| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
|
|
| |
udisks and storaged have merged again, development now continues at
https://github.com/storaged-project/udisks
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
In the common case when we got a DBusException we previously skipped the
time.sleep() and thus were just racing through the retry loop without actually
giving the child process any time to start up.
Also raise a proper exception on timeout instead of os.abort(). The latter did
not flush stderr fast enough to actually see the failure message, and as this
happens in the parent pid we actually can raise an exception meaningfully.
LP: #1585382
|
| |
|
|
|
|
|
| |
Simpler command to run integration tests, and drop the comment about "if you
are not davidz" as we don't hardcode user names any more.
|
| |
|
|
|
|
|
|
|
| |
GMainLoop already handles SIGTERM, but we have our own signal handler for
SIGINT, so use that.
https://bugs.freedesktop.org/show_bug.cgi?id=97349
|
|
|
|
| |
https://bugs.freedesktop.org/show_bug.cgi?id=96816
|
|
|
|
| |
https://bugs.freedesktop.org/show_bug.cgi?id=97592
|
|
|
|
|
|
|
|
|
|
| |
The parent mount point dir (/run/media/user) already controls access
permissions, so there is no need to set umask to 077 for exfat either. This
makes UDISKS_FILESYSTEM_SHARED="1" actually work for exfat.
Similar to commit ffcb11b1 for vfat and ntfs.
https://bugs.freedesktop.org/show_bug.cgi?id=97348
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When the udev device is not initialized it may not contain all of its
properties yet, which makes udisks provide an incorrect representation
of the to its clients. This is particularly problematic during the
system initialization phase, where udev is also in it coldplug phase.
udev will send an event when each device becomes initialized, so we can
safely skip it during coldplug and rely on hotplug even handling
instead.
https://bugs.freedesktop.org/show_bug.cgi?id=96832
|
|
|
|
| |
***Warning*** PKG_NAME is deprecated, you may remove it from autogen.sh
|
| |
|
|
|
|
|
|
|
|
| |
This is hopefully a better fix than 5c859c99df to
https://bugs.freedesktop.org/show_bug.cgi?id=85477
https://bugs.debian.org/767457
https://launchpad.net/bugs/1460602
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This reverts commit 5c859c99df.
The BLKRRPART ioctl is dangerous in that it causes pairs of remove/add
uevents for all partitions which in turn confuse our internal data
structures and cause the block devices to briefly disappear from the
filesystem.
Rereading the partition table is also unnecessary since parted,
libblockdev, and udevd do that already.
Thus, 5c859c99df might actually make things worse instead of fixing the
bug it is intended to fix.
|
| |
|
|
|
|
|
|
|
|
| |
Linux C libraries are moving away from implicit includes of
sys/sysmacros.h, so pull in this header directly in the files
that use these funcs.
https://bugs.freedesktop.org/show_bug.cgi?id=95028
|
|
|
|
| |
https://bugs.freedesktop.org/show_bug.cgi?id=95394
|
|
|
|
| |
https://bugs.freedesktop.org/show_bug.cgi?id=96522
|
|
|
|
|
|
| |
https://bugs.freedesktop.org/show_bug.cgi?id=85477
https://bugs.debian.org/767457
https://launchpad.net/bugs/1460602
|
| |
|
| |
|
| |
|
|
|
|
|
| |
Fix previous commit: We still must reset ptype to 0 for the force-removal test
to work, just do it *after* the CD drive was created successfully.
|
|
|
|
|
| |
Don't reset scsi_debug/ptype to 0 immediatley after the add_host request. This
seems to be racy and sometimes creates a normal hard disk drive.
|
|
|
|
|
|
| |
Fixes building with musl-libc.
https://bugs.freedesktop.org/show_bug.cgi?id=93111
|
|
|
|
| |
https://bugs.freedesktop.org/show_bug.cgi?id=92901
|
|
|
|
|
|
|
|
|
|
|
|
| |
Right before the "remove" event, the UUID of mdraid devices seems to
change sometimes.
Without catching this, we would create a new object for the new UUID
and erroneously dispatch the "remove" event to it. Consequently, the
original object would not receive this event and would continue
watching a non-existent device.
https://bugs.freedesktop.org/show_bug.cgi?id=91326
|
|
|
|
| |
https://bugs.freedesktop.org/show_bug.cgi?id=92488
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch provide option to build and run udisks without ACL.
v2: as replacement of ACL is used chown call
v3: do not change uid, change gid
v4: fix indentation & formating issues
v5: one missed empty line
v6: add g_set_error for chown
Thanks Peter Wu for helping me with improving code.
https://bugs.freedesktop.org/show_bug.cgi?id=54866
|
|
|
|
| |
https://bugs.freedesktop.org/show_bug.cgi?id=92486
|
|
|
|
|
| |
Always replace the current polkitd, and wait until the one on the bus is
actually our's, not the previously running one.
|
|
|
|
|
| |
Fix Polkit.test_removable_fs test case to run standalone, instead of assuming
that the previous test case already creates a file system.
|
| |
|
|
|
|
|
| |
Turn write_stderr() into a classmethod, so that we can call it from
classmethods.
|
|
|
|
|
|
|
| |
Fixes warning with current PyGI:
sys:1: PyGIWarning: UDisks was imported without specifying a version first.
Use gi.require_version('UDisks', '2.0') before import to ensure that the right version gets loaded.
|
|
|
|
|
|
|
|
|
|
| |
glib 2.45 now requires calling g_option_context_add_main_entries(), otherwise
it crashes on a NULL OptionContext->main_group.
Don't prematurely free the description as g_option_context_set_description()
does not copy the string.
https://launchpad.net/bugs/1478369
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
Formatting fails for a partition containing a partition table (e.g. Fedora
hybrid iso). Consequently the device is in wrong state, so check it before
modification and fail with a comprehensible error message.
https://bugs.freedesktop.org/show_bug.cgi?id=76178
|
|
|
|
|
|
| |
UDF stores permissions, so there is no reason to override it.
https://bugs.freedesktop.org/show_bug.cgi?id=90298
|
|
|
|
| |
https://bugs.freedesktop.org/show_bug.cgi?id=63893
|
|
|
|
|
|
|
|
|
| |
A security analysis revealed a possible issue in the UDisksSpawnedJob code.
Reportedly the NSS modules might not be available if the fork() happened
at an inconvenient time. This patch moves retrieval of UID information in
the parent process just before forking.
https://bugs.freedesktop.org/show_bug.cgi?id=73669
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In systemd user sessions, some of a user's processes can exist outside
the scope of any particular session:
└─user.slice
└─user-1000.slice
├─user@1000.service
│ ├─2089 /lib/systemd/systemd --user
│ └─dbus.service
│ ├─ 2233 /usr/bin/dbus-daemon …
│ ├─ 2297 /usr/lib/gvfs/gvfsd
│ …
└─session-2.scope
├─ 2102 gnome-session
├─ 2376 /usr/bin/gnome-shell
…
If processes outside sessions don't have access to devices on
those sessions' seats, then gvfsd won't be able to mount devices.
Conversely, there is no privilege boundary between the sessions and
the non-session processes - in particular, the user's processes can
usually ptrace each other and write to each other's configuration
files - so isolating them doesn't make a great deal of sense.
I'm specifically looking for one or more *active* sessions on the
device's seat because in a situation like this:
…
├─ alice
│ ├─ graphical session on seat0, tty7, active
│ └─ system --user
├─ bob
│ ├─ graphical session on seat0, tty8, inactive
│ ├─ graphical session on seat1, active
│ └─ systemd --user
└─ chris
├─ ssh session on no seat, active
└─ systemd --user
the desired behaviour is that alice controls seat0 devices and bob
controls seat1 devices, corresponding to their respective
physically-present locations; bob should not have control over
seat0 devices until he returns to seat0, and chris should not have
control over either.
Despite udisks_daemon_util_on_same_seat being documented,
it is not actually public API or ABI: it is part of libudisks-daemon.la,
which is a convenience library statically linked into udisksd and
the tests. As such, it's harmless to replace it with _on_user_seat.
I'm renaming it in order to force compilation failure if a branch
has other callers for the old semantics; if it was not renamed,
passing a pid_t where a uid_t was expected would have compiled,
but silently produced wrong results.
https://bugs.freedesktop.org/show_bug.cgi?id=89501
https://bugs.debian.org/780004
|
|
|
|
|
|
|
| |
Extend the hack to show the main partition of isohybrid iso9660
isos to work with udf isos too.
https://bugs.freedesktop.org/show_bug.cgi?id=89711
|
|
|
|
|
|
|
|
| |
$(prefix)/lib means the binary could wind up under e.g. /usr/lib rather
than /usr/lib64, or other more appropriate locations depending on the
distro.
https://bugs.freedesktop.org/show_bug.cgi?id=89736
|
|
|
|
|
|
|
| |
g_variant_new() doesn't like NULL pointers for empty arrays, so allocate an
empty array if we cannot read the attribute.
https://bugs.freedesktop.org/show_bug.cgi?id=91011
|
|
|
|
|
|
|
|
| |
sys.stderr is line-buffered by default which means that the status
updates (e.g. '[wait for data]') do not show up until after the test has
finished. Introduce a wrapper which writes to and flushes stderr.
https://bugs.freedesktop.org/show_bug.cgi?id=91043
|