As this class is a subject of the library hook data, there must be a
solid understanding of the member's alignment and padding across
different architectures. By reordering the layout, this patch provides
a clearer way of adding new members to the class.
Bump the TypeInformationVersion field in qtHookData, to notify the
Qt Creator developers that the offset of QFilePrivate::fileName was
changed and dumpers should be adapted.
Change-Id: Ied8b69bdeb9da50ff05aba2107bc75509674b18e
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
It works as follows:
- user calls write(const QByteArray &);
- this function keeps a pointer to the chunk and calls a regular
write(data, len);
- write(data, len) calls a virtual writeData();
- subclass calls a new QIODevicePrivate::write();
- QIODevicePrivate::write() makes a shallow copy of
the byte array.
Proposed solution is fully compatible with existing subclasses.
By replacing a call to d->writeBuffer.append() with d->write(),
subclasses can improve their performance.
Bump the TypeInformationVersion field in qtHookData, to notify the
Qt Creator developers that the offset of QFilePrivate::fileName was
changed and dumpers should be adapted.
Change-Id: I24713386cc74a9f37e5223c617e4b1ba97f968dc
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
Use pro2cmake with '--api-version 2' to force regenerate
projects to use the new prefixed qt_foo APIs.
Change-Id: I055c4837860319e93aaa6b09d646dda4fc2a4069
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
There is no reason for keep using our macro now that we have C++17.
The macro itself is left in for the moment being, as well as its
detection logic, because it's needed for C code (not everything
supports C11 yet). A few more cleanups will arrive in the next few
patches.
Note that this is a mere search/replace; some places were using
double braces to work around the presence of commas in a macro, no
attempt has been done to fix those.
tst_qglobal had just some minor changes to keep testing the macro.
Change-Id: I1c1c397d9f3e63db3338842bf350c9069ea57639
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
I'd have preferred to use QArrayDataPointer<ushort> for QString, but
that option wasn't the best one. QArrayDataPointer try to do some
operations using QArrayDataOps and that would expand to unnecessary
code. What's more, the existing code expected to be able to modify and
access the d pointer.
Instead, this commit introduces QStringPrivate (named differently from
QStringData to catch potential users), which contains the three
members. This POD class is also used in QJsonValue to store the
"inlined" QString. QHashedString in qtdeclarative will need a similar
solution.
Change-Id: I33f072158e6e2cd031d4d2ffc81f4a8dbaf4e616
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Add QGenericArray to simplify operations. This class can be shared by
other tool classes. If there is nothing else to share it, we can move
the code onto qvector.h. The one candidate is QList.
All tests pass and valgrind is good.
Change-Id: Ieaa80709caf5f50520aa97312ab726396f5475eb
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Don't manage the ref-count yourself, as this requires the code to use
the QAtomic copy ctor, which we want to remove going forward. Using
QSharedData, we can let the compiler write the code for us.
Since 'ref' this way moves to the first spot in the list of effective
members, creating a 4B hole between itself and 'msecs', swap 'status'
and 'msecs' to fill the hole:
offset: 0 8 16 24
| | | |
without v v v v
adj.mnt: |*R*| | msecs | S | U | TZ....
before: | msecs | S | U |*R*| | TZ...
after: |*R*| S | msecs | U | | TZ....
This keeps the padding out of the critical first word, which improves
latency. That said, for accessing the members the old layout surely was
optimal. This layout optimizes copies and pessimizes access to 'msecs'
on 32-bit platforms without the Critical Word First optimization.
Requires adjustments to tst_toolsupport and the qhooks version.
Also default members using NSDMI, consequently drop the manual default
ctor.
Change-Id: I3c48e68694ad29b28a13aa47ea0f283fae52edd7
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Adn create that data structure on demand on the heap.
This reduces the size of QObjectPrivate if there are no
connections. If we have connections, it'll use the same
amount of allocations and memory as before.
Change-Id: I900f6980a2cd8a5f72c3ad18697b5dd49100217d
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
Measurements show that it's just almost as fast to simply query
the connectionlist directly and avoid both the memory
overhead of the bitfield and the associated bookkeeping.
For connected signals, the difference is not relevant at all.
With a signal that was never connected, removing the bitfield will
cause signal emission to be ~2.5% faster. And if you ever disconnect
from a signal, the bitfields might not be accurate and this can
cause a major slowdown.
Here are some numbers to validate this. All times are measured
in ms for 100M signal emissions:
without change with change
string based connect: 3817 3836
pointer based connect: 4552 4571
not connected: 493 479
disconnected: 2113 559
Change-Id: Ia2c85036afaa7f991b883c8ff812f69cf4580f7e
Reviewed-by: Olivier Goffart (Woboq GmbH) <ogoffart@woboq.com>
Reviewed-by: hjk <hjk@qt.io>
On 32-bit arm platform, qint64 gets aligned differently than on 32-bit
x86. First difference between the platforms on QFilePrivate member
offset happens in QFileDevicePrivate::cachedSize:
- On 32-bit x86 it's offset is 148 (4-aligned)
- On 32-bit arm it's offset is 152 (8-aligned) and offsets of all the
remaining members are +4 compared to x86
- On 64-bit architectures the offsets are the same
Change-Id: If110da27ea08504e78b167c0a21599420eaa9630
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Storing them in a single byte is the first step towards the Short
QDateTime Optimization.
The bump in the "private version" by 10 is to accommodate possible
changes in the Qt 5.7 branch.
Change-Id: Id5480807d25e49e78b79ffff144a59420457bcf0
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Also, bump the TypeInformationVersion field in qtHookData, to notify
the Qt Creator developers that the offset of QFilePrivate::fileName
was changed and dumpers should be adapted.
Change-Id: I71bc5f509b733c0ab3430cd47ff08961f0388839
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
By default, the read chunk size is QIODEVICE_BUFFERSIZE and the write
chunk size is 0 (which means that we don't use the internal write buffer).
Derived classes may override these values to define the size of
QIODevice's write buffer or to optimize the read buffer use.
Bump the TypeInformationVersion field in qtHookData, to notify the
Qt Creator developers that the offset of QFilePrivate::fileName was
changed and dumpers should be adapted.
Change-Id: Ib732bc94be8da8a5514a6e5dcc04445895f130d8
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
Some sequential devices allow data to be partitioned into several
channels that have the property of independently sequenced delivery.
Supporting such devices uniformly requires a unified API which provides
the user with a uniform concept of multistreaming.
This patch is based on QProcess's multiplexing model and introduces
the following features:
- ability to get the number of channels;
- multiple internal read/write buffers;
- channel selection functions;
- notification signals on channel activity.
To keep the source code compatible with single-channel implementations,
introduce a private class that references the current read buffer and
hides multistreaming internals from the user.
Bump the TypeInformationVersion field in qtHookData, to notify the
Qt Creator developers that the offset of QFilePrivate::fileName was
changed and dumpers should be adapted.
[ChangeLog][QtCore] Added multistreaming to QIODevice.
Change-Id: Idcaa6a618927c101c4c7284d2a633913be6a6ee2
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@theqtcompany.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
Introduce a transaction mechanism that gives the ability to read the
data atomically. Current implementation supports transactions for both
types of devices. For sequential devices, it records the whole input
stream during transaction. For random-access devices, device position
is saved when transaction starts. If an error occurs, the application
may be able to recover the input stream by rolling back to the start
point.
Also, QIODevice::peek() was rewritten to make use of transactions
internally. The replacement of QIODevicePrivateLinearBuffer by
QRingBuffer is closely entangled with that, which makes it unfeasible
to do separately.
Bump the TypeInformationVersion field in qtHookData, to notify the
Qt Creator developers that the offset of QFilePrivate::fileName was
changed and dumpers should be adapted.
[ChangeLog][QtCore] Added QIODevice's startTransaction(),
commitTransaction(), rollbackTransaction(), isTransactionStarted()
functions to support the read transactions.
Task-number: QTBUG-44418
Change-Id: I3564b343ebeeaaf7c48a1dcdb7ef0a7ffec550f2
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@theqtcompany.com>
This extends the test suite introduced in 497f0af1f7 for
a known-to-be-good case also for 32 bit systems.
Change-Id: Ia231bcb9b0102c28483d932be18767662b7a6afd
Reviewed-by: Christian Stenger <christian.stenger@theqtcompany.com>
The keyword no longer has a meaning for the new CI.
Change-Id: Ibcea4c7a82fb7f982cf4569fdff19f82066543d1
Reviewed-by: Simon Hausmann <simon.hausmann@theqtcompany.com>
Qt Creator needs to be able to determine structure sizes and data
member offsets for certain private types even in the absence of debug
information.
It is sufficient to keep and test the actual data sets on the Qt Creator
side, as long library provides a hint which data set is needed.
So far, HookDataVersion was meant to be used for that purpose. To
make it more explicit, this patch introduce a TypeInformationVersion
field in qtHookData.
Change-Id: Ia1c3c6f62f314d63c4df289ef456f047c5e84cf4
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Qt copyrights are now in The Qt Company, so we could update the source
code headers accordingly. In the same go we should also fix the links to
point to qt.io.
Outdated header.LGPL removed (use header.LGPL21 instead)
Old header.LGPL3 renamed to header.LGPL3-COMM to match actual licensing
combination. New header.LGPL-COMM taken in the use file which were
using old header.LGPL3 (src/plugins/platforms/android/extract.cpp)
Added new header.LGPL3 containing Commercial + LGPLv3 + GPLv2 license
combination
Change-Id: I6f49b819a8a20cc4f88b794a8f6726d975e8ffbe
Reviewed-by: Matti Paaso <matti.paaso@theqtcompany.com>
Most of the QIODevice functions already have a locally cached
"sequential" flag. Make the rest of them follow this strategy.
This eliminates the need to use private caching members.
Change-Id: I0edb2c9b7c5f411c5bee25c425e7b40e3c9021d3
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Reviewed-by: hjk <hjk@theqtcompany.com>
This extends the test suite introduced in 497f0af1f7 for
a known-to-be-good case.
Change-Id: Ib574ecfdb8e9d91985d6df8c092896581d1d06ff
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
This upstreams parts of Qt Creator's tst_offset test to serve as
an early warning system by testing private implementation details
that are used in Qt Creator's data structure "pretty printing"
facility.
While the tested implementation details can be changed without
breaking binary or source compatibility, downstream tools like
Qt Creator depend on them.
If this test breaks, you are kindly asked to coordinate with the
downstream stakeholders to avoid tool breakage.
Change-Id: I1286efcec9bef105f80c2163a4f66f5c43f3a218
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>
