The global replace in changeset bd80903cf987 was a bit too sweeping: a comment
mentioning the OS function wait() (which exists) was inadvertently changed to
talk about an OS function suspend() (which does not).
Add ll_convert<TO, FROM> template, used as (e.g.):
ll_convert<std::string>(value_of_some_other_string_type);
There is no generic template implementation -- the template exists solely to
provide generic aliases for a bewildering family of llstring.h string-
conversion functions with highly-specific names. There's a generic
implementation, though, for the degenerate case where FROM and TO are
identical.
Add ll_convert<> specialization aliases for most of the string-conversion
functions declared in llstring.h, including the Windows-specific ones
involving llutf16string and std::wstring.
Add a mini-lecture in llstring.h about appropriate use of string types on
Windows.
Add LL_WCHAR_T_NATIVE llpreprocessor.h macro so we can detect whether to
provide separate conversions for llutf16string and std::wstring, or whether
those would collide because the types are identical.
Add inline ll_convert_wide_to_string(const std::wstring&) overloads so caller
isn't required to call arg.c_str(), which naturally permits an ll_convert
alias.
Add ll_convert_wide_to_wstring(), ll_convert_wstring_to_wide() as placeholders
for converting between Windows std::wstring and Linden LLWString, with
corresponding ll_convert aliases. We don't yet have library code to perform
such conversions officially; for now, just copy characters.
Add LLStringUtil::getenv(key) and getoptenv(key) functions. The latter returns
boost::optional<string_type> in case the caller needs to detect absence of a
given environment variable rather than simply accepting a default value.
Naturally getenv(), which accepts a default, is implemented using getoptenv().
getoptenv(), in turn, is implemented using an underlying llstring_getoptenv().
On Windows, llstring_getoptenv() returns boost::optional<std::wstring> (based
on GetEnvironmentVariableW()), whereas elsewhere, llstring_getoptenv() returns
boost::optional<std::string> (based on classic Posix getenv()).
The beauty of generic ll_convert is that the portable LLStringUtilBase<T>::
getoptenv() template can call the platform-specific llstring_getoptenv() and
transparently perform whatever conversion is necessary to return the desired
string_type.
Add windows_message<T>(error) template, with an overload that implicitly calls
GetLastError(). We provide a single concrete windows_message<std::wstring>()
implementation because that's what we get from Windows FormatMessageW() --
everything else is a generic conversion to the desired target string type.
This obviates llprocess.cpp's previous WindowsErrorString() implementation --
reimplement using windows_message<std::string>().
This also introduces LLContinueError for exceptions which should interrupt
some part of viewer processing (e.g. the current coroutine) but should attempt
to let the viewer session proceed.
Derive all existing viewer exception classes from LLException rather than from
std::runtime_error or std::logic_error.
Use BOOST_THROW_EXCEPTION() rather than plain 'throw' to enrich the thrown
exception with source file, line number and containing function.
The change from LLProcessLauncher to LLProcess introduces the possibility of a
NULL (default-constructed) LLProcessPtr. Add certain static LLProcess methods
accepting LLProcessPtr, forwarding to nonstatic method when non-NULL but doing
something reasonable with NULL. Use these methods in LLPLuginProcessParent.
Now LLProcess explicitly requests APR to limit the handles passed to any child
process, instead of wantonly passing whatever happens to be lying around the
parent process at the time.
This requires the latest APR build.
Also revert LLUpdateDownloader::Implementation::mDownloadStream to llofstream
(as in rev 1878a57aebd7) instead of apr_file_t*. Using APR for that file was a
Band-Aid -- a single whacked mole -- for the problem more systemically
addressed by apr_procattr_constrain_handle_set().
On Windows, calling CreateProcess(bInheritHandles=FALSE) is the wrong idea. In
that case, CreateProcess() passes NO handles -- even the files you've
explicitly designated as the child's stdin, stdout, stderr in the STARTUPINFO
struct! Remove LLProcess code to tweak bInheritHandles; we should also remove
the corresponding (useless) APR extension.
Instead, given that the Windows file-locking problem we've observed is
specific to the viewer installer .exe file downloaded by the background
updater logic, use APR file I/O for that specific file. Empirically, both
llofstream and std::ofstream seem to make the open file handle inheritable;
but apr_file_open() documentation says: "By default, the returned file
descriptor will not be inherited by child processes created by
apr_proc_create()." And indeed, it does appear to sidestep the locking problem.
On Windows, Bad Things happen when apr_proc_create() is allowed to pass TRUE
to CreateProcess(bInheritHandles). For instance, the open handle for a new
installer executable file being downloaded by the background updater gets
inadvertently passed to a couple slplugin.exe instances. When the viewer
finishes downloading, closes the file and tries to remove it, Windows balks
because the file is still open by another process. Require an apr_suite
package that includes the new Linden apr_procattr_inherit_set() extension, and
call it to turn off CreateProcess(bInheritHandles).
A static LLProcessPtr variable won't be destroyed until after procedural code
has shut down APR. The trouble is that LLProcess's destructor unregisters
itself from APR -- and, for an autokill LLProcess, attempts to kill the child
process. All that is ill-advised after APR shutdown.
Disable use of apr_pool_note_subprocess() mechanism. This should be another
viable way of coping with static autokill LLProcessPtr variables: when the
designated APR pool is cleaned up, APR promises to kill the child process. But
whether it's an APR bug or a calling error, the present (now disabled) call in
LLProcess results in OUR process, the viewer, getting SIGTERM when it asks to
clean up the global APR pool.
It seems that on Windows, even 32K is too big: one in three load-test runs
fails with a duplicated block. Empirically, reducing it to 4K makes it much
more stable -- at least we can run successfully 100 consecutive times, which
is a step in the right direction.
On Windows we ran into trouble trying to write a biggish (~1 MB) buffer of
data to the child process's stdin pipe with a single apr_file_write() call.
The child actually received corrupted data -- suggesting a possible bug in
either APR or Windows pipes; the same test driving the same logic worked fine
on Mac and Linux. Empirically, iterating over chunks of the buffered data is
more robust.
Previous "read N of M bytes" wording implied that the child had M bytes to
send, but we only read N of them. In reality we have no idea how many bytes
the child is trying to send, only how many the OS is willing to deliver at
this moment. To me, "filled N of M bytes" more clearly implies that M is the
buffer size.
We were using uniform macro to report the APR function and its C++ parameter
expressions. But specifically for apr_proc_create() failure, better to report
the command we're attempting to execute.
We can't count on every child process reading everything we try to write to
it. And if the child terminates with WritePipe data still pending, unless we
explicitly suppress it, Posix will hit us with SIGPIPE. That would terminate
the calling process, boom. "Ignoring" it means APR gets the correct errno,
passes it back to us, we log it, etc.
Previously one might get process-terminated notification but still have to
wait for the child process's final data to arrive on one or more ReadPipes.
That required complex consumer timing logic to handle incomplete pending
ReadPipe data, e.g. a partial last line with no terminating newline. New code
guarantees that by the time LLProcess sends process-terminated notification,
all pending pipe data will have been buffered in ReadPipes.
Document LLProcess::ReadPipe::getPump() notification event; add "eof" key.
Add LLProcess::ReadPipe::getline() and read() convenience methods.
Add static LLProcess::getline() and basename() convenience methods, publishing
logic already present elsewhere.
Use ReadPipe::getline() and read() in unit tests.
Add unit test for "eof" event on ReadPipe::getPump().
Add unit test verifying that final data have been buffered by termination
notification event.
Clarify wording in some of the doc comments; be a bit more explicit about some
of the parameter fields.
Make some query methods 'const'.
Change default LLProcess::ReadPipe::getLimit() value to 0: don't post any
incoming data with notification event unless caller requests it. But do post
pertinent FILESLOT in case caller reuses same listener for both stdout and
stderr.
Use more idiomatic, readable syntax for accessing LLProcess::Params data.
If caller runs (e.g.) a Python script, it's not very helpful to a human log
reader to keep seeing LLProcess instances logged as /pathname/to/python (pid).
If caller is aware, the code can at least use the script name as the desc --
or maybe even a hint as to the script's purpose.
If caller doesn't explicitly pass a desc, at least shorten to just the
basename of the executable.
This way a caller need not spin on isRunning(); we can just listen for the
requested termination event.
Post a similar event containing error message if for any reason
LLProcess::create() failed to launch the child.
Add unit tests for both cases.
That is, trying to instantiate a ReadPipeImpl while another already existed
would throw an LLEventPump::DupPumpName exception. Fortunately this behavior
is easily bypassed.
If it's useful to have contains() to tell you whether incoming data contains a
particular substring, and if it's useful for contains() and peek() to accept
an offset within that data, then it's useful to allow you to get the offset of
a desired substring within that data. But of course a find() returning offset
needs something like std::string::npos for "not found"; borrow that
convention.
Support both find(const std::string&) and find(char); the latter permits a
more efficient implementation. In fact, make find(string) recognize a string
of length 1 and leverage the find(char) implementation.
Given that, reimplement contains(mumble) as shorthand for find(mumble) != npos.
Implement find() overloads using std::search() and std::find() on
boost::asio::streambuf character iterators, rather than copying to std::string
and then using string search like previous contains() implementation.
Reimplement WritePipeImpl::tick() and ReadPipeImpl::tick() to write/read
directly from/to boost::asio::streambuf data, instead of copying to/from a
temporary flat buffer.
As long as ReadPipeImpl::tick() keeps successfully filling buffers, keep
reading. Previous implementation would only handle a long child write over
successive tick() calls. Stop on read error or when we come up short.
Also add "len" key to event data on LLProcess::getPump(). If you've used
setLimit(), event["data"].length() may not reflect the length of the
accumulated data in the ReadPipe.
Add unit test with stdin/stdout handshake with child process.
Add LLProcess::FileParam to specify how to construct each child's standard
file slot, with lots of comments about features designed but not yet
implemented. The point is to design it with enough flexibility to be able to
extend to foreseeable use cases.
Add LLProcess::Params::files to collect up to 3 FileParam items. Naturally
this extends the accepted LLSD syntax as well.
Implement type="" (child inherits parent file descriptor) and "pipe" (parent
constructs anonymous pipe to pass to child).
Add LLProcess::FILESLOT enum, plus methods:
getReadPipe(FILESLOT), getOptReadPipe(FILESLOT)
getWritePipe(), getOptWritePipe()
getPipeName(FILESLOT): placeholder implementation for now
Add LLProcess::ReadPipe and WritePipe classes, as returned by get*Pipe().
WritePipe supports get_ostream() method for streaming to child stdin.
ReadPipe supports get_istream() method for reading from child stdout/stderr.
It also provides getPump() returning LLEventPump& so interested parties can
listen for arrival of new data on the aforementioned std::istream.
For "pipe" slots, instantiate appropriate *Pipe class.
ReadPipe and WritePipe classes are pure virtual bases for ReadPipeImpl and
WritePipeImpl, respectively: all implementation data are hidden in the latter
classes, visible only in llprocess.cpp. In fact each *PipeImpl class registers
itself for "mainloop" ticks, attempting nonblocking I/O to the underlying
apr_file_t on each tick. Data are buffered in a boost::asio::streambuf, which
bridges between std::[io]stream and the APR I/O calls.
Sanity-test ReadPipeImpl by using a pipe to absorb the Python "SyntaxError"
output from the successful syntax_error test, rather than alarming the user.
Add first few unit tests for validating FileParam. More tests coming!
When we reimplemented LLProcess on APR, necessitating APR's funny callback
mechanism to sense child-process status, every isRunning() or getStatus() call
called the APR poll function that calls ALL registered LLProcess callbacks. In
other words, every time any consumer called any LLProcess::isRunning() method,
all LLProcess callbacks were redundantly fired. Change that so that the single
APR poll function is called once per frame, courtesy of the "mainloop"
LLEventPump. Once per viewer frame should be well within the realtime duration
in which it's reasonable to expect child-process status to change.
In effect, this changes LLProcess's public API to introduce a dependency on
"mainloop" ticks. Add such ticks to llprocess_test.cpp as well.
Include logic to engage Linden apr_procattr_autokill_set() extension: on
Windows, magic CreateProcess() flag must be pushed down into apr_proc_create()
level. When using an APR package without that extension, present
implementation should lock (e.g.) SLVoice.exe lifespan to viewer's on Windows
XP but probably won't on Windows 7: need magic flag on CreateProcess().
Using APR child-termination callback requires us to define state (e.g.
LLProcess::RUNNING). Take the opportunity to present Status, capturing state
and (if terminated) rc or signal number; but since most of the time all caller
really wants is to log the outcome, also present status string, encapsulating
logic to examine state and describe exited-with-rc vs. killed-by-signal.
New Status logic may report clearer results in the case of a Windows child
process killed by exception.
Clarify that static LLProcess::isRunning(handle) overload is only for use when
the original LLProcess object has been destroyed: really only for unit tests.
We necessarily retain our original platform-specific implementations for just
that one method. (Nonstatic isRunning() no longer calls static method.)
Clarify log output from llprocess_test.cpp in a couple places.
Windows 7 and friends tend to create a process already implicitly allocated to
a job object, and a process can only belong to a single job object. Passing
CREATE_BREAKAWAY_FROM_JOB in CreateProcessA()'s dwCreationFlags seems to
bypass the access-denied error observed with AssignProcessToJobObject()
otherwise.
This change should (!) enable OS lifespan management for SLVoice.exe et al.
On Posix, these and the corresponding getProcessID()/getProcessHandle()
accessors produce the same pid_t value; but on Windows, it's useful to
distinguish an int-like 'id' useful to human log readers versus an opaque
'handle' for passing to platform-specific API functions. So make the
distinction in a platform-independent way.
On Posix platforms, the OS argument mechanism makes quoting/reparsing
unnecessary anyway, so this only affects Windows.
Add optional 'triggers' parameter to LLStringUtils::quote() (default: space
and double-quote). Only if the passed string contains a character in
'triggers' will it be double-quoted.
This is observed to fix a Windows-specific problem in which plugin child
process would fail to start because it wasn't expecting a quoted number.
Use LLStringUtils::quote() more consistently in LLProcess implementation for
logging.
If LLProcess can't set the right flag on a Windows Job Object, the object
isn't useful to us, so we might as well discard it.
quote() is sufficiently general that it belongs in LLStringUtil instead of
buried as a static helper function in llprocess.cpp.
Much as I dislike viewer log spam, seems to me starting a child process,
killing it and observing its termination are noteworthy events.
New logging makes LLExternalEditor launch message redundant; removed.
The idea is that, with the right flag settings, this will cause the OS to
terminate remaining viewer child processes when the viewer terminates --
whether or not it terminates intentionally. Of course, if LLProcess's caller
specifies autokill=false, e.g. to run the viewer updater, that asserts that we
WANT the child to persist beyond the viewer session itself.
Nat Goodspeed