The UnitBase reference must be a member of the class.
Change-Id: Ia0b10ccb7f0f3419470f014a23c7a48d3b390239
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
In WebSocketHandler, we often already have
locked the weak_ptr, so we can pass that
shared_ptr around.
Change-Id: Iaedceff0acbfd747bdf89771f9309ff6f6642b53
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
This merges the two constructors, since they
overlap and initialize most members identically.
Change-Id: Idd2af580606be92bb5733b1a0f1e1e63b84b6498
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
This avoids wakeup warning us that the thread is not started to wakeup.
Change-Id: Ifd5c483717024cb1c30521433d0d10acead01e2f
Signed-off-by: Michael Meeks <michael.meeks@collabora.com>
With multi-tests, the instance of UnitBase may
change while an old test thread is still running.
Here we capture the instance of the test so we
do not publish events into a different (newer)
test.
Change-Id: I3ca2c058cbd81d4077b140f24f60f3f20f12b9b5
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
If the connection is closed right after the response was sent
then it's wise to add 'Connection: close' header to the response
so that the client optimizes its behaviour: e.g. does not reuse
the socket for further http requests. Normally a client should
retry a request if the reuse of an old socket fails and that
should solve the problem but still this is an overhead.
Signed-off-by: Gabriel Masei <gabriel.masei@1and1.ro>
Change-Id: I29f1498610c567024def3beb1ad7014f2c15a232
ERR_get_error cleared the returned error,
which meant the subsequent call to print
the error in string form, using
ERR_print_errors_cb, didn't find anything
to report.
Change-Id: If131a8cc0d2c1d8bbf705ed38f144b38abf6c8c6
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
This was a regression introduced in
14d96957cd.
The result was that multi-part logs were
not flushed and therefore not written to
the output.
Change-Id: Ib22a1da83209b00872d91ca05a940226cbb4552d
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Eg. if only the version was changed, config.h was updated, and
everything was recompiled.
New structure, maintain these manually:
config.h.in - for configured variables that are stable if
configuration parameters are unchanged
config_version.h.in - for version/hash related variables
----
config_unused.h.in - still generated by autoheader with all
variables, don't use it
Signed-off-by: Aron Budea <aron.budea@collabora.com>
Change-Id: Id9a50a9f1e798a3b3814778d8683b7d7cb57bb29
This improves the performance of fuzzing
and removes the checks that can't be guaranteed.
And fixes a test failure.
Change-Id: I987fe15b098c00d9a3d60077f0581d2ef35e306c
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
This should cap the loop either when we timeout
or when termination is flagged.
Change-Id: I78a6b05df99d6bac5286e4b1e4957829ea91c500
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
This is a full round-trip http fuzzer.
It can achieve >1000 iterations per second
on a single 2 Ghz core, even while going
through the network loopback layer.
The advantage is that more networking code
is fuzzed this way, including not just
the http code, but also the sockets.
Change-Id: I75d21bd0e25221ee6621097a2605d62c4bb2ae4d
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Without this patch, we may end up repeatedly
sending close-frame. This impacts tests only.
Change-Id: I3c9a0cb14c9103d002508b1199647f6ff9761230
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
We cannot invoke the socket from outside the
SocketPoll and we must flag for async shutdown.
Change-Id: I223e24f5fe48a79a9d751c7cf5cdf9d720c22f7d
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
We now use our http::Socket and SocketPoll
instead of Poco.
Change-Id: I728f9c975b67da76dbbc78e94f1fb5f224c9bc57
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Before ed5569da30
we invoked handlePoll on all sockets in
each poll. But that's not true anymore.
Since i = _pollStartIndex; (line 432),
when we assigned previ = i; (line 451)
we made previ == _pollStartIndex, which
immediately broke the loop
while (previ != _pollStartIndex)
at line 438, causing handlePoll to be
called on one socket at a time (and
likely *not* the socket that had an event).
It's inefficient to poll and ignore the
result.
Now we do handle revents on multiple
sockets correctly and efficiently for
each ppoll() by going through all sockets
and handling based on whatever revents
gave us. We need to invoke the handler
of each socket to handle timeouts (i.e.
if we don't call the handler until
the socket gets an revent, we wouldn't
be able to support timeouts properly).
Notice that we still start at
a different socket each time (although
it's not entirely clear what advantage
that has, it's maintained all the same)
as the aforementioned patch does.
Change-Id: I9c0a482792aee7df5d44f4b63f173e574db36f97
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Apparently this is an issue for proxy.php.
Change-Id: I18090f17f22fe3145e7fa2364992a34a7e722f1b
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
No need to complain when bind() didn't fail.
And assert when/if we are going to overflow.
Change-Id: I66f5ff1963c0afc6652afb5a7d017e197d945736
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
More consistent logging of the socket FD.
Change-Id: Iee24ad4eb067969abfc06f78a768563351d264c7
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
A corner case where read doesn't return 0
on a closed socket can result in a stuck
state where we attempt to read when we
do get ECONNRESET (which we didn't check
on reads).
This makes the interface of readIncomingData
the same as writeOutgoingData by returning
the actual return value of the read syscall.
So now we handle both return 0 as well as
error codes returned on failed calls.
The logic hasn't changed, just that now
we handle errors better and similar to the
write case.
Change-Id: I0b38a63da4e6c92a482948478d5d8d446e0b8b58
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
This enables code that was protected with
EnableExperimental in the socket logic (and one
case in DocBroker). These changes are now deemed
safe to enable permanently.
Change-Id: Ie62f5d7bd281ade90f38d654b51b104b8d1f14bc
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Without knowing whether the write succeeded
or failed, we cannot trust errno has our
error or some earlier and unrelated error.
This was caught when there were two sockets,
one of which disconnected. The write to the
disconnected one returned -1 and set errno
to ECONNRESET. We subsequently wrote to the
second socket, which succeeded. However,
because errno wasn't reset, and since
writeOutgoingData didn't return anything
to indicate the success, errno's ECONNRESET
value meant the second socket was also
disconnected, which was incorrect.
writeOutgoingData now returns the last return
value from writeData so we can make informed
decision as to whether to check errno or not.
Also, to avoid incorrecly propagating errno,
we now capture errno only when readData and
writeData return -1. This has the nice
side-effect that we reset errno to 0 when
no errors occur during our call.
Change-Id: I911b31390f37cc71938bc4a6ae75393dbf24bb9d
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
OpenSSL 3 deprecated the manual DH parameter
functions. Instead, it encourages the use
of the built-in parameters. Since this
API also works on the 1.1 version, we only
need the manual parameters for older versions.
Change-Id: I900cc11c3ca09f1d85b7d88cfbf537d802f69846
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Since both implementations are identical, there
really is no benefit to having two version.
Change-Id: I4a5288243291c0d5706df8e8870b918fab425317
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
The last burst of data arrives with POLLHUP.
If during the read we hit any errors, including
EAGAIN, for whatever reason, we may lose the
data if we don't attempt reading again later.
The only way to guarantee that we do not have
any more data to read, when we get POLLHUP,
is to poll until we either get no POLLIN or
read returns 0 indicating the socket is closed,
Oddly, read(2) sometimes returns 0 without
POLLHUP ever being set, and sometimes we
do get POLLHUP while POLLIN still set. This
implies that poll and read aren't consistent
in how they detect and report the close
condition. Luckily, with this patch we can
handle all cases without any complications.
This was caught through simulateSocketError,
which when it injects EAGAIN on the very
last read, when the socket is closed, causes
error as the response is partially or
completely lost (because we never read it).
The behavioral change should only be
observable with EnableExperimental.
Change-Id: I77260f98d5dd5050c5f9b202b9effd501191336b
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
This extends the detection of connection termination
for writes via ECONNRESET and EPIPE.
For some reason ECONNRESET was not detected.
This patch avoids extra polling to detect connection
termination and makes the cleanup faster.
The changes are guarded by EnableExperimental to avoid
any unexpected change of behavior.
Change-Id: I2ae3803f025d3cf756f86460e47aedc984249509
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Ensure that rendersearchresult: is terminated, and identified as binary.
Change-Id: I84e83c63351cd0aac0923bf7c833b14c1be2c051
Signed-off-by: Michael Meeks <michael.meeks@collabora.com>
Since the hostname argument is passed
to both the base class of SslStreamSocket
and SSL_set_tlsext_host_name, and since
the base class's getter, also called
hostname(), is hidden by the argument,
we cannot move it.
An empty hostname can result in 403 Forbidden
from the server due to missing Server Name
Indication (SNI).
Change-Id: I27990f64f17ec3c81a4dd543a078807629cd0c20
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
And improved socket logging in general while
making them more consistent.
Change-Id: I1ed7f2561476ca5370af91079d5d616804396f8e
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
While SSL is handshaking, there can be no general
application data communication. During that early
stage of connecting we have data to send (the
request, headers, etc.) and so we poll on POLLOUT.
Naturally, we also always want to poll on POLLIN,
because we can never know when there is data to
read (especially true for web-sockets).
The problem is when SSL will not send data just
yet because it is handshaking. It is typically
waiting for handshake negotiation data to read,
so when we POLLOUT, poll immediately returns, but
writing (via SSL_write) fails with WANTS_READ
error. This goes on in a busy-loop until the
negotiation data is available for read and the
handshake is completed. Very inefficient.
The solution is to poll on whatever SSL needs
during the handshake, exclusively. Once the
handshake is complete, we poll on whatever we
need. However, SSL can renegotiate at any time,
so we also merge with what it needs.
In addition, we avoid the unnecessary read when
poll doesn't give us POLLIN in revents, since the
read will more likely than not fail (except in
the rare case when data becomes available in the
interim). Notice that SSL_read will return
SSL_WANTS_READ when there is no data, which
is misleading (since SSL isn't in need of data to
read at all, nor are we, for that matter).
Best not to do noisy reads unnecessarily.
These changes are disabled by default and can
be enabled via the experimental_features option.
Change-Id: I6a7ed7d871ed257b30062cc720a8b8c7acbab3b7
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
This prevents the kit from becoming a zombie
when it has an early failure to connect to WSD.
Change-Id: I4e8203b2cc3d80822308236e5b47be0c0c96e9ae
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
In 'debug' log-level we expect a detailed, but
still readable output. Having one area with
disproportionately large number of logs reduces
the overall utility of the log output.
This patch reduces a number of redundant log
entries, including errors that are already
logged. It also reduces the level of some
others from 'information' to 'debug' and
from 'debug' to 'trace'.
The goal is to make 'debug' level as useful as
possible to read the progress and be able to
understand what was going on, such that one is
able to decide which area to dig deeper into.
Then, trace level could be used to get more
insight into that area, if necessary. For
example, when investigating a test failure,
one first enables 'debug' logs and reads through.
Once a section between two debug entries is
identified as being of interest, enabling 'trace'
level logs becomes more productive as it's
now possible to easily reach the first DBG
entry and read through until the second one.
It's unfortunate that we don't have per-area
control for enabling/disabling logs, so it
is common to see more and more 'debug' log
entries added all around, making logs
less and less readable.
It is also a limitation of the levels we have
that we really only have 3 usable levels:
one, two, many. That is, 'information' for
the most important events, 'debug' for
technical details needed to investigate issues,
and 'trace' for everything else. ('warning'
and 'error' aren't really 'levels'; they have
semantics that makes them special-cases.)
So we have to avoid degrading one into the
other, or have differences without distinction.
If any of these entries are needed to be
displayed more frequently, changing them
back to 'debug' or even 'information' should
be done. Though for me they seem special
cases that don't benefit most log readings.
Change-Id: Id2c6a9dc027483b81a066b0b4b50a298c5eff449
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
This adds dumpState to DocumentState, SaveManager,
and StorageManager classes, and dumps all the
missing members.
Also, normalize the format and make it symmetric
and consistent.
Change-Id: Ie0cc8e07d13de60c33d64cd621abf4e815a4ef94
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Now the remaining tests that didn't have a name
get one and the logs are unified between old-
and new-style tests. Mostly.
This makes sure that all logs and assertions
properly log the test name and make test
failures easier to debug and fix.
Change-Id: Id159ffacc81642a6ec594c5143498408adab67cf
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
This reduces some logs from debug to trace,
especially in Socket where the amount of
activity is high, but the logs in question
don't add value in debug level.
Also supressed an error when superfluous.
This makes debug-level logs much more readable
and relevant to the main events.
Change-Id: Ie2698b56f320d15a09060cf2877477d9da0e1582
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Don't stop processing all messages when one caused an exception.
Try/catch where messages are processed in the loop so we don't
ignore other messages:
* handleTCPStream
* handleIncomingMessage
Signed-off-by: Szymon Kłos <szymon.klos@collabora.com>
Change-Id: Idb8c9cf72a7912e7e508e429821b4ee32c9b122b
This avoids memory hanging around, _buffer.size() is not data
left as seen in pkill -USR1 output:
coolwsd[25677]: Buffer size: 0 offset: 421811
Change-Id: I0c2eef9435230cf864854cbc95cd587e39d240ac
Signed-off-by: Michael Meeks <michael.meeks@collabora.com>
SteamSocket::eraseFirstInputBytes() removes from the beginning
of std::vector, which is generally slow. If the buffer becomes
too big, which it may under a load, then the function will get
slow, which in turn will likely lead to the buffer getting even
bigger because of accumulated backlog.
The Buffer class is optimized for removal at the beginning,
so use it instead of std::vector, including some API additions
for it to be an in-place replacement where it's used.
Signed-off-by: Luboš Luňák <l.lunak@collabora.com>
Change-Id: I4cf7ec56c908c7d3df391dc3f8e230ad32abb162
We don't want to always start with the most recently added socket.
Signed-off-by: Tor Lillqvist <tml@collabora.com>
Change-Id: Ic4b4bf6c19c5d119e6e6f9b398789a4c77b47a10
HA Proxy doesn't tolerate this very happily cf.
https: //serverfault.com/questions/502443/ignore-port-numbers-in-haproxy-host-header-matches/502630
Change-Id: Id285f8acd0e168a734cabd9eccc4a01fe323ab84
Signed-off-by: Michael Meeks <michael.meeks@collabora.com>
For some servers we receive failure with HTTP 403 Forbidden in WOPI::CheckFileInfo
"Reason: The client software did not provide a hostname using Server
Name Indication (SNI), which is required to access this server"
fixes#2771 : https://github.com/CollaboraOnline/online/issues/2771
Signed-off-by: Szymon Kłos <szymon.klos@collabora.com>
Change-Id: I761b179580481f8882a4526c1d8be4f1c14ad929
Add profile zone around kitPoll function and SocketPoll constructor to
improve trace event generation
Signed-off-by: Gopi Krishna Menon <krishnagopi487.github@outlook.com>
Change-Id: Ia46e9add59a57935997649fe39861a8d851e1ff0
Also clear its input buffer explicitly.
Change-Id: I8badbb96d98eaf10433a65fcfd13b0d6d5893594
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Ignore input in a somewhat gentler way.
Signed-off-by: Michael Meeks <michael.meeks@collabora.com>
Change-Id: I758302dc3bb1aa87f9fbfa726f73f4b9339e08c2
This is really a garbage in -> garbage out situation.
Signed-off-by: Miklos Vajna <vmiklos@collabora.com>
Change-Id: Ic1c33f44081f259e5cf5994ad901e1593fe8dfcf
Given that "digit" is always less than 16, we can avoid the "- digit"
in the condition (because the division turns that into 0 anyway), making
it effectively constexpr.
Signed-off-by: Miklos Vajna <vmiklos@collabora.com>
Change-Id: Iaf9e53d3543f2237c00768f214114a02a4746020
...Thread.
Conflicts:
net/Socket.hpp
net/SslSocket.hpp
Signed-off-by: Dennis Francis <dennis.francis@collabora.com>
Change-Id: I92b8f4b52e7bd60b69305c1916eed8a14a4c1560
This can be used to accelerate unit tests by allowing us to ensure
that all pending work is done between a 'processtoidle' message being
sent and an 'idle' being returned. This can be much faster than waiting
a defined length of time.
Change-Id: I6d96de205d1fd34a22e0ddbd2709bf806a97b33c
Signed-off-by: Michael Meeks <michael.meeks@collabora.com>
We currently only recognize HTTP/x.y. Anything
else is invalid.
Change-Id: I45fa7e33052e3847b8a2d9768b12d8eac43f6b0a
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
The Socket instance owns the protocol handler
by design. This implies that the Session,
which is also the protocol handler, cannot
itself own the Socket.
Here we change that ownership into a weak_ptr
and pass the exclusive ownership of the Socket
to SocketPoll. We track the connection state,
so we can create a new socket when we are
disconnected.
Change-Id: Id6bf6506b0c8e2e0564f91977e1e7189d934a68f
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Setting the state to Error in the parser has the
unwanted side-effect of not triggering the proper
callbacks. The better design is to observe the API
and return error (-1) upon hitting parsing errors.
The caller will then invoke the finishing function
that sets the state and triggers the callbacks.
Change-Id: I4957b0fb431f477d71fb21a8dd601d132c3d6a21
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
When the socket is destroyed, it may invoke
the onDisconnect handler, which updates
the Response instance. If the order of
destruction is incorrect, we may have
a null instance.
Also make the State atomic to make sure
its visibility is consistent across threads.
The shared_ptr instances will fence and will
always be consistently visible across threads,
but if the State isn't in sync, then we can
seem to have an inconsistent state. The State
instance will be set and accessed from different
threads in an async request.
Change-Id: Ia2a48285d140646f3640f59634e14a46e4a4d640
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
If we haven't had a complete response or a timeout,
then we have an incomplete transmission, which is
an error.
Change-Id: I1b775760186021e0a704eba2c9f3647efa4f5ed0
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Reusing a SocketPoll is encouraged rather than
creating throwaway instances. Session now supports
external SocketPoll instances for the sync members
as it does for the async (which is unavoidable).
Also, use TerminatingPoll to break the poll
when the terminating the process is flagged.
Change-Id: Ide6f674080ce20bb764c482871e48c94d83d4214
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
While we are processing data in the poll handler, we can have
another helpful thread from the core sending things to us:
loolforkit(Socket::assertCorrectThread(char const*, int))
loolforkit(WebSocketHandler::sendFrame(std::shared_ptr<StreamSocket> const&, char const*, unsigned long, unsigned char, bool) const)
loolforkit(WebSocketHandler::sendMessage(char const*, unsigned long, WSOpCode, bool) const)
loolforkit(Document::postMessage(char const*, int, WSOpCode) const)
loolforkit(Document::sendFrame(char const*, int, WSOpCode))
loolforkit(ChildSession::sendTextFrame(char const*, int))
loolforkit(ChildSession::loKitCallback(int, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&))
loolforkit(Document::GlobalCallback(int, char const*, void*))
/core/instdir/program/libsofficeapp.so()
/core/instdir/program/../program/libfwklo.so()
/core/instdir/program/libscfiltlo.so()
/core/instdir/program/libexpwraplo.so()
/core/instdir/program/libooxlo.so(oox::core::FastParser::parseStream(com::sun:⭐:xml::sax::InputSource const&, bool))
/core/instdir/program/libooxlo.so(oox::core::FastParser::parseStream(com::sun:⭐:uno::Reference<com::sun:⭐:io::XInputStream> const&, rtl::OUString const&))
/core/instdir/program/libooxlo.so(oox::core::XmlFilterBase::importFragment(rtl::Reference<oox::core::FragmentHandler> const&, oox::core::FastParser&))
/core/instdir/program/libscfiltlo.so()
/core/instdir/program/libcomphelper.so(comphelper::ThreadTask::exec())
/core/instdir/program/libcomphelper.so()
/core/instdir/program/libuno_salhelpergcc3.so.3(salhelper::Thread::run())
/core/instdir/program/libuno_salhelpergcc3.so.3()
/core/instdir/program/libuno_sal.so.3()
/lib/x86_64-linux-gnu/libpthread.so.0()
/lib/x86_64-linux-gnu/libc.so.6(clone)
This is most likely from eg. threaded parsing of various file formats,
or progress messages while loading / saving, sent while the SolarMutex
is locked.
We assume that:
+ there will only ever be one thread running inside kitPoll
this is enforced by vcl/headless.
+ and so, we can safely mutate SocketPoll and Socket buffer
state from this thread.
+ there are only two entry points for callbacks from threads
in core - which are both instrumented, and if necessary
their work is shifted to this thread.
Memory corruptions matching the above trace, suggesting that
multiple writes have collided between kitPoll and an event
callback have been seen in the wild.
Change-Id: I5b084cbfec1ea326b6e17c9e5590a8c8e35b3854
Signed-off-by: Michael Meeks <michael.meeks@collabora.com>
We send lots of large tile data this way, this should help
accelerate tile data transfer marginally.
Change-Id: I1deab7845c09dc65b1f44e9f9fc762f70ce94cc3
Signed-off-by: Michael Meeks <michael.meeks@collabora.com>
Avoiding default values for parameters makes the code easier to read.
Especially true for bool parameters. But sure, just a question of
taste.
Change-Id: I473f70bdfafe3a9ccfb325def8760d78fee7e9a6
Signed-off-by: Tor Lillqvist <tml@collabora.com>
Seems like this was a bug in clang-format
that is fixed in clang-12.
Change-Id: I8144179d2b290f5d875af41dbd44dedcf4152639
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Instead of supporting char appending,
we now support appending literal strings
with size capturing. This covers both
single- and multi-character appending
without the need for an explicit size.
Change-Id: Iee2c20b7aa2cdb6863c88e91cd770205719c2ba6
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
(cherry picked from commit 96e3e88a014960a5249fc83a5b2f3b785ac8737d)
this prevents compile error with gcc (Ubuntu 5.4.0-6ubuntu1~16.04.12)
../net/HttpRequest.cpp:340:9: error: assuming signed overflow does not occur when assuming that (X + c) >= X is always true [-Werror=strict-overflow]
if (off + VersionLen >= available)
Signed-off-by: Andras Timar <andras.timar@collabora.com>
Change-Id: I46df5d43d0997f6a1e61c05a4df7f9d71af65511
net/HttpRequest.cpp:126:19: error: declaration of 'end' shadows a member of 'this' [-Werror=shadow]
kit/Kit.cpp:163:29: error: missing initializer for member 'statfs::f_bsize' [-Werror=missing-field-initializers]
kit/Kit.cpp:171:14: error: 'OVERLAYFS_SUPER_MAGIC' was not declared in this scope
Signed-off-by: Miklos Vajna <vmiklos@collabora.com>
Change-Id: Ibbd35ab5af3adad403ed22a0aeb70917b9e21970
When writing to the socket, it's always more efficient
to fill the buffer up to the hardware limit for each
write. This is doubly important for efficiency with
SSL, due to the overhead of encrypting multiple
small buffers instead of one large one.
Currently we don't write more than one message
at a time, primarily due to limitations in
the Poco sockets in the unit-tests, which
have a hard time consuming multiple WS frames
with a single poll (subsequent calls to poll
doesn't enter signalled state until new data
arrives, possibly because the data is read and
buffered internally, making the whole scheme
of using poll unreliable and meaningless).
Change-Id: Ic2e2cf1babfb5ab4116efd93f392977ba234d92b
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Since it's most efficent to send full blocks at
once (especially so when SSL encryption is done),
we now perform write when the buffer has fewer
than the send-buffer-size bytes. Previously,
we were only getting more data to write to the
socket when the buffer was completely empty,
potentially wasting valuable opportunity to
write more data to the socket while we're at it.
Change-Id: I69a18a042dc2e551ebc5e1af40dae091da3f3d13
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
SocketPoll is captured as weak_ptr into
WebSocketSession and a much better shutdown
support is now available. The new logic
can do async-shutdown after flushing and
will do sync-shutdown if SocketPoll is
no longer around.
Change-Id: Ia206cab58a13f20f7aeb3a6d8c57afee731c8231
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
std::atoi() assumes a null-terminated string and our strings are not
always null-terminated. So add a version that takes a length parameter,
this way we don't have to copy strings around.
Also switch to this in http::StatusLine::parse().
Signed-off-by: Miklos Vajna <vmiklos@collabora.com>
Change-Id: I449b356c1b9948c562434618596e8e3b38656088
And remove the httpheader one, which is not useful, since it uses Poco
for the actual parsing, it did not find anything. (If we switch away
from Poco there in the future, it's easy enough to restore it.)
Also fix some problems found by the fuzzer.
Signed-off-by: Miklos Vajna <vmiklos@collabora.com>
Change-Id: I254247c46ecc78c9c3e75aac4f10c441b0e10fb3
And fix an unhandled std::length_error it found.
Signed-off-by: Miklos Vajna <vmiklos@collabora.com>
Change-Id: I571cdd71caeda84820f2c64088966936637ce2bf
This makes it simpler to call poll repeatedly
within a given timeout, by first calculating
the deadline and passing it to poll to handle
the math.
Change-Id: I7a9311286ec368c853de40f10523b7b98c8f8106
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Useful for debugging, especially for unit-tests.
Change-Id: I820e584203a643a396d26c8817b9af1e750232c7
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Now that the WebSocket functionality is available in
WebSocketHandler, we can cleanup Socket.
Change-Id: Ia3703ab5a2f00fb835eb80493bc400a91dabefad
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
A utility class to create a WebSocket connection based
on WebSocketHandler, http::Request, and SocketPoll.
Change-Id: I00dfd6fc88e210860e0e378587ef1c0a0a65bb5c
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
This adds support to Web-Socket upgrade handshake
validation. We never had a proper client-side
Web-Socket, so never needed this.
Change-Id: If4b8e42daeab13a430179128750f6a4f17452f28
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
This simply reduces the public surface area
of the WebSocketHandler class.
Change-Id: I93be58923f429bceab74a9e239ad0fad11390449
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
According to the RFC, HTTP/1.1 assumes persistent
connection unless the client or server sends
'Connection: close' header. See
https://tools.ietf.org/html/rfc7230#section-6.3
Now we fully support disconnecting when we
get this header, which is important for when
reusing the same http::Session (we need to
reconnect in such a case).
Change-Id: Ib278eff21be32c1414478296e90e2e807ab24132
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Since the return value of syncRequest and syncDownload
was derived from Response anyway, there is no reason
to have multiple values for callers to look at.
This simplifies the API.
Change-Id: I0f136e515dd0ef6eda84f6a7cd662b260809d2f1
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Improved handling of http::Request and http::Response
in Socket directly with better error handling and
automatic socket shutdown on error.
We shouldn't need to manually serialize into the
socket's buffer when sending a Request or Response.
Change-Id: I6dfcd2fe4f6b88cd2fa8c749045ce25a8c52fe9a
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
We shouldn't need to erase from the in-buffer manually
anywhere. This utility encapsulates this functionality
well and protects against off-by-one errors.
Change-Id: I53616444ac9811a1e22d23d96a9cac8965b8518b
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
A full URI with scheme can now be used to
create http::Session instances.
Change-Id: I3e556e3e1aa247ccb93749d9bd6736a15290f442
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
The Socket now accepts http::Response and
serializes it before sending it.
Change-Id: Id36e2d91b21d168da72ccdbd7e509ec08021b78e
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
It's now possible to set a body in http::Response
and it will get serialized out with the header.
Change-Id: If0f01a226bbd7b5ab092d0a246c8b46553a107e0
Signed-off-by: Ashod Nakashian <ashod.nakashian@collabora.co.uk>
Ashod Nakashian
Michael Meeks
Gabriel Masei
Aron Budea
Andras Timar
Henry Castro
Szymon Kłos
Gleb Popov
Luboš Luňák
Tor Lillqvist
Gopi Krishna Menon
Miklos Vajna
Dennis Francis
merttumer