When IPv4 and IPv6 are supported by a server, QHttpNetworkConnection
will start up two connections and pick the network layer of the one
that finish first. In this case the channel with index 1 is used for
IPv6. When IPv6 wins, there is no channel at index 0. This situation
needs to be respected and we should try to use existing channels first
when there is a next request.
This is especially important when TLS session resumption is used.
Creating a new channel will cause to lose the ephemeralServerKey
used in the first connection.
Fixes: QTBUG-93295
Pick-to: 5.15 6.1
Change-Id: Ic9dc6a24ef793a29c2652ad37bc11120e2e6ceef
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
[ChangeLog][QtNetwork][QLocalSocket] The waitFor*() functions on
Windows now support duplex operation, as they already did on Unix.
As a side effect, this restores the behavior that a single call to
waitForReadyRead() won't emit both readyRead() and disconnected(),
which also matches Unix behavior. The groundwork for that misbehavior
was laid by incorrect refactoring in d1a671b69 already, but at this
point it was harmless, as the pipe couldn't be newly closed after a
successful read. That changed with f265c87e0, which made the queuing
of signals async.
Change-Id: I1eb80e8f147bb58825143e0fe1e4300c59ae0fbb
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
The name of attribute has a suffix 'Attribute' which was missing.
Pick-to: 6.1
Fixes: QTBUG-93416
Change-Id: I9ce96efb38caa03957ddbdb13a5e61767e8e74f1
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
An include statement appeared after opening the Qt namespace, thus
injecting symbols in there. Move it outside.
Change-Id: I8e95e821b36ad4e4ceed5b0645bf8ebf7e531e06
Pick-to: 5.15 6.0 6.1
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
With Qt 6 we made HTTP/2 default, which exposed missing handling of
401 Unauthorized (and 407 Proxy Authentication Required).
In HTTP/1.* we would handle this after the response had finished, while
handling the status code. For h2 this path isn't used since it is
heavily reliant on the structure we have for HTTP/1.* (one request per
channel). So we must handle the status code and header directly.
Having that part fixed exposed another issue - when resetting/rewinding
uploaded data we were not resetting the 'totallyUploadedData' counter in
the reply (this, in turn, exposed another small issue). Because of that
we did not actually send any data on the retry, only sending the
content-length followed by no data.
Finally, the small issue mentioned in the previous paragraph was how we
check if we have uploaded all our data. It was only checking if the
byte-device was atEnd(), which it was. But only because it had not yet
prepared any data for us.
Fixes: QTBUG-91284
Pick-to: 6.1 6.0 5.15
Change-Id: I798d105b02688b18a02897cc476f19f57a47f98f
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
And adjust configure.json files that were stale
Amends d385158d52
Change-Id: I851838a12c3773a6e8119ebc7f1de941ae7fe224
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
The topic existed for QRegExp but is not relevant in QRegularExpression.
Pick-to: 6.1
Change-Id: Ife9d42b5b03163cfefff19725409a3feac1cddb2
Reviewed-by: Paul Wicking <paul.wicking@qt.io>
When cross compiling from a case sensitive file system, casing
matters, and mingw headers and import libraries consistently
use lowercase.
This was uncovered by d385158d5213ef568b7629e2aa4a818016bbffac;
prior to that, the schannel TLS plugin didn't end up built (at
least when cross compiling).
Fix other similar cases that can be found by grepping the repo.
Change-Id: Ia696e17b7aaa979d7b7f5b0801383f338a8b585b
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
All TLS (and non-TLS) backends that QSsl classes rely
on are now in plugins/tls (as openssl, securetransport,
schannel and certonly plugins).
For now, I have to disable some tests that were using OpenSSL
calls - this to be refactored/re-thought. These include:
qsslsocket auto-test (test-case where we work with private keys),
qsslkey auto-test (similar to qsslsocket - test-case working with
keys using OpenSSL calls).
qasn1element moved to plugins too, so its auto-test have to
be re-thought.
Since now we can have more than one working TLS-backend on a given
platform, the presence of OpenSSL also means I force this backend
as active before running tests, to make sure features implemented
only in OpenSSL-backend are tested.
OCSP auto test is disabled for now, since it heavily relies on
OpenSSL symbols (to be refactored).
[ChangeLog][QtNetwork][QSslSocket] QSslSocket by default prefers 'openssl' backend
if it is available.
[ChangeLog][QtNetwork][QSslSocket] TLS-backends are not mutually exclusive anymore,
depending on a platform, more than one TLS backend can be built. E.g., configuring
Qt with -openssl does not prevent SecureTransport or Schannel plugin from being
built.
Fixes: QTBUG-91928
Change-Id: I4c05e32f10179066bee3a518bdfdd6c4b15320c3
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Add SOLARIS cmake platform definition.
Add settings for QT_DEFAULT_MKSPEC so that qplatformdefs.h can be found.
Solaris has its gssapi symbols in libgss.
Solaris supports @ORIGIN.
Solaris ld does not support --dynamic-list needed for reduce relocations.
Make solaris fail the reduce relocation test.
getauxval is specific to GNU libc and some other libc implementations on
Linux but sys/auxv.h is not. The bootstrap uses sys/aux.h as the only
indication for getauxval. This breaks builds on Solaris, so only make
sys/auxv.h an indicator for getauxval on linux or glibc based systems.
Solaris uses X11 so add it to the X11_SUPPORTED list.
Solaris network libraries for sockets etc are in socket and nsl.
ifreq does not have a member ifr_ifindex on Solaris, it uses
ifr_index. Add test to check if ifr_index is a member of ifreq.
The first struct in the in_addr union on solaris is defined as four
uint8_t, therefore four arguments are needed for its initializer list.
Change-Id: Ieed4c1bbac8559a7ae1db9c4e1e91f609f150270
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Xcode allows building a project targeting either the device or
simulator sysroot in one single build dir, but for the sysroot
switching to work there should be no linker or compiler flags
referencing absolute paths of a specific sysroot.
During CMake configuration of a project targeting iOS, all found
system libraries will be within one single sysroot, either the device
one or the simulator one, whichever one was passed to
CMAKE_OSX_SYSROOT. CMake will then generate the Xcode project
and pass those absolute paths, which makes sysroot switching within
Xcode not work.
To avoid that, the CMake documentation recommends passing linker and
framework flags of the form '-lfoo' and '-framework bar' instead of
absolute paths. Xcode then takes care of setting the correct framework
search path.
Zlib is one of the libraries found in the iOS sysroot and thus passed
as absolute path.
To avoid that, create a new FindWrapZLIB find script. The target it
creates will pass the absolute path to the library on non Apple
platforms and an -lz linker flag on Apple platforms (macOS and iOS).
To avoid issues with target global promotion when system PNG package
is found, ensure that a found ZLIB::ZLIB target is promoted to global
manually in src/gui/configure.cmake.
Pick-to: 6.1
Change-Id: I8bd8649be4f680a331ad51925f27cb9d13ac5e5f
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
Because take() is deprecated, and these pointers are meant to leave the
scope in some branches.
Pick-to: 6.1
Change-Id: I5432d91a28f4c5c8c17fadf7ce3bcd41716e216a
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Presumably the code at some point would do a
QByteArray::fromRawData-style thing. But now it doesn't do that so
the current code was a bit strange. It would map the content of the file
to memory only to then copy the content into a QByteArray. Then it
reparents the file to the QBuffer, keeping it alive even if its not
needed.
Fixes: QTBUG-92838
Pick-to: 6.1 6.0 5.15
Change-Id: I88f8cd1b64e0fd13d08b5cc4df44661e216da340
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
For QProcess, there is no point in suppressing recursive
QWPR::readyRead() emission, as the former manages this logic itself. On
top of that, the non-recursive nature of QWPR::readyRead() indirectly
disallowed reading from the channels inside
QProcess::waitForReadyRead(), if that is called from a slot connected
to QProcess::readyRead().
QWPW had two signals, one allowing recursion and one not.
This commit allows recursion of QWPR::readyRead() and
QWPW::bytesWritten(), and moves recursion suppression to the higher-
level classes. This makes the code more uniform and efficient, at the
cost of a few duplicated lines.
Change-Id: Ib20017fff4d92403d0bf2335f1622de4aa1ddcef
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
As suggested in the API review.
Amends 37bd7b5733
Pick-to: 6.1 6.1.0
Change-Id: Ic3e8567f349568dc3b4dbf79be27c304b39480cf
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
On Windows, on exit, the backend will deref (and then implicitly start
deletion of) a COM object. This object tries to communicate with an
object in another thread, though it seems this other thread quits before
the main thread in _most_ cases.
To get around this we move the deletion to earlier in the program. While
this is only reported as a Windows issue it makes for more consistent
behavior if all platforms behave the same.
Document and test that recreation of QNI works as expected after the
destruction (and recreation) of QCoreApplication.
Amends: 0875626e22
Fixes: QTBUG-92568
Pick-to: 6.1 6.1.0
Change-Id: Iffc07f38673019aa059efd4d64d2ad706a03f6fe
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Reviewed-by: Oliver Wolff <oliver.wolff@qt.io>
It wasn't meant to be public.
Pick-to: 6.1 6.1.0
Change-Id: Ifa7fff48f82b96bdfa277677d3988dc8881b8c2a
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
And the code for loading whatever backend is sorted first.
Though, looking at the code it would've never returned 'true' anyway.
Pick-to: 6.1 6.1.0
Change-Id: I7bc2c740e8cb5343e5843cb1d65715d236b92a25
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Takes advantage of Linux's and Android's support for abstract namespace
when binding sockets, which is independent of the filesystem (see man
entry for unix domain sockets).
To make QLocalServer and QLocalSocket use an abstract socket address,
one needs to set the socket options to QLocalServer::AbstractNamespaceOption.
Fixes: QTBUG-16090
Change-Id: Ia9f9c9cc1ac5c28f9d44b0a48d854a7cfbd39b11
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
We must stop the pipe writer before closing the handle on which it
operates.
Pick-to: 5.15 6.0 6.1
Change-Id: I4765dd8393167fe2453653aba76a097b8ace8e3d
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
It should be disconnected() instead of disconnectedFromServer().
Pick-to: 6.0 6.1
Change-Id: I2b816e9c92d675ea5a7a26b4752e831c30950ef2
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
This design decision was intentional and reflects the fact that we
cannot have inheritance tree 'linear' as before, since the base
interface is an abstract class without any data-members.
Also, fix a warning about size_t -> int implicit conversion.
Fixes: QTBUG-91866
Change-Id: Icbc73c22d8adf72bbd8949125d4404ea6da69c6d
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
(cherry picked from commit 1a7faeef94dcf6826c6ee564a1bfb31ebb952e56)
This can only be completed in Qt 7, due to BC guarantees. I've only
updated the source code for QHostAddress and its unit test and even then
I did not touch the documentation. This needs to be completed in Qt 7.0
Change-Id: I26b8286f61534f88b649fffd166b695882f8f3b5
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Because this is misleading:
socket.bind(QHostAddress::AnyIPv4);
The conversion of enum to int has preference over the construction of
QHostAddress from enum, so that ends up calling bind(6), which attempts
to bind to privileged port 6 and just returns with error. Meanwhile,
socket.bind(QHostAddress::AnyIPv4, 0);
does construct the QHostAddress because that is preferred to converting
the int to the BindMode enum.
Ideally we'd simply add the overload to QAbstractSocket but we can't do
that because QHostAddress depends on QAbstractSocket. So I've added a
Qt7 task to invert that dependency.
Change-Id: I26b8286f61534f88b649fffd166b683266597796
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
It was reasonable while backends were a part of QtNetwork.
Now if moving them outside (or just trying to implement
a new backend as a plugin), accessing data-members of
QSslConfigurationPrivate means that any plugin knows
about memory layout actual only for the version of Qt
it was built with/for. Instead, we have to use the
public class. Since it does not have all needed setters
and some data-members have no access at all, we
provide an API in QTlsBackend (which stays a part
of QtNetwork) that knows the actual memory layout.
Task-number: QTBUG-65922
Change-Id: I5ca1de4f982b4b11d9a87c4b40413367dcb83c16
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
(cherry picked from commit 31cc0df7607a4d5887812c304aac0001c2cd7705)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
All backend-specific code is now separated and removed
from QSslSocket(Private) code. The original code is mostly
preserved to avoid (as much as possible) regressions (and
to simplify code-review).
Fixes: QTBUG-91173
Task-number: QTBUG-65922
Change-Id: I3ac4ba35d952162c8d6dc62d747cbd62dca0ef78
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
(cherry picked from commit 9391ba55149336c395b866b24dc9b844334d50da)
Commit 21d3916817 added the 64-bit
version, so qsizetype now works cross-platform. The casts were added to
make qtbase compile on commit df853fed66.
Change-Id: I26b8286f61534f88b649fffd166c409c5c232230
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Silence compiler warnings from deprecated QScopedPointer::swap after
fe9d7bf759.
For Qt 7, both QSslCipher and QNetworkAddressEntry should be made into
implicitly shared classes.
Change-Id: Idfd5ec4b5a0f156f212d57684822a3cd1d88de1a
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
So to avoid QScopedPointer::swap deprecation. Side note,
QSslError is not implictly shared (although it should be, but
that's a job for Qt 7).
Change-Id: I42f7abffa81d72aac5af157074a0c3cbd20ba253
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Having a `break` after a `return` is pointless as it's never reached.
Change-Id: I30877e926c006fac45681f547e97a55410f02e43
Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
[ChangeLog][QtNetwork][QNetworkInterface] Fixed the reporting the "DNS
eligibility" factor (QNetworkAddressEntry::dnsEligibility) for global IP
addresses. Previously, QNetworkInterface interface erroneously reported
all global addresses as eligible, regardless of whether they had already
been deprecated by the OS or were temporary in the first place.
From the tst_QNetworkInterface::dump output (netmask & broadcast
suppressed for readability, IPv6 prefixes changed to protect the
innocent, but the local part is exactly as found on my system):
address 0: 192.168.26.33/24 dns-eligible preferred:33546998ms valid:33546998ms
address 1: 2001:db8::800/128 dns-eligible preferred:264817000ms valid:264817000ms
address 2: fd00::800/128 dns-eligible preferred:forever valid:forever
address 3: 2001:db8::2f5b:342c:fc53:b9b2/64 dns-ineligible preferred:12422000ms valid:264817000ms
address 4: fd00::9351:adff:333d:5c8d/64 dns-ineligible preferred:12421999ms valid:531402999ms
address 5: fd00::7297:9516:fbb4:17ac/64 dns-ineligible preferred:0ms valid:445585999ms
address 6: fd00::fdc8:e620:691:3b44/64 dns-eligible preferred:forever valid:forever
address 7: 2001:db8::ae82:a01:5a8a:e210/64 dns-ineligible preferred:0ms valid:264816999ms
address 8: 2001:db8::c673:e0a2:8927:2118/64 dns-eligible preferred:264816999ms valid:264816999ms
address 9: fe80::bd89:b998:4aeb:a5d0%bond0/64 dns-ineligible preferred:forever valid:forever
Prior to this commit, only address 9 was showing as ineligible.
Addresses 1 and 2 come from DHCPv6 and are thus always eligible.
Addresses 3, 4, 5, and 7 are temporary addresses added on account of
RFC 4941 and are therefore ineligible (they have the IFA_F_TEMPORARY
flag set). Note how 5 and 7 have also stopped being preferred. Address 3
is the one I see when I go to ip6.me and address 4 is the one used to
reach the router.
Addresses 6, 8, and 9 are the "permanent" addresses that would normally
be based on the MAC address of the interface, but are actually in
"stable privacy" mode for me (RFC 8064). Address 9 is the link-local
one, which makes it ineligible, leaving the other two stable addresses
eligible.
Addresses 1, 3, 7, and 8 are global, from a DHCPv6-PD delegation from my
ISP, which is why they have finite lifetime. Addresses 2, 4, 5, and 6
are from my local Unique Local Address (ULA) prefix, which is why 2 and
6 are valid forever.
Pick-to: 6.1
Change-Id: If8b43dc9678c4b4ba9c1fffd1668fdcae873c6bd
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Renamed the private part of the namespace to QTlsPrivate.
Fixes: QTBUG-91531
Change-Id: Id2d9c458d8d958b35e437eee42889177dad0a7a7
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
(cherry picked from commit e4efa00c72c7011f4e5c27722c2a522b30274610)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>
This is to avoid false positives.
By default files are large if uncompressed size > 10 MB. Only configurable internally.
Also add auto tests.
Task-number: QTBUG-91392
Pick-to: 6.0 6.1
Change-Id: I32258cb7c957f2a23a05157ba4ed5c0af2ba585e
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
The Android build failed, because the namespace QNativeInterface could
not be found. Include the header file where it's defined.
Change-Id: If53eda4d4d4e6d5e66787d74c714215721ba0b60
Reviewed-by: Assam Boudjelthia <assam.boudjelthia@qt.io>
Lars Schmertmann
Assam Boudjelthia
Alex Trotsenko
Timur Pocheptsov
Giuseppe D'Angelo
Morten Sørvig
Mårten Nordheim
Alexandru Croitor
Volker Hilsheimer
Martin Storsjö
Niclas Rosenvik
Qt CI Bot
Albert Astals Cid
Piotr Mikolajczyk
Thiago Macieira
Robert Löhning
Tang Peng
Paul Wicking
Øystein Heskestad
Joerg Bornemann