure due to clang 3.4
FreeBSD recently updated to Clang 3.4 and the TestFormatters test case
started failing as it omits the C1 complete object constructor when not
needed.

llvm.org/pr19011

llvm-svn: 204652
2014-03-24 18:30:59 +00:00
Jim Ingham 74287f4a12 Adapt test to avoid short string types.
llvm-svn: 204295
2014-03-20 02:21:17 +00:00
Jim Ingham c81a99eb48 More recent compilers emit debug info for the length property of NSString, so the length will be correctly
reported as NSUInteger.  Adopt the test case to handle either possibility.

llvm-svn: 204288
2014-03-20 01:08:27 +00:00
Jim Ingham 4e2084d792 expectedFailureDarwin for this test case.
llvm-svn: 204287
2014-03-20 00:19:16 +00:00
Jim Ingham c49d0d4fe6 I changed the logging test to just test that logging doesn't crash and does produce output. That's
about all it is useful to test.

llvm-svn: 204284
2014-03-19 23:55:54 +00:00
Jim Ingham a7d63fc610 These logging tests depend on the exact output of lldb logging, which is brittle
and not particularly useful.  Skipping till we get a chance to make a test that 
actually tests something we care about.

llvm-svn: 204283
2014-03-19 23:50:48 +00:00
Jim Ingham 040e26a945 Remember to run ranlib after ar to make useful archive files.
llvm-svn: 204282
2014-03-19 23:38:22 +00:00
Jim Ingham 1b15136b52 Remove unnecessary dependence on vecLib.
llvm-svn: 204281
2014-03-19 23:25:11 +00:00
Jim Ingham 7675035235 Check if "golden_output.txt" exists before trying to delete it rather than just raising an uncaught error.
llvm-svn: 204258
2014-03-19 19:45:37 +00:00
Jim Ingham 3a520a9373 Put a workaround for flaky unwinder behavior on Darwin (<rdar://problem/16363195>).
llvm-svn: 204254
2014-03-19 17:50:46 +00:00
Jim Ingham 3a0bcc032a Add a test to check the C's constructor's name is C::C, which will fail everywhere now,
so mark as expected fail.  And find another way to check that we hit our constructor
breakpoint, so we don't need to expected fail the Disassembly tests.

llvm-svn: 204209
2014-03-19 01:54:41 +00:00
Jim Ingham ac57f43072 Don't use the Disassembly test to report the fact that we don't decorate up
constructors with their class name, that seems inessential.  We can make another
test specifically for this if we want to.

llvm-svn: 204208
2014-03-19 01:41:19 +00:00
Jim Ingham 39771db637 Work around clang's ignoring __attribute ( align(16) ) in this test case.
llvm-svn: 204207
2014-03-19 00:50:08 +00:00
Jim Ingham 8e522094b2 Expected failure printing std::string::c_str() due to flubbing the typedef lookup.
llvm-svn: 204206
2014-03-19 00:30:31 +00:00
Steve Pucci 44ba171ae3 Inform users when *not* running dosep.ty multithreaded.
The gain with multithreading is large, but turning it on requires
an environment variable and so is hard for users to discover.  This
gives users a way to discover the feature by printing out a message
when the environment variable is not set.

llvm-svn: 204018
2014-03-16 18:23:59 +00:00
Jim Ingham 20e1458795 Test cases for stepping out past frames with no debug info if that is turned on.
llvm-svn: 203993
2014-03-15 00:57:48 +00:00
Jason Molenda b00cc1f92f Refactor the Queues test case a little based on feedback from Jim.
Add queues tests for when libBacktraceRecording is absent.

llvm-svn: 203864
2014-03-13 23:04:22 +00:00
Jason Molenda 846952f5d4 Also check if the queues are serial or concurrent.
llvm-svn: 203755
2014-03-13 05:43:18 +00:00
Jason Molenda 7f8b9111a5 Add a quick test case for some of the queues debugging support.
It should only run on Darwin systems, and only when a couple of
libraries are available.

llvm-svn: 203754
2014-03-13 05:37:51 +00:00
Jim Ingham 4b4b2478fc This commit reworks how the thread plan's ShouldStopHere mechanism works, so that it is useful not only
for customizing "step-in" behavior (e.g. step-in doesn't step into code with no debug info), but also 
the behavior of step-in/step-out and step-over when they step out of the frame they started in.

I also added as a proof of concept of this reworking a mode for stepping where stepping out of a frame
into a frame with no debug information will continue stepping out till it arrives at a frame that does
have debug information.  This is useful when you are debugging callback based code where the callbacks
are separated from the code that initiated them by some library glue you don't care about, among other
things.

llvm-svn: 203747
2014-03-13 02:47:14 +00:00
Jason Molenda 53b8ea1c89 Need to use rm -rf on dSYM bundle directories.
llvm-svn: 203332
2014-03-08 01:53:27 +00:00
Ed Maste c5a3c9f483 Use predefined $(RM) in clean rule, and -r only for (expected) directories
llvm-svn: 203271
2014-03-07 19:11:00 +00:00
Ed Maste 49f359aea4 Fix malloc thread step-out test on FreeBSD
After hitting the malloc() breakpoint on FreeBSD our top frame is actually
an inlined function malloc_init.

  * frame #0: 0x0000000800dcba19 libc.so.7`malloc [inlined] malloc_init at malloc.c:5397
    frame #1: 0x0000000800dcba19 libc.so.7`malloc(size=1024) + 9 at malloc.c:5949
    frame #2: 0x00000000004006e5 test_step_out_of_malloc_into_function_b_with_dwarf`b(val=1) + 37 at main2.cpp:29

Add a heuristic to keep stepping out until we come to a non-malloc caller,
before checking if it is our desired caller from the test code.

llvm.org/pr17944

llvm-svn: 203268
2014-03-07 19:02:20 +00:00
Ed Maste 0990e05d43 Fix test 'make clean' target for empty $(DSYM) on FreeBSD
A freebsd issue with rm prevents '[g]make clean' from working if $(DSYM)
is empty (fts_open(3) fails if passed an empty argument).

To work around this, simplify the clean target by using three separate
rm invocations: one for the common files, one for the case of non-empty
$(DYLIB_NAME), and one for non-empty $(DSYM).

Issue diagnosed (and reported to FreeBSD) by John Wolfe.

llvm.org/pr17933

llvm-svn: 203253
2014-03-07 17:20:50 +00:00