Disable SSL compression by default since this appears to be the a likely
cause of the currently hyped CRIME attack.
Change-Id: I515fcc46f5199acf938e9e880a4345f2d405b2a3
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Peter Hartmann <phartmann@rim.com>
The testAndSet operation would mean another thread could see the value
of 1 and proceed to write(2)/sendto(2) before SIGPIPE had been ignored.
If the pipe or socket were already closed by then, a SIGPIPE would be
delivered to the application with its default action: terminate.
Change-Id: I62dc8f5fa14c1dd453d13e4053c642bd78fbc468
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Reviewed-by: Peter Hartmann <phartmann@rim.com>
"HTTP/1.1 100 CONTINUE\r\n"
If the header from a server is splitted between two packets
the first packet contains "HTTP/1.1 100" and the second one
contains " CONTINUE\r\n", one space (0x20) is skipped. After
processing the line looks in this way "HTTP/1.1 100CONTINUE".
QHttpNetworkReplyPrivate::readStatus(QAbstractSocket *socket)
is called twice, if a http header is splitted as above.
The function always removes whitespace from the beginning of a packet,
even if it is the second part of a http header.
QHttpNetworkReply returns QNetworkReply::RemoteHostClosedError
due to damaged http header during processing.
Improvement of unit test.
Task-number: QTBUG-27161
Change-Id: Ifc2949f62473209b4032185effbf5078b4130cda
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Use an include guard in headers to ensure the header is not included
more than once. Make the header guard match its file name.
Also, cpp files should include their own headers first (but below config.h)
Change-Id: Iecf5da23c0f8e6d457f67657b88ef7557bde9669
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Make C++ class constructors that can be used with only one
required argument 'explicit' to minimize wrong use of the class.
Change-Id: I12ad5b6eb1794108c6b7464a2573e84068733b03
Reviewed-by: Sergio Ahumada <sergio.ahumada@nokia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
All qdocconf settings that should be used by all Qt modules are now in
qt-module-defaults.qdocconf.
Change-Id: I2a0315a55db3fcbb0160c4392d2da98611043d83
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@nokia.com>
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
The signal has nothing to do with any action on the remote side. It is
emitted when the class has written the bytes to the operating system.
More likely than not, the bytes have not been even sent yet, much less
read by the other side.
Change-Id: Ia04d37ffc8c0788173d3d29f49c5759bcdef6afa
Reviewed-by: Qt Doc Bot <qt_docbot@qt-project.org>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Also fixing some includes and re-enabling the test
Change-Id: I4a061e106c2e55db39b8000729737a93e3d7714a
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
This is an automated change performing the following replacements:
join\("(.)"\) -> join('\1')
join\(QLatin1String\("(.)"\)\) -> join(QLatin1Char('\1'))
join\(QStringLiteral\("(.)"\)\) -> join(QLatin1Char('\1'))
Change-Id: I5b75877ba192fa1357e67fee70dff7c0475991e8
Reviewed-by: João Abecasis <joao.abecasis@nokia.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
According to Qt doc, "Network proxy is not used if the address used in
connectToHost(), bind() or listen() is equivalent to
QHostAddress::LocalHost or QHostAddress::LocalHostIPv6. This is not the
case in current implementation.
Change-Id: I6b8a40c1e8bd8aad9504d8f939b87eda6e93337c
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Even though I really think the change was the right thing to do, it seems
like Windows people don't like this change because of some Windows Data Types
specific rules.
This reverts parts of the commit 56d5c909af.
Change-Id: I2c67d9b1bab36fc63937ef386aef56d2a4472a04
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
The TRUE and FALSE macros are obsolete and should be replaced with
true and false (all lower case) respectively.
Change-Id: Iee352e8173500683e6319be0abbf5bacf29016e0
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
To fix a compile error when QT_NO_NETWORKPROXY is defined.
Change-Id: If0433387cff2e6574fe24721a0ce2bfc41b0eb47
Reviewed-by: Markus Goetz <markus@woboq.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
https://codereview.qt-project.org/#change,33974 only works in the case
of a developer build. Not in a normal prefix build.
Change-Id: I3a3e5029cefaa9f83c5deb71665f0efa9d812819
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@nokia.com>
We should be using the global qdocconf for the common variables. This
change also allows you to just specify -installdir without using a
templatedir.
Change-Id: I207d279d9b5199212e896fc5ccab5c212b1896c6
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@nokia.com>
Add SslProtocol enums TlsV1_1 and TlsV1_2 and use the appropriate OpenSSL
methods when they're selected (TLSv1_1_client_method, TLSv1_2_client_method,
TLSv1_1_server_method and TLSv1_2_server_method). This allows us to
explicitly use TLS 1.1 or 1.2.
Task-number: QTBUG-26866
Change-Id: I159da548546fa746c20e9e96bc0e5b785e4e761b
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
These two variables are only ever accessed under mutex protection,
and don't otherwise look like they could be changed by the hardware,
so remove the volatile qualifier.
Change-Id: I714451bb3e80778b971a901d53fe13e1b01dd84f
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Since there is non-atomic data that is protected by 'triedResolve',
the (outer) read from triedResolve needs to have acquire, and the
store needs to have release semantics. The release implied by the
mutex unlock is not good enough because it only synchronises-with
the locking of the same mutex, which not all threads execute.
Change-Id: If46b3ea6ccfdd66ca41ce44d4f45bef2c2c30f72
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
A volatile bool read/store is documented on MSVC to have acquire/release
semantics, respectively, but that doesn't need to be true for MinGW, so
use explicit memory ordering.
Apply the same fix to the Unix implementation, too.
Change-Id: Ica466cec50beed830aafa4e3384d82f02e1a47e8
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
There are too many references to the QWidget lib documentation
in there. On the other hand this keeps snippets working.
Change-Id: I7dd63b7fba1758accea2663f7b427940a8857e32
Reviewed-by: Martin Smith <martin.smith@nokia.com>
Network operations whose content size exceed 2G are not unheard of, and
they probably shouldn't be cached, so use the correct toX() function, one
that doesn't truncate the value in the header.
Change-Id: Ie4e68a3120c85952dd2cc9feb73c5844e20b9aa3
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
@p replaced by \a, \QD replaced by {Qt Designer}, couple of enums etc.
Change-Id: I315510690bf52e42db519292b4122fa24c73bbc9
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@nokia.com>
The hashing functions for QDateTime and QHostAddress did not get the
noexcept keyword because they might allocate memory. QDateTime doesn't
do it now, but it could in the future. QHostAddress does allocate
memory today.
Change-Id: Ia5f80942944bfc2b8c405306c467bfd88ef0e48c
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
There's a Q_DECLARE_METATYPE at the end of this file.
Change-Id: Ib63aa4397ccba8740d46cf1ee7af712f75d05765
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Fixed some bad camel case, marked some parameters with \a,
and changed a \enum to a \typedef. Removed some obasolete
text.
Change-Id: I1e69dd9f3b74129230770addc80fe1c125faf268
Reviewed-by: Richard J. Moore <rich@kde.org>
Version numbers for CE are higher then for the desktop,
so insert a check for this, as the api is not working as
for windows 7 and above.
Change-Id: I79baba9183e11c714cffd0bb43327c197043d6e0
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
-fixed generated HTML file names
-removed qt-network group in qt/qtdoc and removed \ingroup instances
Change-Id: Ida5ded5da30e6db1978382f6742a5da8036e055e
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Reviewed-by: Martin Smith <martin.smith@nokia.com>
Reviewed-by: Laszlo Papp <lpapp@kde.org>
this is more logical, following the qt_plugin and qt_tool scheme.
Change-Id: Ib3b2abec6728cdab260e15128b1cd78e8e6f5d6a
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Under Windows it's quite possible that OpenSSL, DBUS or MySQL is
not installed into a central place. If -I and -L is
passed at configure time, it is added to all targets,
and if that path contained a conflicting header things would go
wrong.
Change-Id: Ic3338c49aa6eaa91b3abf5341e709ef604bf7aab
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
We need to handle FD_CLOSE separately on Windows as this will be sent
only once. When we get FD_CLOSE we need to check if there is more data
available for reading. It there is this might indicate that there is
another FD_READ that we need to handle after the FD_CLOSE. So in this
case we will manually create another close event.
Task-number: QTBUG-19409
Task-number: QTBUG-25386
Change-Id: Ie19906bc3f64fb6a85a508a5ab12caac5d70ccdb
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
msvc cannot use the same library for debug and release builds
if openssl libraries are linked statically into the network library.
Change-Id: Ic27ede2d9531b94aff4c50c1699947ce72caf286
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
If the error code from WSAConnect is WSAEWOULDBLOCK, then the
operation proceeds but the outcome is not known at that time. We then
check SO_ERROR's value to detect errors. But if that call returns 0
this could indicate that the value is still not know. In this case
we try one more time to increase the chance of getting the correct
value.
This fixed the tst_QNetworkReply::getFromUnreachableIp auto test
on Windows.
Change-Id: I25008aca062b2f823e3d93ebb0ae456d7e4a6ecc
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
The Q_DECLARE_METATYPE()s were outside of #ifndef QT_NO_SSL for classes
that are defined inside.
Expand the #ifndef block.
Change-Id: I45b73a24032fb2a79fd80d91282b782daa8a8f68
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
Reviewed-by: Rafael Roquetto <rafael.roquetto@kdab.com>
Using the nullary version has the advantage that multiple calls
during a program run are much more efficient, since an inlined
atomic is used to store the result. It also ensures that
Q_DECLARE_METATYPE(T) has been used, whereas qRegisterMetaType<T>("T")
will happily register anything. So I've added the macro where it
was missing, or moved it to a central place when it existed
hidden.
In tst_qnetworkreply, this became a bit tricky, because a private
header is conditionally included, so moved the Q_DECLARE_METATYPE()
into a conditional section, too.
Change-Id: I71484523e4277f4697b7d4b2ddc3505375162727
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
This is the result of running util/normalize --modify
from Qt 4.7 with manual review.
Change-Id: I3f89d5138ea9905c42ed581991426e72c90d4069
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Marc Mutz <marc.mutz@kdab.com>
Reviewed-by: Olivier Goffart <ogoffart@woboq.com>
URLINFO was removed from qfeature.txt but QT_NO_URLINFO macro was not
removed. As QUrlInfo is now private and only used for FTP, the class can
be disabled when FTP is disabled.
Change-Id: Ic63b066f8ff9ad1eea0073ab75b622e5739ac5a8
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
Reviewed-by: Jonas Gastal <jgastal@profusion.mobi>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Change-Id: I19d3b2e9a5180b13deb828b55195404ef20be295
Reviewed-by: Daniel Teske <daniel.teske@nokia.com>
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
There may be duplicated debug plugins, loading both will cause
crash.
Change-Id: Icc2a3643c318844bc7f2e149a6434e95de2449b2
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
If the Windows version doesn't support dual stack then we should bind
to IPv4. So we should change the socketProtocol to
QAbstractSocket::IPv4Protocol as well.
Change-Id: I7f9d0bdd861f82d720e347d3fa968198de720d1a
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
The reqExp used to handle wildcards in the path was broken. So we
always searched the working directory and not the specified path.
Autotest where passing because of a hack used for Windows paths
where we removed the first two chars in the path string.
This fix will not use nativeSeparators thus removing the Windows hack
and fix the regExp to match wildcard chars.
Task-number: QTBUG-23573
Change-Id: I56fadbb67f25b8ce9c0f17cb6232e0bdb9148b1c
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
If a request is aborted while under load, the abort signal can be
queued in front of a signal emitted from the httpReply.
The abort slot is deleting the httpReply and setting it to null.
So when the queued slot is processed the httpReply is null and
caused an MMU fault.
Removed qWarning from existing null checks, as these are expected
if abort is called with precise timing so that it races with the
reply finishing on the socket.
Task-number: QTBUG-26245
Change-Id: I0a7e0223fda1bc01d117fe8a993c7f6e43fd72ff
Reviewed-by: Jonas Gastal <jgastal@profusion.mobi>
Reviewed-by: Richard J. Moore <rich@kde.org>
QPair is small and is no problem to include from
qdatastream.h. However, including QDataStream from qpair.h means
including QIODevice and QObject too.
Change-Id: I344321e9f68438008ec329a165135c3a346c6058
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
As required by RFC6265, if the date can't be parsed then the
attribute should be ignored but not the whole cookie.
Task-number: QTBUG-15794
Task-number: QTBUG-16798
Change-Id: I2ed2fbbaf2d00b194995349c24b174032a599703
Reviewed-by: Richard J. Moore <rich@kde.org>
As recommended by RFC6265.
This fixes the optional-domain0042 test case.
Task-number: QTBUG-15794
Change-Id: I6dd459797afcb52fa2a78437f8481f5abc6f3105
Reviewed-by: Richard J. Moore <rich@kde.org>
Url encoding of paths is no longer used. This matches the
current release behaviour of Firefox, Chrome and MSIE browsers.
RFC6265 does not allow this type of encoding.
This fixes remaining path test cases in the IETF test suite.
Currently the path0027 test is passed by Firefox but failed by
Chrome and MSIE, so there is a potential compatibility issue.
However it is a corner case with a malformed cookie.
Task-number: QTBUG-15794
Change-Id: I9b02bb5adc32d614f512d314d06f2c60894aa2b0
Reviewed-by: Richard J. Moore <rich@kde.org>
The ';' separator takes priority even inside a quoted string.
Quotation marks have no special meaning, they are not parsed and
regenerated anymore. This means it is not possible to include
the ';' character inside a cookie value.
Other characters are returned transparently, including [",\]
Task-number: QTBUG-15794
Task-number: QTBUG-26002
Task-number: QTBUG-11641
Change-Id: I4eefef5c6ac7753d5a21c226169e264578521fe9
Reviewed-by: Richard J. Moore <rich@kde.org>
If unparsable, ignore the max-age attribute but process the rest of
the cookie normally.
If max age <= 0, set expiration time to "earliest representable time"
To keep this a safe value for conversions, time_t of 0 is used.
This fixes cases 0019 and comma0005 in the test suite.
Due to this change, cookies may be sent after they should have expired
in case the max-age was malformed. Previously they would have been
discarded immediately, which is more likely to break web services.
Task-number: QTBUG-15794
Change-Id: I7882af8eb37db156785e4e358ca639e90c94f8d0
Reviewed-by: Richard J. Moore <rich@kde.org>
RFC6265 clarifies that unknown cookie attributes should be ignored,
including the version attribute which was defined by RFC2109 but
not used correctly in practice.
This fixes case 0008 in the test suite with minimal risk.
Task-number: QTBUG-15794
Change-Id: I6f15e8e5e2e5f1ed168fc733a5c84d606a452252
Reviewed-by: Richard J. Moore <rich@kde.org>
This also allows cookie values to contain commas to increase compatibility like
most popular browsers do even though the RFC still reserves them for future uses.
Reviewed-by: Peter Hartmann <peter.hartmann@nokia.com>
Task-number: QTBUG-21456
(cherry-picked from 8ba781b01e900148fec2e9d26485369b3295487f)
Change-Id: Ib09ab2411dddf7f99de1c0c31680428b7412fc7e
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
This is needed for the QSslSocket. When we read on that socket we will
only read from the QIODevice buffer to get the unencrypted data.
So when the readNotifier has been turned off on the plainsocket there
is nothing to trigger it to be turned on again.
This will add a readData with zero size when we have read everything
from the buffer. This is so that we get a call into the socket to
check if the readNotifier should be turned on again.
Change-Id: I3b63e33de007db823e964480903186eb1b8caac2
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Removes the readBuffer from the QAbstractSocket since data is already
buffered in the QIODevice.
Change-Id: I4e50b791fd2852455e526fa2c07089d4d3f0b2a4
Reviewed-by: Prasanth Ullattil <prasanth.ullattil@nokia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The documentation is misleading, as using the new API actually
causes WSASocket to fail.
New behaviour:
On windows vista or earlier: skip the new API, use old one
On windows 7: try the new API first, if it fails try the old one
On windows 8: try the new API only
The windows 7 behaviour is because we don't know if the service
pack has been installed or not. (And IT departments may have
specifically installed/blocked the hotfix)
Task-number: QTBUG-26224
Change-Id: I6da47959919caee0cd2697f1ae1fca46aa33c1ff
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Mikhail Vorozhtsov
When doing a NTML authentication it would close the connection after
the second phase before replying when the reply was a HTTP 1.0 version
reply. Since the Proxy-Connection header is set to Keep-Alive in this
case we want to ensure we do not close in this siutation.
Task-number: QTBUG-26037
Change-Id: Icaaf2277efc0d05a946c52fb42b5191964e2e0fe
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Server responses may arrive in more than one packet, though this
is rare due to nagle algorithm.
Also fixed IPv6 addresses being discarded from server responses,
which was caught by the new autotest.
Task-number: QTBUG-18564
Change-Id: I32d9e2978037fb3e1fff27b7e618b5da6d222f28
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
Qt 4.7.3 sent no Content-Type header by default.
This was fixed independently on 4.8.0 and 4.7.4 branches, with
different defaults.
Since this is often used for web service logins, the 4.7.4
default of x-www-form-urlencoded seems more likely to work.
The warning message is left in place, since not specifying the
content type is still an application bug.
Task-number: QTBUG-23350
Change-Id: I30bf50fd216ee9894d0168e904cea1ed4251ec68
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
The httpThread was using deleteLater the finished signal of the
thread to call the deleteLater slot. If the QNetworkAccessManager is
deleted when the application is closed then then fished is emitted
but we never return to the eventloop so the deletion is never done.
This will delete the httpThread directly instead of using deleteLater
Task-number: QTBUG-25487
Change-Id: I1fdbd4eca01e8bd8b3a98936298e5c78217752b4
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
absorb module.prf into qt_installs.prf, as that's where it belongs.
add qt_install_module option and automatically set it in
qt_module_config. make qt_installs use that option.
Change-Id: I860616f3a29a456f7b88ddaffa09375400c8911e
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
The QNetworkReply::uploadProgress signal is intended for updating UI
elements such as a progress bar.
Limit the signal emissions to 10 per second to prevent overloading
the UI with updates.
As with the downloadProgress choke, this is implemented by dropping
signals that occur within 100ms of the previous emission.
The 100% signal is always emitted (bytesSent == bytesTotal)
When the upload size is initially unknown, this behaviour is still
provided by the upload device emitting a suitable readProgress
signal when EOF is reached.
Task-number: QTBUG-20449
Change-Id: I77e03c8a49109106e1c375ee00380293fd326b63
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
The QNetworkReply::downloadProgress signal is intended for updating
user interface components (e.g. a progress bar).
Limit signal emissions to 10 times per second, with an additional
signal just before the finished() signal to provide the 100% progress.
For the size of download where a progress bar is necessary, this
update frequency seems sufficient.
The implementation is done by dropping signals which would be emitted
less than 100ms after the previous signal emission.
Task-number: QTBUG-20449
Change-Id: I9c2dbe16c70f3270cbf98f3c74cf9d9a3f0ab900
Reviewed-by: David Faure <faure@kde.org>
Reviewed-by: Markus Goetz <markus@woboq.com>
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
qdatetime.h uses std::min/max and on Windows windows.h (or some subsequent
header file) may under certain circumstances define min/max as macros.
The easiest way to prevent the windows header files from doing that is to
define NOMINMAX in the place right before windows.h is included. The other
way is to define min and max to min/max themselves to prevent windows.h
from doing its evil thing.
If a user of Qt (WebKit in this case) chooses the approach of defining
min/max to themselves and then includes qdatetime.h, then a subsequent
inclusion of windows.h doesn't work because qdatetime.h undefines min/max.
We should not enforce the type of workaround needed, therefore this patch
removes the workaround from qdatetime.h and requires user code that
happens to include windows header files before qdatetime.h (seldom case)
to choose either workaround.
Change-Id: I7347eec7369491a065e894cff557004e069453d5
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
When Qt is being compiled and is using the bundled zlib, QtCore needs
to export the zlib symbols, since zlib is needed in other Qt
libraries as well. That gives a danger of a potentially disastrous
symbol clash if the user later on links with both Qt and an external
zlib (ref. e.g. QTBUG-15071).
This commit enables a zlib compilation flag that causes all zlib
symbols to be redefined with a prefix. Hence, they will not clash
with a standard zlib.
A minor drawback is that zlib.h will now have #defines for a few
semi-normal identifiers. Hence, a couple of more changes are done:
In the private qzip code, the identifer crc32 had to be renamed.
QHttpNetworkReplyPrivate needed no change, but as a defensive measure
the #include <zlib.h> is moved from the _p.h file to the .cpp file,
to avoid surprising compilation errors later in code that include
that header.
This commit does not in itself solve the issue of how to let Qt
libraries outside of qtbase use the same bundled zlib, but it is
a prerequisite for that.
Change-Id: If84105901a8c90d35009faffe660c85a6bd2fee5
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Change-Id: Ibe7cb1ca9cdcedd3f09dd4f865907e3f0fa6aef3
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
With Happy Eyeballs we will connect one IPv4 and one IPv6 channel and
pick the network layer depending on which connects first. When the
first channel is connected we can close the other one.
Before this we let the other connection finish connecting and then
closed it. This will close the other one as soon as the first one
is connected.
Change-Id: Ib2ab3f949704fd39dc0584bd31b9bcaf75ce35f7
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
The use of QWeakPointer for tracking QObject pointers is to be
deprecated.
Change-Id: If460ca7f515db77af24030152f4bd56e1a5fae7c
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
For raw deflate compressed data with no zlib or gzip header
we need to call initInflate2 with -MAX_WBITS.
The first call to inflate will asume that the data has a header,
but if that call fails with a Z_DATA_ERROR we can try once more
with changed windowBits incase the data was raw compressed data
without a header.
Task-number: QTBUG-25513
Change-Id: Ib37c286c6da6d5395581717d0e76a0dbd5df289c
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
It is useful for abort() to be a slot for connecting cancel and timeout
signals.
Change-Id: Ib0146d66fffe3f57b683dfbce0863e378b8c076d
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
When handling signals from the session, a pointer is needed.
Also the QNetworkReplyImpl needs to access the manager's session.
So, the manager should have a strong and weak reference.
The strong reference is held during connection establishment.
The weak reference is held all the time, though it will become
null when the session is destroyed in idle.
The non static member function getNetworkSession() is used to
create strong references from the weak reference where required.
Task-number: ou1cimx#1004278
Change-Id: I4b5b36b1d996b98e659d993969006c61b4440c15
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
For bearer management to work correctly, we need to know when there
are no network replies active. Previously this was implemented
using qFindChildren, but that doesn't work when the user reparents
QNetworkReply.
QtWebkit does this (actually sets parent to 0).
Also the qFindChildren implementation was racy if multiple requests
were finished in parallel. Again, likely to be triggered by webkit
loading page elements.
Task-number: QTBUG-15812
Change-Id: I181a9ba6611c7c4b6fffa2d84fe4029d89e8f596
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
QNtlmWindowsHandles was placed inside Q_OS_WIN in some places and in
other places inside Q_OS_WIN32. It seems that Q_OS_WIN is correct
define everywhere.
In addition placed "InitSecurityInterfaceW" inside L macro to convert
argument to wide characters in WinCE. WinCE GetProcAddress takes
LPCWSTR instead of LPCSTR.
This fixes the QtNetwork build for WEC7, where Q_OS_WIN32 apparently
is not defined. Should it be defined, is another story...
Change-Id: Id309d20c46b66139e2cb2e62349067848d8ebb4e
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
When the QNetworkAccessHttpBackend was merged into the
QNetworkReplyHttpImpl there was some code and comments comming from the
backend that was left in the new class. This removes some of these
leftovers from the QNetworkAccessHttpBackend.
Change-Id: Ifa118160438e2740fb9bf52907066096d8de9ae7
Reviewed-by: Markus Goetz <markus@woboq.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Check for blacklisting in case the application has blacklisted
a cert before windows has (currently unlikely as the blacklist is
hardcoded in Qt)
Don't need to check for time validity because that's already checked
by the windows API.
Change-Id: I34da5c4a8a0f8851b9b7668fc421a93c360c8588
Reviewed-by: Richard J. Moore <rich@kde.org>
Calling a static function non statically causes an unused variable
warning for the optimised away d pointer.
sscanf causes an insecure functions warning. (Even though it was used
safely in this case)
Change-Id: I07700e2155284ef3ebbe7d604ed59b2e61ee7f95
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
In case the HTTP server returns more data after the end of the
compressed data stream, inflate will return Z_STREAM_END, which
is a normal informative error code.
This was handled in 4.8, but lost in 5.0.
Also catch all ZLIB negative error codes rather than only three.
Task-number: QTBUG-25823
Change-Id: Ibdbbd3dd6fa81a0880c477cb080ad35f2d7116f0
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
If the QNetworkSession change while we are handling a QNetworkRequest
we should try to resume this request if possible. In that case we set
the offset for where to continue the request. Abort the operation
in the http thread. And then post the request again. The offset will
then be set using the range header in postRequest() so that the
operation is resumed.
Task-number: QTBUG-25762
Change-Id: Ib7946905bcc633f8cc3dd7a7ad1ae74579e6bf56
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
In tests when IPv6 is not present QSKIP IPv6 tests.
Task-number: QTBUG-23660
Change-Id: I02abc7322d765a93cbf661e53c76257f03dca73e
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
A couple of people reviewing the toText() method (which is new in 5.0)
have said that since the string returned is human readable it should
be a QString not a QByteArray. This change follows their advice.
Change-Id: Ibade9a24870805f7fbe2d299abeb9c6e964f0cf4
Reviewed-by: Girish Ramakrishnan <girish.1.ramakrishnan@nokia.com>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
This is for behaviour consistency with Qt on unix, as well as the
socket close issues described in the task.
Task-number: QTBUG-4465
Change-Id: Ida95650d8a9bd7b5bc3d3926d22e20a6d7eeb30b
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Incase there is no bytesAvailable on the socket there is no need to
try to read.
Change-Id: Ied9e4b8d86854b356a733fab4948739db206b6ad
Reviewed-by: Prasanth Ullattil <prasanth.ullattil@nokia.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
When we get a remoteHostClosed we should try to read everything from
the socket before we close the channel.
Change-Id: Iaa87d79ea16d69735f6ba3e8b3b4a0f86fbd5f73
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Reviewed-by: Prasanth Ullattil <prasanth.ullattil@nokia.com>
Reviewed-by: Richard J. Moore <rich@kde.org>
This prevents http POST/PUT from hanging if the QIODevice being uploaded
returns -1 from read.
Task-number: QTBUG-24738
Change-Id: I76500cc4f0101cc8e5da5f1dc105508b3f519a3c
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
MinGW installations on case-sensitive filesystems expect
lowercase names of include-libraries and (usually) include
files.
When crosscompiling on Debian 6 (targeting MS Windows) linking
fails because mingw is looking for non-existent include-libraries.
Using lowercase names solves this.
Change-Id: Id3454f4ed8ba42b6ea93d65d9c0ce567db6712df
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
When the scope ID is not set, Mac and Windows will not transmit
packets to link local addresses. This patch implements setting
the scope in the native socket engines and adds a test case.
(it was partially implemented already, though UDP specific code
paths were missed in the unix engine)
Task-number: QTBUG-25634
Change-Id: I23300bdc9856e38458078e913daaa59cd05a74b5
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
The scope ID is a required part of the address for link local
addresses (to solve the problem of the same link local address
being created by two machines on different networks).
It is required to send packets to a link local address on Mac and
Windows, although Linux multicasts if scope is missing.
Task-number: QTBUG-25634
Change-Id: Ie2bb09df8d261eefcb81716bafeb1475f0bed5fe
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
The CSS for the footer is not completely correct, but at this
time it is better to have something than nothing.
Change-Id: I7371e1e458a2abafcdb0fca5564ad73e209d64c3
Reviewed-by: Jerome Pasion <jerome.pasion@nokia.com>
Make sure that we always have a reply set when we try to read.
Change-Id: Ib3ce9063bffc9810fc2a4fb52ee1a65b7f8cdfa0
Reviewed-by: Prasanth Ullattil <prasanth.ullattil@nokia.com>
Added the setReadBufferSize functionallity again by limiting the
amount that the delegate read from the channel. Each time that data is
fetched from the reply buffer, we communicate back to the thread so
that more data can be fetched.
Task-number: QTBUG-25327
Change-Id: I2f9950196e64acd09bc8da50c1116f2c9deacad4
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Http proxy cannot support server sockets or udp sockets, so don't
offer it as the system proxy (in the generic environment variables
implementation)
Change-Id: I234ab1024952b2630e668ddc35c665efade8e2b9
Reviewed-by: Adrien Bustany <adrien@bustany.org>
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
Ensure comma between elements (757 missing), single space and curly-
braces around title elements, etc.
Change-Id: Id16c3fda7fc47a12a0682f8720214f4990609a97
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Reviewed-by: Casper van Donderen <casper.vandonderen@nokia.com>
- Adds support for doing NTLMv2 integrated authentication via logged in
users domain credentials using the Windows SSPI api.
Task-number: QTBUG-17332
Change-Id: I4e840f7fb54ce1ace8a6151868f59f413d232295
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
The files in this change were still in qtbase/doc/src or required for
it. qtbase/doc/src should now only contain example documentation and
images for the example documentation.
Change-Id: Ia7ca8e7fd2b316e77c706a08df71303bc8294213
Reviewed-by: Marius Storm-Olsen <marius.storm-olsen@nokia.com>
- Fix multi-threading bug where NTLM/Digest auth would fail when
concurrent requests were on the wire. The fix for this is too not
internally share QAuthenticationPrivate pointers, since the challange
values would get overridden in different threads. This was failing
because the internal QAuthenticationPrivate members would have been
set with the status/values of the current request which would mess
up the state of the new request. As currently implemented, the helper
functions inside QAuthenticationPrivate can't call detach to implement
proper copy on write symantics.
- Fix issue where if user was set via constructor, the NTLM domain
parsing would not occur. Parsing of DOMAIN\user is now redone if
proxy type is determined to be NTLM.
Task-number: QTBUG-15472
Task-number: QTBUG-17322
Task-number: QTBUG-18794
Task-number: QTBUG-13063
Task-number: QTBUG-16585
Change-Id: I8a898c51fb04fab6fb08d96d88dd73be0c87af5d
Reviewed-by: Aron Rosenberg <aronrosenberg@gmail.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
This change moves the snippets and images to the modularized
directories.
Change-Id: Idacff866735d00b048d65318bc4c3ee1dfa16310
Reviewed-by: Marius Storm-Olsen <marius.storm-olsen@nokia.com>
This change moves the snippets and imagesto the modularized directories.
Change-Id: If14912692a7f72d7de345eaf88d9ec9752310fca
Reviewed-by: Marius Storm-Olsen <marius.storm-olsen@nokia.com>
Only call qdoc for projects which sets the QMAKE_DOCS variable to
point to a qdocconf file.
Exclude examples/ and tests/ from the qdoc run, by adding
no_docs_target
to CONFIG for those projects.
Change-Id: Ic856c8f19db59309302d0602b3e99735609e525a
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
Reviewed-by: Casper van Donderen <casper.vandonderen@nokia.com>
This is to avoid a deadlock that happens when a user thread is
accessing the QNetworkConfigurationManager at the same time the
plugin emits a signal.
i.e.
plugin is holding engine lock
user thread is holding manager lock and blocked trying to acquire
the engine lock
In the manager slot, it tries to acquire the manager lock.
By using queued connection, there are no locks held at the time the
manager slot is called.
Change-Id: I95f28028b5e77f77b2b9b7e31cbd1b78a8fe3097
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
QNetworkSession::open can synchronously emit an error, therefore
we need to queue this.
Otherwise QNetworkReply::finished is emitted before the user has
had a chance to connect the signals.
Task-number: QTBUG-18824
Change-Id: I703d5e31d2934afafabdf0a77ea3aaf5336e8dec
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
QNetworkProxyFactory used to check only for the http_proxy environment
variable in systemProxyForQuery. This patch makes it look as well in
https_proxy, ftp_proxy and all_proxy. http_proxy is still used as a
fallback value.
Change-Id: I7934af70d191cd17dbce3b3789260ae1a8332986
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
QNetworkAccessManager was using the AutoCloseConnection property
of QNetworkSession to close it when idle. However this property is
only implemented for polling engines and not event driven engines.
Instead, release the network session reference.
If another request comes in, it will be resurrected from the shared
session weak reference.
If not, then after 2 minutes when the connection caches are flushed
the ref count will reach zero and cause the QNetworkSession to be
destroyed (which closes it)
Task-number: QTBUG-15812
Change-Id: I2963bdf13fb02e3ac269489ea463669142f3c5f3
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
Windows 7 accumulates networks over time with the status
"not present". This is so it can remember whether a given network
was a public, home or work network next time you use it.
By default, these networks are not returned when enumerating
network interfaces, but we specified a flag to request all networks.
These networks are generally not useful so use the default behaviour
of not reporting them.
Task-number: QTBUG-18824
Change-Id: I4edc4cea83a0e97a88ac649b7d8af95a8a600b89
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This operation should be a no-op anyway, since at this point in time,
the fromAscii and toAscii functions simply call their fromLatin1 and
toLatin1 counterparts.
Task-number: QTBUG-21872
Change-Id: I94cc301ea75cc689bcb6e2d417120cf14e36808d
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Reviewed-by: Richard J. Moore <rich@kde.org>
This functionality was in Qt4, and apparently lost in the
QNetworkAccessHttpBackend -> QNetworkReplyHttpImpl change
Change-Id: Ia354cfd2c10808a6543173ec017d911f1696e7fb
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
Although we created an enum for pause modes to make 5.x binary
compatible with 5.0, the enum value is not well named.
In 5.1, we propose to add PauseOnProxyAuthentication to the enum.
PauseOnNotify is not clear what it means, while PauseOnSslErrors is.
Any new notification in a minor release would need a new enum value
otherwise applications would get pauses they did not expect.
Task-number: QTBUG-19032
Change-Id: I4dbb7467663b37ca7f0551d24a31bc013968bedc
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
If a website presents the complete certificate chain in the handshake
i.e. site -> intermediate CA -> root CA then openssl gives
a different error (SelfSignedCertificateInChain)
Because of this windows feature, that either means the site is
signed by an untrusted CA, or the CA trust status is unknown because
we don't have the root cert in the cert store.
In any case, calling the windows verification function results
in a trusted chain & the root being added to the cert store.
Task-number: QTBUG-24827
Change-Id: I2663ea2f86cd0b4dfde105d858ec1b39a340c1f6
Reviewed-by: Richard J. Moore <rich@kde.org>
The \since 5.0 directive was missing from many places.
Task-number: QTBUG-24001
Change-Id: I191ba8891ae66d78f923164bcab2fccb16eabef9
Reviewed-by: Casper van Donderen <casper.vandonderen@nokia.com>
With the new functions linked
Task-number: QTBUG-24001
Change-Id: I9fd2de746a6342a1f4f182189e7f2529f092c003
Reviewed-by: Casper van Donderen <casper.vandonderen@nokia.com>
If we're not going to verify the peer, or we know in advance that
windows won't have a CA root then don't ask it to verify the
certificate chain.
The test case started failing in CI when the windows cert fetcher
was integrated due to timing change. I've relaxed the timing
requirement of the test to avoid it being unstable.
Task-number: QTBUG-24827
Change-Id: I694f193f7d96962667f00aa01b9483b326e3e054
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
Tested manually using the localfortuneserver example and multiple
login sessions. The linux autotest isn't suitable for windows due
to pipe permissions not appearing in the filesystem.
Task-number: QTBUG-25147
Change-Id: I5ea4db81d1870dc45bd483fa8d0b06afede3b722
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Added \since 5.0 and corrected the enum documentation to match the
code.
Task-number: QTBUG-24001
Change-Id: I8e284b6eaaf68c46bbd0e07f4378f0c442f2edd4
Reviewed-by: Casper van Donderen <casper.vandonderen@nokia.com>
Most FTP servers do not support the ABOR command (they don't process
control channel traffic until the data channel is finished). Or they
require the telnet IP & sync procedure on the control channel (this
requires sending TCP urgent data, which we do not support).
Following behaviour of most browsers and GUI FTP clients, abort
downloads by resetting the data channel.
Abort of uploads needs no change, because the client closes the data
channel rather than the server.
Task-number: QTBUG-25494
Change-Id: I69e7b7c04d709d26def9a4a7081074e1cbf69701
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
Previously, we just disconnected signals from QFtp and allowed it
to continue downloading in the background.
Task-number: QTBUG-25494
Change-Id: I891c2fff88ef1ee554d1ccf821a3f7998eeb8406
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
Android's resolv.h is completely crippled, so we can't use the libc resolver.
Change-Id: I07e310e3561c719f4fe0bf08e58b5194f5a76e4a
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
qVariantValue and qVariantCanConvert are Compatibility members, while in
Qt4.8 they are marked as Qt 3 Support Members.
qVariantFromValue and qVariantSetValue are Obsolete members.
Change-Id: Ie8505cad1e0950e40c6f6710fde9f6fb2ac670fd
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
The QNetworkProxyFactory class considered the http_proxy environment
variable, but not the no_proxy one. This commit adds no_proxy handling,
loosely modeled after the way curl does it.
Change-Id: Ibb9e5ffcb30fed5c95dc9fc3bc4177e20d025a50
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Q_DECLARE_METATYPE must be outside of the qt namespace.
System headers must be included outside of the qt namespace.
Change-Id: I2f48b1df87e5edae2baee6ce813af08d3e011dc0
Reviewed-by: Sergio Ahumada <sergio.ahumada@nokia.com>
Reviewed-by: Toby Tomkins <toby.tomkins@nokia.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
The const char *-based API is deprecated and will be removed in Qt5.
Change-Id: I36f6dc761e3b5a087e38db29b761c3e9237958b4
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Put MacOS-specific code in #ifdef blocks so that it is not compiled for
iOS.
Change-Id: I303b7dbb63eb298e92351951731868df6a38bffa
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
... to conform to RFC 2046 (section 5.1.1). Apparently IIS had problems
without the new line.
Task-number: QTBUG-25429
Change-Id: Ia619bbdcebd407b2716bc467323634e4c8d77bcd
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
While linux returns the local address of dual mode sockets as "::"
i.e. the IPv6 address, OSX returns "::FFFF:0.0.0.0" i.e. the IPv4 address.
In the unix socket engine, treat both of these as candidates for
being a dual mode socket. This should hopefully cover other unixes
which we haven't tested.
Task-Number: QTBUG-25445
Change-Id: If1251018d54da98618223ba538c00ba54ac4cb4e
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The previous solution didn't work on systems where the libraries
aren't in the same location.
Now we search for both libcrypto and libssl and load them if their
versions match, even if they are in different directories.
Task-number: QTBUG-25398
Change-Id: I37164638890586947d07670d8a59fc53a84f9c42
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Reviewed-by: Richard J. Moore <rich@kde.org>
Calling peek() for qsslsocket caused socket data to be copied into
qiodevices buffer and therefore make it unaccessible in qsslsocket.
Cherry picked form 4.8-branch & modified to Qt5 API changes
(int -> qintptr)
Original commits:
commit 621f18955082fc73471e75d1f8c35c2dcd4befeb
Author: Shane Kearns <ext-shane.2.kearns@nokia.com>
commit 68b1d5c17aa38d5921bdade2b0e0cb67c6c90513
Author: Kalle Viironen <kalle.viironen@digia.com>
Task-number: QTBUG-18498
Change-Id: I6be4b19baec2f3197537f5e7b61432040ec84ad2
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
If the "bypass proxy for local addresses" option is enabled in
the windows proxy configuration, then do not use the proxy for
any IP address in the subnet of any network interface.
As the systemProxyForQuery api is now offering HTTP proxy tunnels
for TCP sockets, this change avoids local ad-hoc network
connections being routed through the proxy.
In the case where the local address was on a different interface to
the proxy server, it may have been unreachable through the proxy.
For example IP over USB or Bluetooth.
Change-Id: I0842732832a7795112be029d923ed168edc008d6
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Since we are calling q_SSL_load_error_strings to load error strings
we should call ERR_free_strings to free the memory again.
Task-number: QTBUG-15732
Change-Id: Ie41291bb0e1434f82025378edfca51930712a8aa
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Using the policy change signal from QNetworkSession.
If the new policy disallows background requests and this is a
background request, then generate an error.
This results in a TCP RST on the socket, and a
BackgroundRequestNotAllowedError on the QNetworkReply.
If the reply is already finished, no action is taken.
Change-Id: I4ff5c681a8b7b852727bb95f03664d666f4efe07
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
Replace "contains(QT_CONFIG, coreservices)" with "!ios" in config files.
Replace "QT_NO_CORESERVICES" with "Q_OS_IOS" in source files.
Change-Id: Id3b02316b245a24ce550e0b47596d18a4a409e4f
Reviewed-by: Morten Johan Sørvig <morten.sorvig@nokia.com>
The used functionality is partially not available under WinCE.
qnetworkfunctions_wince.h encapsulates the needed extra symbols.
It only needs to compile as the functions are loaded dynamically.
Change-Id: Ieb9010d4f6c7f94d39918b869ac503976e094e49
Reviewed-by: Andreas Holzammer <andreas.holzammer@kdab.com>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Previously the cache was not enabled if the request had AlwaysNetwork
set. This removes the check for the CacheLoadControlAttribute when
checking if the cache should be enabled.
Task-number: QTBUG-15805
Change-Id: I3d0722fbc1e6c77c5b947a2af0a09e2de8f4f9df
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Put MacOS-specific code into #ifdef blocks to enable compilation on iOS.
Change-Id: I0bb3846f457d1b3a56d99fe182b1718bc8429117
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
ce only has the utf16 variant desktop has all variants
of the DnsQuery call. Unified it to the utf16 variant.
Change-Id: Ic126c3f53da71c85a41a5c24c85970335700bcf0
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Andreas Holzammer <andreas.holzammer@kdab.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
The cache expiration calculation was previously removed to avoid
calculating it every time an object is fetched from the cache.
This is because the expiration date is calculated from max-age or
expires headers when the object is added to the cache.
However this broke the case where the server only provides a
last-modified header. In this case, the expiration date needs
to be checked at fetch time in order to add the warning if the
object is >24h old.
Change-Id: Iad381da0f91c27ac603abbaa94f268700ad2297f
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
When errors are detected synchronously in _q_startOperation,
they were not reported. This is because unlike the generic
QNetworkReplyImpl the function is called directly rather than
using a queued connection.
In order to report errors, use the _q_error and _q_finished slots
so that signals are emitted after returning to the event loop
i.e. after the application had a chance to connect the QNetworkReply
Change-Id: I8a7bbe79a934f4634fb4e0572ebb5479dfc5f489
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
Change-Id: I7d8ea41299408377042a9f0d0a672e1a6fb57e7d
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
If background requests are not allowed, don't even attempt to start
the network session, just fail the request immediately.
After this change, a QNAM with mixed requests queued should behave
as follows when background requests are disabled:
- background requests at head of the queue fail
- first foreground request starts the session and succeeds
- remaining background requests fail
- remaining foreground requests succeed
If policy is changed on the fly, then running background requests
are not aborted. However queued background requests won't be started.
Change-Id: Ic9aba1eb59ca41b166a08d2ed09418e1b6af6b60
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
This allows the system to publish usage restrictions to applications
related to the network in use.
Currently there is only one restriction defined:
NoBackgroundTrafficPolicy, which means that non user initiated traffic
should be avoided (e.g. background downloads).
For example this policy could be applied to save battery or data transfer
charges.
Change-Id: I49e26c0f3650d2b92f4ec51981aae9435b717b49
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Switched the error code.
Republish the error string from the bearer plugin if possible.
Change-Id: I9e4ac7a9914fbf2e87fe8fd3a5175deda6d933d2
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
This error will be used when background network requests are not allowed
according to the current policy of the bearer plugin.
For example, to save power when battery is low on a portable device.
Change-Id: I866e115f8fdd046134da99ea895b7c1df0375f26
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
This is to replace the UnknownNetworkError which occurs when the
internal QNetworkSession fails to start (e.g. no usable WLAN available)
Change-Id: I2b14577c22e0acf8ff07be7e932f0dfe9ac89c33
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
Set the QNetworkSession attribute to match the QNetworkRequest that
triggered it.
Currently if there are a mix of normal and background requests queued
and background connections are not allowed by policy:
- background requests at the head of the queue should fail
- first foreground request should successfully start a connection
- remaining requests succeed regardless of background attribute
Change-Id: If0e3ec0b8a5096e3d7cd6df85884c6f53172d233
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
This is so that the ConnectInBackground flag can be set on the
QNetworkSession internal to QNAM according to pending requests.
Change-Id: If0cc62f5117ed8febbbda7b7f6de62b11b274258
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
Windows ships with a minimal set of CA roots.
When using windows API to verify a certificate, it will fetch the
root certificate from windows update (assuming it is part of the
Microsoft trust program).
As we are using openssl, this does not happen transparently.
If SSL errors occur which indicate a broken chain then attempt
to fix it using the windows API before emitting sslErrors.
If the system CA certs are not in use (a CA bundle has been set
on the socket or as the global configuration), then this is skipped.
This is so an application can continue to use its own cert bundle
rather than trusting the system certs.
Key usage is specified, so that windows will return not trusted
status if the root is not suitable for SSL (server auth or
client auth OID).
Testability:
- to test, must delete the CA cert(s) from the "third party
root certification authorities" section of the cert store
using mmc.exe.
- If the workaround of installing the windows XP cert bundle was
performed, then you also need to delete certs from the "trusted
root certification authorities" section.
This is dangerous, be careful not to delete the required
certificates which are documented on MS website
- Naturally, modifying these areas of the cert store requires
elevated privilege.
Task-number: QTBUG-24827
Change-Id: I5cfe71c8a10595731f6bbbbabaaefa3313496654
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Removed some MacOS source code files from iOS build. Use unix standard
paths for now (iOS-specific implementation will come later).
Change-Id: I8b2731b431b3a379a1ec4ec07d227e886209e3e9
Reviewed-by: Morten Johan Sørvig <morten.sorvig@nokia.com>
Just like qMalloc/qRealloc/qFree, there is absolutely no reason to wrap these
functions just to avoid an include, except to pay for it with worse runtime
performance.
On OS X, on byte sizes from 50 up to 1000, calling memset directly is 28-15%
faster(!) than adding an additional call to qMemSet. The advantage on sizes
above that is unmeasurable.
For qMemCopy, the benefits are a little more modest: 16-7%.
Change-Id: I98aa92bb765aea0448e3f20af42a039b369af0b3
Reviewed-by: Giuseppe D'Angelo <dangelog@gmail.com>
Reviewed-by: John Brooks <john.brooks@dereferenced.net>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Add support for automatically searching the Mac System Preferences
for proxy server username/password. If a user has put credentials
in the SystemPreferences->Network->Interface->Proxies area, we
will now look in the KeyChain for those files. This will
automatically pop up a Permissions dialog from the OS if valid
credentials were found which match the server we are trying to
access.
Task-Number: QTBUG-22033
Change-Id: Ic7952afab4d16a65a87bb2f97a928c1c91167fe7
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Adds support for fetching and parsing Proxy Auto Config files if one
is specified in the Mac System Preferences
Task-Number: QTBUG-2069
Task-Number: QTIFW-28
Change-Id: I91feb999222187e7467f2c41383904cf0cff8633
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Avoid the conversion to a temporary QString -- just hash the address
as a byte array.
Change-Id: Ic35cdbbc3ee66c32a28d911bd27de0092395979f
Done-with: Shane Kearns
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
The installed path of libssl may include an element describing the
architecture, e.g. x86_64-linux-gnu or i386-linux-gnu.
In most cases, the libraries already loaded (static dependencies of
Qt, such as libc) will include the path where libssl is installed.
Use dl_iterate_phdr to find the paths. This is a linux specific
function, but it does provide "/lib/<arch>" and "/usr/lib/<arch>"
at the point ssl symbols are being resolved when running the
qsslsocket autotest (which has less dependencies than a typical
Qt app).
Task-number: QTBUG-24694
Change-Id: I9af8081f41bb85c2fcff450a2acda5672a7f7518
Reviewed-by: Harald Fernengel <harald.fernengel@nokia.com>
The current scheme is to use IE's default config.
If that fails get the winhttp config.
That's ok. The problem is that if you run a program as a service
getting the IE config will set the fAutoDetect flag.
But later the call to WinHttpGetProxyForUrl mightfail with the
error code ERROR_WINHTTP_AUTODETECTION_FAILED.
this patch just makes sure that we have a fallback winhttp solution
in case the IE proxy is not set.
The new code detcted if the current process is a service, in which case it
will try to default to the system-wide proxy.
Change-Id: I57e9082a46a8422c54f8f069715752c271a3a001
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
If there are still bytesAvailable on the socket and we have
nothing pipelined we should close the channel. This was not
done before as the chunked decoder did not read the last
CRLF. This has now been fixed so this close should be enabled
again.
Task-number: QTBUG-24875
Change-Id: I115b60f8d78bf1b83d3fd5887ab3eb9940b6bf00
Reviewed-by: Jonas Gastal <jgastal@profusion.mobi>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
If the current host name is an ip address we can create the QHostInfo
directly instead of performing a reverse lookup.
Task-number: QTBUG-18881
Change-Id: If239481e455f4f7fb74d978f685dc5d8e9f22c0a
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Jonas Gastal <jgastal@profusion.mobi>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Way back in the mists of time, someone added SO_REUSEPORT to socket binding,
which was great, because otherwise it meant that multiple UDP sockets couldn't
share the same port on OS X (as platforms with SO_REUSEPORT apparently don't
support rebinding with SO_REUSEADDR).
However: SO_REUSEPORT also means that *any* bind on a port will succeed, which
is most definitely not wanted in the case of TCP sockets, so check the socket
type before performing the actual bind.
Also test that multiple listens don't take effect.
Change-Id: I2f8d450bcfb8a7f3abd8918a4e789a850281dd13
Done-with: Thiago Macieira
Done-with: Shane Kearns
Task-number: QTBUG-6305
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
On OS X and Windows, this was not working, because the socket was being bound
in v6 mode (due to ::Any being for dual mode), but the address passed was a v4
address, meaning it took the wrong codepath. Linux, strangely, apparently works
anyway.
This is fixable in OS X (by using the v6 join path when bound in v6/dual mode),
but the same fix doesn't work on Windows, failing with WSAEADDRNOTAVAIL.
Don't allow this behaviour, and provide a sane error message telling the user
what to do instead.
Done-with: Shane Kearns
Task-number: QTBUG-25047
Change-Id: Iaf5bbee82e13ac92e11b60c558f5af9ce26f474b
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
This was never set on Windows Xp or later as we only checked for this
in the old interfaceListingWin2K() function.
Task-number: QTBUG-10170
Change-Id: Ib0808a945d9d1b31019967a69e668ec216702799
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
When FTP login fails we fail to remove the entry from the cache.
This is because the cache key is created from the url with the
userInfo. So this needs to be set again to match the key used
when inserted.
Task-number: QTBUG-11824
Change-Id: Ib3fd2d737581653ae59c56d0810d42e2d8dc2176
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
The use of any broken-down components of the query now needs
QUrlQuery.
The QUrl constructor and toString() are now rehabilitated and the
preferred forms. Use toEncoded() and fromEncoded() now only when we
need to store data in a QByteArray or the data comes from a QByteArray
anyway. Change to toString() or the constructor if the data was in a
QString.
Change-Id: I9d761a628bef9c70185a48e927a61779a1642342
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
The new code now generates lowercase hex instead of uppercase, so
adapt the unit tests to pass.
Also, "123.0.0" is now considered valid (compatibility with inet_aton).
Change-Id: I07b5125abf60106dc5e706033d60836fb690a41f
Reviewed-by: João Abecasis <joao.abecasis@nokia.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
When chunked encoding is used we should try to read the last CRLF after
the last zero-lenght chunk, with chunk size coded as 0.
Task-number: QTBUG-19480
Task-number: QTBUG-20924
Change-Id: Ida40593fec8788bff713a31cfe6a7c2d86354a91
Reviewed-by: Jonas Gastal <jgastal@profusion.mobi>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
A combination of other fixes resulted in authentication credentials
with a blank username & password being incorrectly added to the
cache only for NTLM.
This caused authentication to be attempted with blank credentials
first (which would fail), before the authenticationRequired
signal was emitted.
It caused a performance issue because of the extra 2 requests to the
server (and possibly delays inserted by the server following a failed
login)
Change-Id: Ic588a20cfe7c24d5e60cd384caff0673a587e484
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
The QLocalSocket is using a QLocalUnixSocket which is a QTcpSocket
that has buffering enabled. Because QLocalSocket is a QIODevice it
also has its own read buffer. By opening the QIODevice Unbuffered
we only use the buffer in the QLocalUnixSocket.
Change-Id: I4b9091522f73b0831b20efb4a2a00c2603fc748b
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
When writeDatagram is called without first binding the UDP socket,
then bind it as QHostAddress::Any.
This allows the same socket to be used to sent to both IPv4 and
IPv6 destination addresses.
Allowing the OS to autobind the socket inside sendTo() may
result in a single protocol socket.
Task-number: QTBUG-5275
Change-Id: I2b76507e8a8a38369c6eafb61ce4191d1d6cc930
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
Otherwise, it crashes if you simply set the same proxy factory instance
twice.
Change-Id: I4b1f1d0eed7630c9a145c0a330c6be1b3fc381e5
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
When binding with the DefaultForPlatform bindmode, then don't set
options on the socket related to sharability, leave them at the
platform default.
This restores compatiblity with Qt 4
Change-Id: I612a3acb976f08446b4eef03ccdcdf84b3477c3c
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
Otherwise, you don't have any socket engine handler created when
directly setting a proxy to a socket.
Change-Id: I35cd7f52331672a6d3e7bcdf817c0fe1fd1e7de3
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
LSB doesn't define some functionality that we expect on Linux, so
fall back to the generic code.
Change-Id: I19e09908b9e9002140535f159ae0745d7ca80b70
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
If there are no bytesAvailable and no reply then the channel
can be closed without any need for this warning.
Task-number: QTBUG-9315
Change-Id: Ifbb1f4732d3173a7807ac5bd0dc78fb5144faaf3
Reviewed-by: Markus Goetz <markus@woboq.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
This is a semi-automatic search, so I'm
reasonably sure that all the exported ones
have been caught.
Change-Id: Ia00eb9194a5f64002bd7e7b894abf6333d1b825e
Reviewed-by: Jonas Gastal <jgastal@profusion.mobi>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Each channel will create a socket that will allocate memory for the
read and write buffers. This change will instead initialize
channels only when they are needed.
Change-Id: I112b4c7b944a7dd345414f06260c92803394eaed
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
This reverts commit ff25691d00.
The change broke qtdeclarative. Several autotests crash because
QHttpNetworkConnection::transparentProxy() calls proxy() on a
null socket.
Task-number: QTBUG-24717
Change-Id: I57e3ccf5d20683f59cf7450083d1fcb3fa1c40fe
Reviewed-by: Kent Hansen <kent.hansen@nokia.com>
If SSL_MODE_RELEASE_BUFFERS is available we should tell OpenSSL
to release memory early.
http://www.openssl.org/docs/ssl/SSL_CTX_set_mode.html
Task-number: QTBUG-14985
Change-Id: Ib6656ebb3c4d67ca868b317ee83ddbf0983953f9
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Reviewed-by: Markus Goetz <markus@woboq.com>
The comodogate 72:03:21:05:c5:0c:08:57:3d:8e:a5:30:4e:fe:e8:b0
certificate is a test certificate and the MD5 Collisions was created
as a proof of concept deliberately made to be expired at the time
of it's creation.
Task-number: QTBUG-24654
Change-Id: Ic8eb417363569fe50bf19cd229658f5e371862f7
Reviewed-by: Richard J. Moore <rich@kde.org>
There was a problem with slow https connections that if the download
was ongoing when quit() was called then it tended to cause a crash
because it tries to shutdown the SSL connection.
Task-number: QTBUG-24594
Change-Id: I6b161bc7a9bb99e41849537462de2d7c92661902
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
If both "automatically detect settings" and "use automatic configuration
script" are selected in internet options, then respect that.
Because of performance issues, these are tried separately, and
disabled if they fail.
Task-number: QTBUG-13957
Task-number: QTBUG-10238
Change-Id: Ibc824d3039afeaf12c5ea82ed95287e5cf4e1776
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
TcpServer requests always returned no proxy, even if socks was available
Tag handling was broken for empty tag (if system proxies were tagged)
Tag handling was broken for unknown tags - now handled the same as if
no tag was given at all.
When there are different proxies for http and https, windows returns
the http proxy first. However we should prefer to use the https proxy
for general sockets, as it's more likely to support the CONNECT method.
Change-Id: I55dcadf2e142367e857f94e55fdbb0c4ddb513a3
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
If both manual proxy settings are configured and the autodetect
setting is enabled, fall back on the manual configuration when
auto detection fails.
Task-number: QTBUG-10428
Change-Id: If008c7c967eec6256ce3c614fff0ec258190d451
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
The PAC file may result in direct connection, in which case
WinHttp returns null strings for the proxies and the connection
type marked as direct.
In this case, return the default list (no proxy)
Change-Id: I601033f56a841bb92ea80a28174bb993b024ad79
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
When the windows internet options are used to specify a
different proxy for each protocol, assume the proxy server type
matches that protocol too.
e.g. "socks=qt-test-server:1080" is both tagged for socks, and
assumed to be a socks server.
"ftp=qt-test-server:2121" is assumed to be an ftp proxy
"ftp=http://qt-test-server:3128" is overridden to be a http proxy
used for ftp.
Task-number: QTBUG-10502
Change-Id: I70615c89d6ede53f0e7d62e6d0754b90d042aa2e
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The test was testing the wrong thing, and passing even though
QNetworkRequest::AuthenticationReuseAttribute was not being
respected, until recently when I fixed username/password in URLs
Now the cache is properly bypassed when this attribute is set to
manual, and the autotest is updated to check this.
Change-Id: I87943515562d0b16b03504f0758ba265758d1c22
Reviewed-by: Martin Petersson <Martin.Petersson@nokia.com>
Loser of the race would try to delete an uninitialised pointer
Task-number: QTBUG-15765
Change-Id: Ie184ee2306e102aa8fbad752ef09b95c3ede00c2
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
QSslCertificate can be copied around into multiple threads,
without detaching. For example, the https worker threads inside
QNetworkAccessManager.
There are const methods, which lazily initialise members of
the private class without detaching (i.e. caching results of
expensive function calls)
These functions now lock the d pointer using QMutexPool to
avoid concurrency related crashes.
autotest crashes 20% of the time in release builds without
the fix, passes 100 times in a row with the fix.
Task-number: QTBUG-20452
Change-Id: I64a01af8159216f2dd6215a08669890f6c029ca8
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Richard J. Moore <rich@kde.org>
Each channel will create a socket that will allocate memory for the
read and write buffers. QNam generaly inits 6 sockets for each
connection. That means that by default 12 such buffers are created.
This will instead initialize channels when they are needed.
Change-Id: Ie3f2cf789e084fd3d17d3b2a9bb3d3a4370b3da4
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
QDoc now has support for Doxygen style commands for italics, bold
and list items. This change applies that change in QDoc to the
actual documentation.
Task-number: QTBUG-24578
Change-Id: I519bf9c29b14092e3ab6067612f42bf749eeedf5
Reviewed-by: Shane Kearns <shane.kearns@accenture.com>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
When there is no QObject subclass with Q_OBJECT macro,
we must at least set an object name
Change-Id: Ib429a9b246d9d6b4b4cfb11593e4f358850677f2
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
The OS provides the error string in this case.
This gives more information to the developer seeing a generic
error.
Change-Id: Ia03642982f3513ee5a8a9fa98d918e948f8d97a5
Reviewed-by: Miikka Heikkinen <miikka.heikkinen@digia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Richard J. Moore <rich@kde.org>
When loading dnsapi.dll as a plugin, we experience crashes because
the calling convention is not specified.
The default is _cdecl, but __stdcall (via the WINAPI macro) should
be used for windows APIs.
Mismatched calling convention results in corruption of local variables,
probably because the stack pointer is incorrect and SP offsets are
used in optimised builds rather than frame pointer offsets.
Since the library has been available since Windows 2000, I don't
think that we need to load it dynamically.
(Unlike the unix version where it isn't part of the LSB)
Also checked that the current release of mingw works.
Task-number: QTBUG-24227
Change-Id: I37c0a6aa0c133799c2a6dd9391ca1435ba2539ea
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Windows unhelpfully writes to only one byte of the output buffer
when getsockopt is called for a boolean option. Therefore we have
to zero initialise the int rather than initialising to -1 as was
done before.
This in general only works for little endian architecture, because
the word would look like 0x01000000 on big endian. So I have added
some compile time asserts in the assumption that windows is always
little endian. This is ok for comparisons with 0/false, but not
comparisons with true or nonzero values.
In the case of IPV6_V6ONLY, it is documented as DWORD (unsigned int)
but on some windows versions it is returned as a boolean triggering
the warning. I removed the warning, as the conversion to int works on
both LE and BE since it is only compared with zero.
Task-number: QTBUG-23488
Change-Id: I3c586d1ada76465fc045a82661f289920c657a4c
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@nokia.com>
Reviewed-by: Andreas Holzammer <andreas.holzammer@kdab.com>
This would break a namespaced Qt when QT_NO_DEBUG_STREAM
was in effect. Unlikely to hit, but nevertheless fixworthy.
Change-Id: Ie2a4cf4334a6a610c84233ab1ca89b928386c91a
Reviewed-by: Richard J. Moore <rich@kde.org>
This abstraction imposed serious performance penalties and is being
dropped from the public API.
In particular, by allowing file names to be arbitrarily hijacked by
different file engines, and requiring engines to be instantiated in
order to decide, it imposed unnecessary overhead on all file operations.
Another flaw in the design with direct impact on performance is how
engines have no way to provide (or retain) additional information
obtained when querying the filesystem. In many places this has meant
repeated operations on the file system, where useful information is
immediately discarded to be queried again subsequently.
For Qt 4.8 a major refactoring of the code base took place to allow
bypassing the file-engine abstraction in select places, with
considerable performance gains observed. In Qt 5 it is expected we'll be
able to take this further, reaping even more benefits, but the
abstraction has to go.
[Dropping this now does not preclude that virtual file systems make an
appearance in Qt at a later point in Qt 5's lifecycle. Hopefully with a
new and improved abstraction.]
Forward declarations for QFileExtension(Result) were dropped, as the
classes were never used or defined.
Tests using "internalized" classes will only fully run on developer
builds. QFSFileEngine was removed altogether from exception safety test,
as it isn't its intent to test internal API.
Change-Id: Ie910e6c2628be202ea9e05366b091d6d529b246b
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
addrlen must be initialize to the amount of space available
in the buffer.
Change-Id: I52945d780bba9d22aeaa7ac5a35a0e54dbea60dc
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Change the ifdefs in our own code (except openssl backend) to use the
new configure flag.
Change-Id: I8774734771c66b22164b5fae8fdb27814ac3df7b
Reviewed-by: Richard J. Moore <rich@kde.org>
Restored the Qt4 behaviour where attempting a https request using
QNetworkAccessManager fails with ProtocolUnsupportedError instead
of timing out or hanging.
Covered by existing autotests.
Task-number: QTBUG-17189
Change-Id: Iceb1ba6558c7d2b1af8ddf8d4ea9315a5b44d970
Reviewed-by: Richard J. Moore <rich@kde.org>
Connecting to an IPv4 mapped IPv6 address (e.g. ::FFFF:127.0.0.1)
requires the IPV6_V6ONLY socket option to be cleared.
This was causing tst_qtcpserver::ipv6ServerMapped autotest to fail.
The same change is not required on MacOS X - the test passes there.
Task-number: QTBUG-24351
Change-Id: I6c08b19f0daa12765da2d44792ffb17299322695
Reviewed-by: Markus Goetz <markus@woboq.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
If the DNS server returns a non authoritative host not found response,
then windows returns WSATRY_AGAIN error code.
This is now reported as HostNotFound and not UnknownError
Change-Id: I212985acd4e85ff4b2bdb6c57ec403405a7695fb
Reviewed-by: Miikka Heikkinen <miikka.heikkinen@digia.com>
Reviewed-by: Richard J. Moore <rich@kde.org>
QGlobalNetworkProxy (a singleton) had two phase construction, with
the second phase being called from QNetworkProxy's constructor.
This isn't necessary, and has been reported as causing deadlocks.
Although constructing socket engine handlers has side effects
(they add themselves to a list on construction and remove themselves
on destruction), this appears to be safe. The socket engine handlers
are only used while holding the list mutex, and any socket engines
created don't have any reference to the factory that created them.
With the new version, it is possible that two instances of
QHttpSocketEngineHandler and QSocks5SocketEngineHandler exist
temporarily if a Q_GLOBAL_STATIC initialisation race occurs.
This appears safe, because the loser of the race deletes its
handlers, which remove themselves from the global list as above.
Task-number: QTBUG-13088
Change-Id: I8cf520da717d8ab7d862ab89c6de13aea6d60ac3
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
QNetworkAccessManager was ignoring the supplied credentials, although
webkit seems to support these urls at a higher level.
Following the behaviour of browsers:
We use supplied credentials if authentication is required.
We add supplied credentials to the authentication cache.
We emit authenticationRequired signal if the credentials were wrong.
We do not use previously cached credentials for that url
Synchronous http requests fail, if the credentials were wrong.
Task-number: QTBUG-18107
Change-Id: If46e8eab1511ba8a0f4bbe0d4efaabc4df0b8ab4
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
In certain circumstances, this could cause the request to time out
(and repeatedly send bad authentication credentials to the server)
instead of failing with AuthenticationRequiredError.
Change-Id: Iff66b32f1d7268f21fd77b6620aae4b5d49d857f
Reviewed-by: Richard J. Moore <rich@kde.org>
This feature was implemented in 4.8, but documentation was not
updated at the time.
Task-number: QTBUG-18181
Change-Id: I657d7ab7aaf43b73b7bf8fd1cb76086522cf5c2b
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Change-Id: If2471bea27f095352ae8c28604e104b896fd97c7
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
-Some of the public classes were missing the \inmodule command.
-Fixed a sentence.
Change-Id: I88ebe12680c744e32253dc01c5ddb4292267caf9
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
QLocalServer could only listen to sockets it created.
Thi is not always possible as sockets may be passed
by socketpair() or have to be created locally by
other means. This adds a similar feature to QLocalSocket
where a native descriptor maybe used.
Change-Id: I43b0af179b3b868dd164d4e1fd312ff4546cf9ff
Reviewed-by: Michalina Ziemba <michalina.ziemba@nokia.com>
Reviewed-by: Tapani Mikola <tapani.mikola@nokia.com>
Reviewed-by: Harald Fernengel <harald.fernengel@nokia.com>
An old coding error meant that the C runtime abort() function was
being called instead of QFtp::abort() when cancelling an FTP download
using QNetworkReply::close()
Task-number: QTBUG-22820
Change-Id: Ib97fda9769b2b55a08c042c66c4444cb6216d2b1
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
There was a time when qsslsocket.h declared its private slots as
Q_PRIVATE_SLOT(d, void _q_connectedSlot())
But now they are correctly declared as
Q_PRIVATE_SLOT(d_func(), void _q_connectedSlot())
so the "#define d d_ptr" hack isn't needed.
Specifically, the define would break moc-generated code that refers
to the member d of a structure (which a future moc revision does,
namely QByteArrayData::d).
Change-Id: Ic94fa4d523fb17e8088973cfc0d090d5cce97267
Reviewed-by: Jonas Gastal <jgastal@profusion.mobi>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This fixes the compilation for Windows CE.
Include Windock2.h before Windows.h.
Change-Id: Iae1ab98239bb75b59c78460e0c0e48dfa1326032
Reviewed-by: Jonas Gastal <jgastal@profusion.mobi>
Reviewed-by: Miikka Heikkinen <miikka.heikkinen@digia.com>
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
QLocalServer had no way to set socket options
that more complicated servers require. The
first set of options allow setting of access
control on the sockets.
Change-Id: If4268c66462fc2e6cf1e70b1d5f56c76d2c69228
Reviewed-by: Harald Fernengel <harald.fernengel@nokia.com>
- Special #define to access HB seems no longer necessary
(it was causing about mismatching DLL import attributes).
Change-Id: I57cc7d57b12a67c1d549b053db81e1f198f87786
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
QSslSocket overrides one version of QAbstractSocket::connectToHost.
Since these functions were made virtual, this now hides the other
overloads.
Added a using statement to make the other overloads visible.
Change-Id: Ia48fdc9bd67936c75c25bb24dabd26a723bb7a05
Reviewed-by: Stephen Kelly <stephen.kelly@kdab.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Otherwise, the HTTP request won't be sent if bearer is not compiled.
Change-Id: I9a8171fde6fe5b2db0405b4af6e1d52ac4528502
Reviewed-by: Shane Kearns <ext-shane.2.kearns@nokia.com>
This file was removed in commit ba9302b8a9
Change-Id: Ib3a70877f6f6848c8cf072faf13da93abbe186f2
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
QBasicMutex is a POD and can be used as a static global object.
in qpicture.cpp factoryLoader is used only once, and under the mutex, so
there is no need for Q_GLOBAL_STATIC for it, it can be a function static
in qhostinfo_unix.cpp the code seemed wrong while compiled with
namespace and QT_NO_GETADDRINFO. I also could get rid of one include
because it was included earlier.
Change-Id: I3c700203c3e067266c20733f4bda8031446dbb86
Reviewed-by: Bradley T. Hughes <bradley.hughes@nokia.com>
SSL context was destroyed on disconnect. This makes it impossible to
decrypt buffered encrypted data. So if there is encrypted data in the
receive buffers, then don't destroy the ssl context until the socket is
destroyed.
Task-Number: QTBUG-23607
Change-Id: I16a7b4fa006647ec73049c90cdbc72686696850f
Reviewed-by: Jonas Gastal <jgastal@profusion.mobi>
Reviewed-by: Richard J. Moore <rich@kde.org>
Unify the includes for resolver function on unix-like platforms
to avoid build failures on BSD platforms.
Change-Id: I9accd7077d5a319a2c93642e011492d0fc779394
Reviewed-by: Holger Freyther <holger+qt@freyther.de>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
- Always use <qt_windows.h> as the last file to be included.
- Remove it from some headers, use Qt::HANDLE instead of HANDLE.
- Clean up #ifdef, use Q_OS_WIN for Windows/Windows CE.
- Add NOMINMAX to qt_windows.h to avoid problems with the
min/max macros.
- Remove <windows.h> from qplatformdefs.h (VS2005)
Change-Id: Ic44e2cb3eafce38e1ad645c3bf85745439398e50
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@nokia.com>
After discussion with Liang, I'm reverting it as he requested. This change put every header into the SYNCQT.HEADER_FILES twice for in-source builds, and the qtMODULEversion.h header did not include a path component.
This reverts commit 2fbc45b58bba860abf67fb28aa1319c9f4ededaf
Change-Id: Ie84cef19193ce5e49072f1f67a41140d9d2673b8
Reviewed-by: Liang Qi <liang.qi@nokia.com>
Create new QAbstractSocket::SocketError value that denotes a error in
the SSL library: SslInternalError
Create QAbstractSocket::SocketError value that denotes a error in data
provided by user cauding an SSL library error: SslInvalidUserDataError
Change-Id: I466a9389d9d7052efd8eddd1a2d6067ba26dfddb
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Shane Kearns <ext-shane.2.kearns@nokia.com>
Implementating this would cause massive SC breakage and give little
gain.
Change-Id: I56a6d302fdd0e8b53d1f9154f3eaf9e1ee429f9d
Reviewed-by: Giuseppe D'Angelo <dangelog@gmail.com>
Reviewed-by: Richard J. Moore <rich@kde.org>
%l has no meaning, should be %li according to the arguments.
Change-Id: Ife9ec524109e021ad723865445e80b6bad51a5c6
Reviewed-by: Richard J. Moore <rich@kde.org>
The debug message could derefence a null pointer, this crashed when
running ssl autotests
Change-Id: I176aaa9f3cf3c6cc1512cdc34db06d4c79f92e73
Reviewed-by: Richard J. Moore <rich@kde.org>
The intention is to add additional pause modes over time, this will
be easier if we can just test if a particular reason for pausing is
turned on. If we don't do this we'll end up having to check for each
enum value every time we check what is enabled.
Change-Id: I6b08f0e819b5593e3f6463c3dd175eff8625e155
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@nokia.com>
Reviewed-by: Shane Kearns <ext-shane.2.kearns@nokia.com>
The QDnsLookup class provides asynchronous APIs for performing
DNS lookups. For now, the following lookups are supported:
- A and AAAA
- CNAME as defined per RFC 1035
- MX as defined per RFC 1035
- NS as defined per RFC 1035
- PTR as defined per RFC 1035
- SRV as defined per RFC 2782
- TXT as defined per RFC 1035
Task-number: QTBUG-10481
Change-Id: I46c1741ec23615863eeca3a1231d5e3f8942495e
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
As in the past, to avoid rewriting various autotests that contain
line-number information, an extra blank line has been inserted at the
end of the license text to ensure that this commit does not change the
total number of lines in the license header.
Change-Id: I311e001373776812699d6efc045b5f742890c689
Reviewed-by: Rohan McGovern <rohan.mcgovern@nokia.com>
It was put in source tree before.
Task-number: QTBUG-20439
Change-Id: Ib52d9c2e83ae375aad259ddc74138bbc728b3ed0
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@nokia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Per ### Qt5 comment and the fact that it was already virtual on
QAbstractSocket.
Change-Id: If2d2b2f9cdec1ef4c5bf625e3ce0d6f2d9a7bdfd
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
QNetworkCookieJar now has the following virtual methods:
virtual bool validateCookie(QNetworkCookie &cookie);
virtual bool insertCookie(const QNetworkCookie &cookie, const QUrl &url);
virtual bool updateCookie(const QNetworkCookie &cookie);
virtual void deleteCookie(const QNetworkCookie &cookie);
Their implementation is such that the behavior the class previously
had(in memory storage of the cookies) is mantained.
Task-number: QTBUG-23145
Change-Id: I1420894d31e8072eca6903c3c7ffd6f06205a257
Reviewed-by: Peter Hartmann <peter.hartmann@nokia.com>
Reviewed-by: Alexis Menard <alexis.menard@openbossa.org>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Shane Kearns <ext-shane.2.kearns@nokia.com>
pause and resume is currently only supported upon emitting the
QSslSocket::sslErrors() signal. The API was added in QAbstractSocket to
also support QAbstractSocket::proxyAuthenticationRequired() in the
future.
This is the first patch to support that feature on the socket level,
another patch will follow to support sslErrors() and
authenticationRequired() in QNetworkAccessManager / QNetworkReply.
Task-number: QTBUG-19032
Change-Id: Ide2918268590ab9a01454ab26cb7fdca3dc840ab
Reviewed-by: Shane Kearns <ext-shane.2.kearns@nokia.com>
The original architecture of the QtNetwork bearer support hosted the
engines in the application's main thread, but this causes some problems.
If the QNetworkConfigurationManager is constructed in a worker thread,
then it is populated asynchronously without any notification when it is
done (the app gets incomplete or missing results)
Fixing that by restoring the earlier behaviour of using blocking queued
connections to wait for the lists to be populated caused a regression,
as some applications deadlock because the main thread is waiting on the
worker thread at this time.
By introducing a dedicated worker thread for the bearer engines,
QNetworkConfigurationManager can be safely constructed in any thread
while using blocking queued connections internally.
Task-number: QTBUG-18795
Change-Id: Iaa1706d44b02b42057c100b0b399364175af2ddb
Reviewed-by: mread
(cherry picked from commit 5f879c55e531165cc2569b03c3796d0f33d0a0b7)
Reviewed-by: Jonas Gastal <jgastal@profusion.mobi>
Reviewed-by: Murray Read <ext-murray.2.read@nokia.com>
Reviewed-by: Alex <alex.blasche@nokia.com>
These defines were there to aid in the commercial
licensing scheme we used long ago, and are no longer needed.
Keep a QT_MODULE(x) define so other modules continue compiling.
Change-Id: I8fd76cd5270df8f14aee746b6cf32ebf7c23fec7
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Change-Id: Iae377505e36ae1239be7ce52c773dc2a4f4a9767
Reviewed-by: Jonas Gastal <jgastal@profusion.mobi>
Reviewed-by: Peter Hartmann <peter.hartmann@nokia.com>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
The following methods have been made virtual:
setReadBufferSize()
socketDescriptor()
setSocketDescriptor()
socketOption()
setSocketOption()
waitForConnected()
waitForDisconnected()
Now that these methods are virtual we no longer need the nasty
polymorphism workarounds for QSslSocket.
Change-Id: I319989b6cdb025ba33d7d53ae90f3a6a3b6b1b7b
Reviewed-by: Richard J. Moore <rich@kde.org>
Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
Reviewed-by: Peter Hartmann <peter.hartmann@nokia.com>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
QAuthenticator used it for the convinience of QHttpSocketEngine only.
QHttpSocketEngine has now been ported to use QHttpNetworkReply to parse
HTTP responses.
Change-Id: Idf6e70aa76613aad6e3d789d81ca1b4fd73575c2
Reviewed-by: Peter Hartmann <peter.hartmann@nokia.com>
Change-Id: I6db7211fcf6b24bd75e360645bbb2fdf1ef8a8bc
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@nokia.com>
Reviewed-by: João Abecasis <joao.abecasis@nokia.com>
- Missing return value
- Wrong format for qint64
Change-Id: Id0de58c85b7c8ed2a62f7237fd23e6c5a5ac92ec
Reviewed-by: Peter Hartmann <peter.hartmann@nokia.com>
Also, use QStringLiteral instead of QLatin1String.
Change-Id: I232fc02a56261929864c2ea66993ef1c74bc1237
Reviewed-by: Peter Hartmann <peter.hartmann@nokia.com>
Reviewed-by: Robin Burchell <robin+qt@viroteck.net>
Reviewed-by: Jonas Gastal <jgastal@profusion.mobi>
