This change opens a socket pair and passes the second socket pair file descriptor down to the debugserver binary using a new option: "--fd=N" where N is the file descriptor. This file descriptor gets passed via posix_spawn() so that there is no need to do any bind/listen or bind/accept calls and eliminates the hanshake unix socket that is used to pass the result of the actual port that ends up being used so it can save time on launch as well as being faster.
This is currently only enabled on __APPLE__ builds. Other OSs should try modifying the #define from ProcessGDBRemote.cpp but the first person will need to port the --fd option over to lldb-server. Any OSs that enable USE_SOCKETPAIR_FOR_LOCAL_CONNECTION in their native builds can use the socket pair stuff. The #define is Apple only right now, but looks like:
#if defined (__APPLE__)
#define USE_SOCKETPAIR_FOR_LOCAL_CONNECTION 1
#endif
<rdar://problem/27814880>
llvm-svn: 278524
Options used to store a reference to the CommandInterpreter instance
in the base Options class. This made it impossible to parse options
independent of a CommandInterpreter.
This change removes the reference from the base class. Instead, it
modifies the options-parsing-related methods to take an
ExecutionContext pointer, which the options may inspect if they need
to do so.
Closes https://reviews.llvm.org/D23416
Reviewers: clayborg, jingham
llvm-svn: 278440
Resumbitting the commit after fixing the following problems:
- broken unit tests on windows: incorrect gtest usage on my part (TEST vs. TEST_F)
- the new code did not correctly handle the case where we went to interrupt the process, but it
stopped due to a different reason - the interrupt request would remain queued and would
interfere with the following "continue". I also added a unit test for this case.
This reapplies r277156 and r277139.
llvm-svn: 278118
This reverts commit r277139, because:
- broken unittest on windows (likely typo on my part)
- seems to break TestCallThatRestart (needs investigation)
llvm-svn: 277154
SendContinuePacketAndWaitForResponse was huge function with very complex interactions with
several other functions (SendAsyncSignal, SendInterrupt, SendPacket). This meant that making any
changes to how packet sending functions and threads interact was very difficult and error-prone.
This change does not add any functionality yet, it merely paves the way for future changes. In a
follow-up, I plan to add the ability to have multiple query packets in flight (i.e.,
request,request,response,response instead of the usual request,response sequences) and use that
to speed up qModuleInfo packet processing.
Here, I introduce two special kinds of locks: ContinueLock, which is used by the continue thread,
and Lock, which is used by everyone else. ContinueLock (atomically) sends a continue packet, and
blocks any other async threads from accessing the connection. Other threads create an instance of
the Lock object when they want to access the connection. This object, while in scope prevents the
continue from being send. Optionally, it can also interrupt the process to gain access to the
connection for async processing.
Most of the syncrhonization logic is encapsulated within these two classes. Some of it still
had to bleed over into the SendContinuePacketAndWaitForResponse, but the function is still much
more manageable than before -- partly because of most of the work is done in the ContinueLock
class, and partly because I have factored out a lot of the packet processing code separate
functions (this also makes the functionality more easily testable). Most importantly, there is
none of syncrhonization code in the async thread users -- as far as they are concerned, they just
need to declare a Lock object, and they are good to go (SendPacketAndWaitForResponse is now a
very thin wrapper around the NoLock version of the function, whereas previously it had over 100
lines of synchronization code). This will make my follow up changes there easy.
I have written a number of unit tests for the new code and I have ran the test suite on linux and
osx with no regressions.
Subscribers: tberghammer
Differential Revision: https://reviews.llvm.org/D22629
llvm-svn: 277139
This finally removes the use of the Mutex and Condition classes. This is an
intricate patch as the Mutex and Condition classes were tied together.
Furthermore, many places had slightly differing uses of time values. Convert
timeout values to relative everywhere to permit the use of
std::chrono::duration, which is required for the use of
std::condition_variable's timeout. Adjust all Condition and related Mutex
classes over to std::{,recursive_}mutex and std::condition_variable.
This change primarily comes at the cost of breaking the TracingMutex which was
based around the Mutex class. It would be possible to write a wrapper to
provide similar functionality, but that is beyond the scope of this change.
llvm-svn: 277011
debugserver jGetSharedCacheInfo packet instead of reading
the dyld internal data structures directly. This code is
(currently) only used for ios native lldb's - I should really
move this ObjectFileMachO::GetProcessSharedCacheUUID method
somewhere else, it makes less and less sense being in the
file reader.
<rdar://problem/25251243>
llvm-svn: 276369
for the fall (northern hemisphere) 2016 Darwin platforms to learn
about loaded images, instead of reading dyld internal data structures.
These new SPI don't exist on older releases, and new packets are
needed from debugserver to use them (those changes are already committed).
I had to change the minimum deployment target for debugserver in the xcode
project file to macOS 10.10 so that debugserver will use the
[[NSProcessInfo processInfo] operatingSystemVersion]
call in MachProcess::GetOSVersionNumbers to get the operarting system
version # -- this API is only available in macOS 10.10 and newer
("OS X Yosemite", released Oct 2014). If we have many people building
llvm.org lldb on older systems still, we can back off on this for the
llvm.org sources.
There should be no change in behavior with this commit, either to
older darwin systems or newer darwin systems.
For now the new DynamicLoader plugin is never activated - I'm forcing
the old plugin to be used in DynamicLoaderDarwin::UseDYLDSPI.
I'll remove that unconditional use of the old plugin soon, so the
newer plugin is used on the newest Darwin platforms.
<rdar://problem/25251243>
llvm-svn: 276254
review it for consistency, accuracy, and clarity. These changes attempt to
address all of the above while keeping the text relatively terse.
<rdar://problem/24868841>
llvm-svn: 275485
for TestNamespaceLookup.py; didn't see anything obviously wrong so I'll
need to look at this more closely before re-committing. (passed OK on
macOS ;)
llvm-svn: 273531
There's uses of "macosx" that will be more tricky to
change, like in triples (e.g. "x86_64-apple-macosx10.11") -
for now I'm just updating source comments and strings printed
for humans.
llvm-svn: 273524
This patch allows LLDB for AArch64 to watch all bytes, words or double words individually on non 8-byte alligned addresses.
This patch also adds tests to verify this functionality.
Differential revision: http://reviews.llvm.org/D21280
llvm-svn: 272916
Summary:
Because PIE executables have an e_type of llvm::ELF::ET_DYN,
they are not of type eTypeExecutable, and were being removed
when svr4 packets were used.
Reviewers: clayborg, ADodds, tfiala, sas
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D20990
llvm-svn: 271899
The error was not getting propagated to the caller, so the higher layers thought the breakpoint
was successfully set & resolved.
I added a testcase, but it assumes 0x0 is not a valid place to set a breakpoint. On most systems
that is true, but if it isn't true of your system, either find another good place and add it to the
test, or x-fail the test.
<rdar://problem/26345962>
llvm-svn: 270014
This is a pretty straightforward first pass over removing a number of uses of
Mutex in favor of std::mutex or std::recursive_mutex. The problem is that there
are interfaces which take Mutex::Locker & to lock internal locks. This patch
cleans up most of the easy cases. The only non-trivial change is in
CommandObjectTarget.cpp where a Mutex::Locker was split into two.
llvm-svn: 269877
Summary:
MonitorDebugServerProcess went to a lot of effort to make sure its asynchronous invocation does
not cause any mischief, but it was still not race-free. Specifically, in a quick stop-restart
sequence (like the one in TestAddressBreakpoints) the copying of the process shared pointer via
target_sp->GetProcessSP() was racing with the resetting of the pointer in DeleteCurrentProcess,
as they were both accessing the same shared_ptr object.
To avoid this, I simply pass in a weak_ptr to the process when the callback is created. Locking
this pointer is race-free as they are two separate object even though they point to the same
process instance. This also removes the need for the complicated tap-dance around retrieving the
process pointer.
Reviewers: clayborg
Subscribers: tberghammer, lldb-commits
Differential Revision: http://reviews.llvm.org/D20107
llvm-svn: 269281
Summary:
This replaces the C-style "void *" baton of the child process monitoring functions with a more
C++-like API taking a std::function. The motivation for this was that it was very difficult to
handle the ownership of the object passed into the callback function -- each caller ended up
implementing his own way of doing it, some doing it better than others. With the new API, one can
just pass a smart pointer into the callback and all of the lifetime management will be handled
automatically.
This has enabled me to simplify the rather complicated handshake in Host::RunShellCommand. I have
left handling of MonitorDebugServerProcess (my original motivation for this change) to a separate
commit to reduce the scope of this change.
Reviewers: clayborg, zturner, emaste, krytarowski
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D20106
llvm-svn: 269205
Summary:
If the remote uses svr4 packets to communicate library info,
the LoadUnload tests will fail, as lldb only used the basename
for modules, causing problems when two modules have the same basename.
Using absolute path as sent by the remote will ensure that lldb
locates the module from the correct directory when there are overlapping
basenames. When debugging a remote process, LoadModuleAtAddress will still
fall back to using basename and module_search_paths, so we don't
need to worry about using absolute paths in this case.
Reviewers: ADodds, jasonmolenda, clayborg, ovyalov
Subscribers: lldb-commits, sas
Differential Revision: http://reviews.llvm.org/D19557
llvm-svn: 267741
Summary:
If the remote uses include features when communicating
xml register info back to lldb, the existing code would reset the
lldb register index at the beginning of each include node.
This would lead to multiple registers having the same lldb register index.
Since the lldb register numbers should be contiguous and unique,
maintain them accross the parsing of all of the xml feature nodes.
Reviewers: jingham, jasonmolenda, clayborg
Subscribers: lldb-commits, sas
Differential Revision: http://reviews.llvm.org/D19303
llvm-svn: 267468
Summary:
When we receive an svr4 packet from the remote, we check for new modules
and add them to the list of images in the target. However, we did not
do the same for modules which have been removed.
This was causing TestLoadUnload to fail when using ds2, which uses
svr4 packets to communicate all library info on Linux. This patch fixes
the failing test.
Reviewers: zturner, tfiala, ADodds
Subscribers: lldb-commits, sas
Differential Revision: http://reviews.llvm.org/D19230
llvm-svn: 267467
In turns out this does make a functional change, in case when the inferior hits an int3 that was
not placed by the debugger. Backing out for now.
llvm-svn: 265647
Summary:
SetThreadStopInfo was checking for a breakpoint at the current PC several times. This merges the
identical code into a separate function. I've left one breakpoint check alone, as it was doing
more complicated stuff, and it did not see a way to merge that without making the interface
complicated. NFC.
Reviewers: clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D18819
llvm-svn: 265560
Summary:
This resolves a similar problem as D16720 (which handled the case when we single-step onto a
breakpoint), but this one deals with involutary stops: when we stop a thread (e.g. because
another thread has hit a breakpont and we are doing a full stop), we can end up stopping it right
before it executes a breakpoint instruction. In this case, the stop reason will be empty, but we
will still step over the breakpoint when do the next resume, thereby missing a breakpoint hit.
I have observed this happening in TestConcurrentEvents, but I have no idea how to reproduce this
behavior more reliably.
Reviewers: clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D18692
llvm-svn: 265525
Summary:
The logic to read modules from memory was added to LoadModuleAtAddress
in the dynamic loader, but not in process gdb remote. This means that when
the remote uses svr4 packets to give library info, libraries only present
on the remote will not be loaded.
This patch therefore involves some code duplication from LoadModuleAtAddress
in the dynamic loader, but removing this would require some amount of code
refactoring.
Reviewers: ADodds, tberghammer, tfiala, deepak2427, ted
Subscribers: tfiala, lldb-commits, sas
Differential Revision: http://reviews.llvm.org/D18531
Change by Francis Ricci <fjricci@fb.com>
llvm-svn: 265418
rnb_err_t
RNBRemote::HandlePacket_stop_process (const char *p)
{
if (!DNBProcessInterrupt(m_ctx.ProcessID()))
HandlePacket_last_signal (NULL);
return rnb_success;
}
In the call to DNBProcessInterrupt we did:
nub_bool_t
DNBProcessInterrupt(nub_process_t pid)
{
MachProcessSP procSP;
if (GetProcessSP (pid, procSP))
return procSP->Interrupt();
return false;
}
This would always return false. It would cause HandlePacket_stop_process to always call "HandlePacket_last_signal (NULL);" which would send an extra stop reply packet _if_ the process is stopped. On a machine with enough cores, it would call DNBProcessInterrupt(...) and then HandlePacket_last_signal(NULL) so quickly that it will never send out an extra stop reply packet. But if the machine is slow enough or doesn't have enough cores, it could cause the call to HandlePacket_last_signal() to actually succeed and send an extra stop reply packet. This would cause problems up in GDBRemoteCommunicationClient::SendContinuePacketAndWaitForResponse() where it would get the first stop reply packet and then possibly return or execute an async packet. If it returned, then the next packet that was sent will get the second stop reply as its response. If it executes an async packet, the async packet will get the wrong response.
To fix this I did the following:
1 - in debugserver, I fixed "bool MachProcess::Interrupt()" to return true if it sends the signal so we avoid sending the stop reply twice on slower machines
2 - Added a log line to RNBRemote::HandlePacket_stop_process() to say if we ever send an extra stop reply so we will see this in the darwin console output if this does happen
3 - Added response validators to StringExtractorGDBRemote so that we can verify some responses to some packets.
4 - Added validators to packets that often follow stop reply packets like the "m" packet for memory reads, JSON packets since "jThreadsInfo" is often sent immediately following a stop reply.
5 - Modified GDBRemoteCommunicationClient::SendPacketAndWaitForResponseNoLock() to validate responses. Any "StringExtractorGDBRemote &response" that contains a valid response verifier will verify the response and keep looking for correct responses up to 3 times. This will help us get back on track if we do get extra stop replies. If a StringExtractorGDBRemote does not have a response validator, it will accept any packet in response.
6 - In GDBRemoteCommunicationClient::SendPacketAndWaitForResponse we copy the response validator from the "response" argument over into m_async_response so that if we send the packet by interrupting the running process, we can validate the response we actually get in GDBRemoteCommunicationClient::SendContinuePacketAndWaitForResponse()
7 - Modified GDBRemoteCommunicationClient::SendContinuePacketAndWaitForResponse() to always check for an extra stop reply packet for 100ms when the process is interrupted. We were already doing this because we might interrupt a process with a \x03 packet, yet the process was in the process of stopping due to another reason. This race condition could cause an extra stop reply packet because the GDB remote protocol says if a \x03 packet is sent while the process is stopped, we should send a stop reply packet back. Now we always check for an extra stop reply packet when we manually interrupt a process.
The issue was showing up when our IDE would attempt to set a breakpoint while the process is running and this would happen:
--> \x03
<-- $T<stop reply 1>
--> z0,AAAAA,BB (set breakpoint)
<-- $T<stop reply 1> (incorrect extra stop reply packet)
--> c
<-- OK (response from z0 packet)
Now all packet traffic was off by one response. Since we now have a validator on the response for "z" packets, we do this:
--> \x03
<-- $T<stop reply 1>
--> z0,AAAAA,BB (set breakpoint)
<-- $T<stop reply 1> (Ignore this because this can't be the response to z0 packets)
<-- OK -- (we are back on track as this is a valid response to z0)
...
As time goes on we should add more packet validators.
<rdar://problem/22859505>
llvm-svn: 265086
to each other. This should remove some infrequent teardown crashes when the
listener is not the debugger's listener.
Processes now need to take a ListenerSP, not a Listener&.
This required changing over the Process plugin class constructors to take a ListenerSP, instead
of a Listener&. Other than that there should be no functional change.
<rdar://problem/24580184> CrashTracer: [USER] Xcode at …ework: lldb_private::Listener::BroadcasterWillDestruct + 39
llvm-svn: 262863
on attach uses the architecture it has figured out, rather than the Target's
architecture, which may not have been updated to the correct value yet.
<rdar://problem/24632895>
llvm-svn: 261279
Summary:
r259344 introduced a bug, where we fail to perform a single step, when the instruction we are
stepping onto contains a breakpoint which is not valid for this thread. This fixes the problem
and add a test case.
Reviewers: tberghammer, emaste
Subscribers: abhishek.aggarwal, lldb-commits, emaste
Differential Revision: http://reviews.llvm.org/D16767
llvm-svn: 259488
Summary:
- The patch solves Bug 23478 and Bug 19311. Resolving
Bug 23478 also resolves Bug 23039.
Correct ThreadStopInfo is set for Linux and FreeBSD
platforms.
- Summary:
When a trace event is reported, we need to check
whether the trace event lands at a breakpoint site.
If it lands at a breakpoint site then set the thread's
StopInfo with the reason 'breakpoint'. Else, set the reason
to be 'Trace'.
Change-Id: I0af9765e782fd74bc0cead41548486009f8abb87
Signed-off-by: Abhishek Aggarwal <abhishek.a.aggarwal@intel.com>
Reviewers: jingham, emaste, lldb-commits, clayborg, ovyalov
Subscribers: emaste
Differential Revision: http://reviews.llvm.org/D16720
llvm-svn: 259344
Summary:
Allows the remote to enumerate the link map when adding and removing
shared libraries, so that lldb doesn't need to read it manually from
the remote's memory.
This provides very large speedups (on the order of 50%) in total
startup time when using the ds2 remote on android or Tizen devices.
Reviewers: ADodds, tberghammer, tfiala
Subscribers: tberghammer, sas, danalbert, llvm-commits, srhines
Differential Revision: http://reviews.llvm.org/D16004
Change by Francis Ricci <fjricci@fb.com>
llvm-svn: 257502
at each public stop to improve performance a bit. Most of the
information lldb needed was already in the jThreadsInfo response;
complete that information and catch a few cases where we could still
fall back to getting the information via discrete memory reads.
debugserver adds 'associated_with_dispatch_queue' and 'dispatch_queue_t
keys to the jThreadsInfo response for all the threads. lldb needs the
dispatch_queue_t value. And associated_with_dispatch_queue helps to
identify which threads definitively don't have any queue information so
lldb doesn't try to do memory reads to get that information just because
it was absent in the jThreadsInfo response.
Remove the queue information from the questionmark (T) packet. We'll
get the information for all threads via the jThreadsInfo response -
sending the information for the stopping thread (on all the private
stops, plus the less frequent public stop) was unnecessary information
being sent over the wire.
SystemRuntimeMacOSX will try to get information about queues by asking
the Threads for them, instead of reading memory.
ProcessGDBRemote changes to recognize the new keys being sent in the
jThreadsInfo response. Changes to ThreadGDBRemote to track the new
information. Also, when a thread is marked as definitively not
associated with a libdispatch queue, don't fall back to the system
runtime to try memory reads to find the queue name / kind / ID etc.
<rdar://problem/23309359>
llvm-svn: 257453
"qserial" to "qserialnum" because "qserial" looks a lot like the
queue type (either 'serial' or 'concurrent') and can be confusing
to read through. debugserver passes these up either in the questionmark
("T") packet, or in the response to the jThreadsInfo packet.
llvm-svn: 257121
(There are changes in the copies of these four files in the FreeBSD base
system, and I've changed these ones to reduce gratuitous diffs in future
imports.)
llvm-svn: 256723
"thread-pcs" key is added to the T (questionmark) packet in
gdb-remote protocol so that lldb doesn't need to query the
pc values of every thread before it resumes a process.
The only odd part with this is that I'm sending the pc
values in big endian order, so we need to know the endianness
of the remote process before we can use them. All other
register values in gdb-remote protocol are sent in native-endian
format so this requirement doesn't exist. This addition is a
performance enhancement -- lldb will fall back to querying the
pc of each thread individually if it needs to -- so when
we don't have the byte order for the process yet, we don't
use these values. Practically speaking, the only way I've
been able to elicit this condition is for the first
T packet when we attach to a process.
<rdar://problem/21963031>
llvm-svn: 255942
Summary:
Since this is within the lldb namespace, the compiler tries to
export a symbol for it. Unfortunately, since it is inlined, the
symbol is hidden and this results in a mess of warnings when
building on OS X with cmake.
Moving it to the lldb_private namespace eliminates that problem.
Reviewers: clayborg
Subscribers: emaste, lldb-commits
Differential Revision: http://reviews.llvm.org/D14417
llvm-svn: 252396
