Android Marshmellow intentionally forbids use of hard links. See
https://code.google.com/archive/p/android-developer-preview/issues/3150
However, instead of using EPERM or another error code that indicates the
hard linking operation itself has a problem but there are no other
problems, Android developers stupidly chose to use EACCES, an errno code
that only indicates permission problems.
In any case, since the call will never succeed, we shouldn't even try.
Task-number: QTBUG-64103
Change-Id: I9e2892cb6c374e93bcb7fffd14fc5d1082bd60a3
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
The RENAME_NOREPLACE flag is supported for all Linux local filesystems,
since that can be easily checked by the VFS layer (it knows which files
exist and which ones don't). For non-local filesystems, the backend
needs support and that might need server-side support too. So we may get
EINVAL errors for those, in which case we fall back to link/unlink,
which in turn can fall back to rename().
EINVAL can also happen if we attempt to make a directory a
subdirectory of itself. In that case, we will attempt to link() it,
which will result in EPERM as we can't hardlink directories. Then we try
rename() again, which should result in the expected EINVAL.
Task-number: QTBUG-64008
Change-Id: Icaa86fc7b54d4b368c0efffd14f09ca23602dd2e
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Convert QSysInfo/QOperatingSystemVersion to __builtin_available where
required or possible, or to QOperatingSystemVersion where
__builtin_available cannot be used and is not needed (such as negated
conditions, which are not supported by that construct).
Change-Id: I83c0e7e777605b99ff4d24598bfcccf22126fdda
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This file uses functions declared in this header, and it is not pulled
in transitively on all platforms.
Change-Id: I6654118883a8dc22dacf1beb7b9b1c662719d25c
Reviewed-by: Tuomas Heimonen <tuomas.heimonen@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Conflicts:
examples/examples.pro
qmake/library/qmakebuiltins.cpp
src/corelib/global/qglobal.cpp
Re-apply b525ec2 to qrandom.cpp(code movement in 030782e)
src/corelib/global/qnamespace.qdoc
src/corelib/global/qrandom.cpp
src/gui/kernel/qwindow.cpp
Re-apply a3d59c7 to QWindowPrivate::setVisible() (code movement in d7a9e08)
src/network/ssl/qsslkey_openssl.cpp
src/plugins/platforms/android/androidjniinput.cpp
src/plugins/platforms/xcb/qxcbconnection.cpp
src/plugins/platforms/xcb/qxcbconnection_xi2.cpp
src/widgets/widgets/qmenu.cpp
tests/auto/widgets/kernel/qwidget_window/tst_qwidget_window.cpp
Change-Id: If7ab427804408877a93cbe02079fca58e568bfd3
Consider the following:
/root/target - a file
/root/path/link -> ../target
/root/path/other/exe - executable
Running from /root/path/other.
exe is:
#include <QDebug>
#include <QFileInfo>
int main()
{
qDebug() << QFileInfo("../link").symLinkTarget()
return 0;
}
The link references /root/target, but the current output is
/root/path/target.
The link doesn't depend on the PWD. It depends on its own directory.
Change-Id: I61e95018154a75e0e0d795ee801068e18870a5df
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
FICLONE only works on (currently) btrfs and xfs, and then only if it's
the same mount, so it's of very limited use. There are a couple other
techniques we may try that do not involve I/O through user-space, though
not immediate:
- sendfile(2) can be used on regular files, even across mountpoints
- sendfile(2) can be used on block devices too, but we need to get the
device's size first
- splice(2) can be used on pipes (FIFOs)
We only implement the first technique (earlier iterations of this patch
implemented all).
Change-Id: I81480fdb578d4d43b3fcfffd14d4b47cd70495a3
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
For some reason, fcopyfile(3) works on directories, so we need to make
we aren't operating in one.
Change-Id: I81480fdb578d4d43b3fcfffd14d4f2f485348515
Reviewed-by: Jake Petroules <jake.petroules@qt.io>
Unified headers now defines _POSIX_THREAD_SAFE_FUNCTIONS but not all
libc functions are available in all Android API versions.
Change-Id: I01c94f0b89e7f8aa8575e7bbda28d9fe41a68ff1
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Making system calls with empty file names is not a good idea. When you
run qmake $srcdir, you see this in strace:
stat("", 0x7ffed229e250) = -1 ENOENT (No such file or directory)
(twice)
I've also inlined the isEmpty() function for better code generation.
Some functions take QSystemError and some don't. That needs to be
corrected at some point, possibly with something like std::expected.
Change-Id: I1eba2b016de74620bfc8fffd14ccbfa162f93631
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
The stat::st_flags and fchflags(2) syscall are marked:
HISTORY
The chflags() and fchflags functions first appeared in 4.4BSD.
Change-Id: I81480fdb578d4d43b3fcfffd14d4fd23bd27e37e
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Reviewed-by: Tobias C. Berner <tcberner@FreeBSD.org>
This system call, new in Linux 4.11, gives us the file birth time. It's
also extensible, representing the fourth generation of stat(2) on Linux
(the original sys_stat(), sys_newstat(), sys_stat64() and now
sys_statx()), not to be confused with glibc's __xstat function, which
wraps a call to stat64. Anyway, the new one is designed to be extensible.
Now we get birth times on ext[34] on Linux too:
Name: .
Path: . (/home/tjmaciei/src/qt)
Size: 4096 Type: Directory
Attrs: readable writable executable hidden nativepath
Mode: drwxr-xr-x
Owner: tjmaciei (1000) Group: users (100)
Access: 2017-07-02T14:47:49.608
Birth: 2016-05-02T13:20:33.097
Change: 2017-07-01T13:37:08.737
Modified: 2017-07-01T13:37:08.737
It's not supported in any other filesystems I have (Linux sources show
xfs has the feature too). Even on ext4, it depends on whether the
filesystem was created with 256-byte inodes, which my /boot fs wasn't.
Change-Id: I8d96dea9955d4c749b99fffd14cda23ed60d5e72
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
It was working on Linux because _GNU_SOURCE gets us POSIX.1-2008
compatibility, but not on macOS or the BSDs. There, we were still stuck
to full second precision.
This commit uses the template trick introduced by the futimes code
(which itself was inspired by commit 2fb42eb4af3444b11e7b1210323637937ef
in QtNetwork). Also note how it adds support for birth time, if the
system's stat struct has that information.
Tested to work on MacOS and FreeBSD. The manual filetest produces:
Name: .
Path: . (/usr/home/tjmaciei/src/qt/qt5)
Size: 1536 Type: Directory
Attrs: readable writable executable hidden nativepath
Mode: drwxr-xr-x
Owner: tjmaciei (1001) Group: tjmaciei (1001)
Access: 2017-07-13T20:03:47.916
Birth: 2017-07-13T20:03:47.916
Change: 2017-07-13T20:04:41.648
Modified: 2017-07-13T20:04:41.648
Linux will require support for statx(2).
Change-Id: I8d96dea9955d4c749b99fffd14cd97d7a8c6d45d
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Took me a long while to understand what it did. Now that I do, I can
also answer the question left behind during the original implementation
in Qt 4.8 (commit 4fd2aced96d9095254d89f9da9c911bd88f15245 in the old
history): how to know if a file exists?
Change-Id: I8d96dea9955d4c749b99fffd14cdae135499a0d3
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
The use as in the code:
futimesat(fd, NULL, &tv)
is not documented to work. The file descriptor should be a directory's
one, not an open file (though the Linux source code seems to handle that
case). This call was done as a fallback to futimes, so it's very
unlikely a system would have futimesat and not futimes.
Both the Linux and the FreeBSD man pages say it's deprecated anyway.
Change-Id: I8d96dea9955d4c749b99fffd14cd94068dc7668a
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
The Unix stat fields "st_ctime" and "st_ctim" mean "change time", the
last time that the file/inode status fields were changed. It does not
mean "creation time". So this commit splits all of the internal API to
"birth" and "metadata change" instead of "creation" to avoid the
conflict.
Change-Id: I149e0540c00745fe8119fffd1463fe78b619649e
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
The renameat2(2) Linux system call, new in 3.16, allows for the atomic
renaming of a file if and only if it won't clobber an existing
file. None of the Linux libcs have enabled this syscall as an API, so we
use syscall(3) to place the call.
If your libc has SYS_renameat2 but your kernel doesn't support it, we'll
keep issuing the unknown syscall, every time. Users in that situation
should upgrade (3.16 is from 2014).
On Darwin, there's a similar renameatx_np (guessing "np" stands for
"non-portable"). I haven't found anything similar on the other BSDs.
Change-Id: I1eba2b016de74620bfc8fffd14ccb4e455a3ec9e
Reviewed-by: David Faure <david.faure@kdab.com>
The rename(2) system call overwrites, so instead of using it, we try to
use the link/unlink pair. This works for regular cases, but can fail if
trying to change case in case-insensitive filesystems, if we're
operating on a non-Unix filesystem (FAT) or, on Linux, if the file
doesn't belong to the calling user (BSDs permit this). For those cases,
we fall back to rename(2).
That means there's a race condition if a new file is created there. But
we at least reduce the likelihood of that happening for regular files.
Change-Id: I1eba2b016de74620bfc8fffd14ccb38fd929e5aa
Reviewed-by: David Faure <david.faure@kdab.com>
To do that, we needed to add virtual id() in QAbstractFileEngine and
override it in QFSFileEngine. It might be useful to return other types
of IDs for the other file engines, but this commit does not attempt that
just yet.
Change-Id: I1eba2b016de74620bfc8fffd14ccafe0762b3c38
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
This deduplicates the code between QFileSystemEngine and QLockFile.
Change-Id: I1eba2b016de74620bfc8fffd14cd005d5fd9beaa
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@qt.io>
QFileSystemEngine::id() will stat(2)/CreateFile in order to get the ID
of the file anyway, so we don't need to use QFile::exists() to check if
the destination exists. Instead, rely on id() returning a null value to
indicate error. On Windows, it's possible that the calls to either
GetFileInformationByHandle or GetFileInformationByHandleEx might fail,
but we ignore those.
Change-Id: I1eba2b016de74620bfc8fffd14ccaebcbed64419
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@qt.io>
This protects against the file having been renamed or deleted. We'll
still operate on the open file, regardless the name it may have on the
filesystem.
Change-Id: I1eba2b016de74620bfc8fffd14cca85cfd672e6d
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
It's not dangling only because of QFileSystemMetaData's construction:
the nativeFilePath() function returns a member variable. Since
QByteArray COWs, the pointer that we stored would not be freed.
But this was dangerous, since any change to the "entry" variable could
cause it to invalidate the member variable and the pointer to become
dangling. This line is only as old as this entire file is.
Change-Id: I8d96dea9955d4c749b99fffd14cda4d8e2cc5e5b
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
It is incorrect to collapse a "symlink/.." segment because the parent
directory of the symlink's target may not be the directory where the
symlink itself is located.
[ChangeLog][QtCore][QDir] Fixed a bug that caused QDir::mkpath() to
create the wrong directory if the requested path contained a symbolic
link and "../".
Change-Id: Iaddbecfbba5441c8b2e4fffd14a3e367730a1e24
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@qt.io>
Reviewed-by: David Faure <david.faure@kdab.com>
Recent Darwin system have a new system call that allows cloning the
contents of a file from another one if the underlying file system (for
example, APFS) supports it.
Change-Id: I90ec53b8abd2b1dc4000070f295e226d0fb4c672
Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
Automount filesystems like /home on many operating systems (QNX and
OpenIndiana, at least) don't like if you try to mkdir in them, even if
the file path already exists. OpenIndiana even gives you an ENOSYS
error.
So instead, let's try to mkdir our target, if we fail because of ENOENT,
we try to create the parent, then try again.
Task-number: QTBUG-58390
Change-Id: Ibe5b1b60c6ea47e19612fffd149cce81589b0acd
Reviewed-by: James McDonnell <jmcdonnell@blackberry.com>
Reviewed-by: David Faure <david.faure@kdab.com>
Change-Id: I62448507f80daf6be72994ee99f0fb1aa107eb78
Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
All strings coming out of POSIX API calls are converted to composed form
by QFile::decodeName. Do the same for realpath(3) output. This is
especially important for HFS+, which will store file names in decomposed
form, and APIs will therefore return strings in decomposed form.
Task-number: QTBUG-55896
Change-Id: I5e51f4e5712ff26bf9644cbcf9a9603995748892
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
Slims down QCFString and leaves only one implementation of converting
back and forth between CF/NS strings and QStrings.
Change-Id: I068568ffa25e6f4f6d6c99dcf47078b7a8e70e10
Reviewed-by: Jake Petroules <jake.petroules@qt.io>
Replace substring functions that return QString with
corresponding functions that return QStringRef where
it's possible.
Create QString from QStringRef only where necessary.
Change-Id: Id9ea11b16947220cd27787c0b529de62d10b6c26
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Pass -xplatform macx-tvos-clang to configure to build.
Builds device and simulator by default.
Added ‘uikit’ platform with the common setup.
Also added QT_PLATFORM_UIKIT define (undocumented).
qmake config defines tvos (but not ios).
tvOS is 64bits only (QT_ARCH is arm64) and requires bitcode to be
embedded in the binary. A new ‘bitcode’ configuration was added.
For ReleaseDevice builds (which get archived and push to the store),
bitcode is actually embedded (-fembed-bitcode passed to clang). For all
other configurations, only using bitcode markers to keep file size
down (-fembed-bitcode-marker).
Build disables Widgets in qtbase, and qtscript (unsupported,
would require fixes to JavaScriptCore source code).
Qpa same as on iOS but disables device orientation, status bar, clipboard,
menus, dialogs which are not supported on tvOS.
Change-Id: I645804fd933be0befddeeb43095a74d2c178b2ba
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@theqtcompany.com>
From Qt 5.7 -> LGPL v2.1 isn't an option anymore, see
http://blog.qt.io/blog/2016/01/13/new-agreement-with-the-kde-free-qt-foundation/
Updated license headers to use new LGPL header instead of LGPL21 one
(in those files which will be under LGPL v3)
Change-Id: I046ec3e47b1876cd7b4b0353a576b352e3a946d9
Reviewed-by: Lars Knoll <lars.knoll@theqtcompany.com>
