David Herrmann [Tue, 21 Jul 2015 10:59:56 +0000 (12:59 +0200)]
sd-bus: fix path of object-manager signals
Each signal of the ObjectManager interface carries the path of the object
in question as an argument. Therefore, a caller will deduce the object
this signal is generated for, by parsing the _argument_. A caller will
*not* use the object-path of the message itself (i.e., message->path).
This is done on purpose, so the caller can rely on message->path to be
the path of the actual object-manager that generated this signal, instead
of the path of the object that triggered this signal.
This commit fixes all InterfacesAdded/Removed signals to use the path of
the closest object-manager as message->path. 'closest' in this case means
closest parent with at least one object-manager registered.
This fix raises the question what happens if we stack object-managers in
a hierarchy. Two implementations are possible: First, we report each
object only on the nearest object-manager. Second, we report it on each
parent object-manager. This patch chooses the former. This is compatible
with other existing ObjectManager implementations, which are required to
call GetManagedObjects() recursively on each object they find, which
implements the ObjectManager interface.
Otherwise copying full directory trees between container and host won't
work, as we cannot access some fiels and cannot adjust the ownership
properly on the destination.
Of course, adding these many caps to the daemon kinda defeats the
purpose of the caps lock-down... but well...
Daniel Mack [Mon, 27 Jul 2015 13:41:53 +0000 (15:41 +0200)]
bus-proxy: augment debug message for dropped broadcasts a bit
Add the PID we are proxying for, as well as the message's sender and
destination string, to the debug message that is printed when the proxy
drops unmatched broadcasts.
This should clear up some confusion in
https://github.com/systemd/systemd/issues/717.
This basically copies the description from systemd.unit to this
man page. Masking can happen also in /run, so strike the part
about /etc, and also add the magic work "mask".
Justification is similar to BPDUGuard rename. "Positive" values
are easier. This is a rather uncommon option, so using a slightly
longer name should not be a problem, and may in fact may make it
easier to guess what the option does without reading the
documentation.
Looking at the kernel commit, "on" seems to be the default value:
commit 867a59436fc35593ae0e0efcd56cc6d2f8506586
Author: Vlad Yasevich <vyasevic@redhat.com>
Date: Wed Jun 5 10:08:01 2013 -0400
bridge: Add a flag to control unicast packet flood.
Add a flag to control flood of unicast traffic. By default, flood is
on and the bridge will flood unicast traffic if it doesn't know
the destination. When the flag is turned off, unicast traffic
without an FDB will not be forwarded to the specified port.
... and it seems to be the reasonable thing to do by default.
Rename to follow the follow the style of other options.
In general "positive" options are preferred to "negative" ones,
because they are easier to describe and easier for humans to
parse (c.f. the shortening on the man page entry).
Old name was slightly misleading, because this flag does not determine
whether DSCP is used overall, but only if it is copied to the
decapsulated packet. Rename to better reflect that.
"Copy" does not imply direction. This is on purpose, because we might
later on enhance the setting to allow/disallow copying in the other
direction, to the encapsulated packet. If that is implemented,
CopyDSCP could understand additional values. This is nicer than
having two separate settings and follows the example of DHCP=.
Also, we try to avoid abbreviations, but we allow acronyms
like MTU, in DiscoverPathMTU=.
This setting was recently added, so it's fine to rename it without
backwards compat.
Michael Olbrich [Fri, 24 Jul 2015 20:25:28 +0000 (22:25 +0200)]
automount: handle state changes of the corresponding mount unit correctly
The expire timeout must be started/stopped if the corresponding mount unit
changes its state, e.g. it is started via local-fs.target or stopped by a
manual umount.
Michael Olbrich [Fri, 24 Jul 2015 20:21:59 +0000 (22:21 +0200)]
automount: don't try to umount if it already happened
Return the token immediately instead. Otherwise the token is never returned
to the kernel, because the umount job is a noop and will not trigger a
state change.
Daniel Mack [Fri, 24 Jul 2015 11:49:11 +0000 (13:49 +0200)]
tree-wide: do not use _cleanup_free_ on const pointers
free() cannot be used with const pointers. However, our _cleanup_free_
handler features cast logic that hides that qualifier, so we don't get a
warning.
David Herrmann [Fri, 24 Jul 2015 10:37:12 +0000 (12:37 +0200)]
sd-bus: don't treat KDBUS_ITEM_TIMESTAMP as unknown item
In bus_kernel_translate_message(), we print a DEBUG message on unknown
items. But right now, we also print this message for KDBUS_ITEM_TIMESTAMP
despite parsing it properly. Fix this!
journal: uppercase first character in verify error messages
In the english language the first character of a sentence is supposed to
be uppercase. Let's make sure this also applies to the journal
verification error messages.
journal: avoid mapping empty data and field hash tables
When a new journal file is created we write the header first, then sync
and only then create the data and field hash tables in them. That means
to other processes it might appear that the files have a valid header
but not data and field hash tables. Our reader code should be able to
deal with this.
With this change we'll not map the two hash tables right-away after
opening a file for reading anymore (because that will of course fail if
the objects are missing), but delay this until the first time we access
them. On top of that, when we want to look something up in the hash
tables and we notice they aren't initialized yet, we consider them
empty.
This improves handling of some journal files reported in #487.
process: return ESRCH when a PID is not valid anymore
so far, when we read something from /proc/$PID we would pass on the
ENOENT from the kernel as error, if the process was missing. With this
change we systematically convert this to ESRCH, which is the more
appropriate error code, and what all the other glibc/syscalls like
kill() use.
All code that calls these functions should be fine with this change. In
fact, one invocation of get_process_exe() in bus-creds.c already assumed
ESRCH would be returned if a process is missing, and this assumption is
now validated after the change.
Christian Hesse [Thu, 23 Jul 2015 19:18:36 +0000 (21:18 +0200)]
virt: handle Virtualbox 5.0 with kvm hypervisor
Virtualbox 5.0 now supports kvm hypervisor. In this case cpuid
identidies as "kvm", which breaks units depending on
ConditionVirtualization=oracle.
So return "oracle" even with kvm hypervisor.
Tom Gundersen [Thu, 23 Jul 2015 11:13:43 +0000 (13:13 +0200)]
resolved: rr - don't read past end of RR when parsing NSEC(3)
We can never read past the end of the packet, so this seems impossible
to exploit, but let's error out early as reading past the end of the
current RR is clearly an error.
Tom Gundersen [Thu, 23 Jul 2015 10:57:58 +0000 (12:57 +0200)]
resolved: packet - fail on invalid zero-length data
Most blobs (keys, signatures, ...) should have a specific size given by
the relevant algorithm. However, as we don't use/verify the algorithms
yet, let's just ensure that we don't read out zero-length data in cases
where this does not make sense.
The only exceptions, where zero-length data is allowed are in the NSEC3
salt field, and the generic data (which we don't know anything about,
so better not make any assumptions).
projects / thirdparty / systemd.git / log
