Age | Commit message (Collapse) | Author |
|
Create and send a DHCP Request message reusing already existing parts
of the code. This causes factoring out IP and UDP header creation and
moving next timeout calculation to be done every time in the timer
callback function independent of DHCP state. Also add an exponential
part to the timer calculation, bail out if there are errors while
resending the DHCP message for the sixth or more times.
|
|
Create a function for handling the full IP, UDP and DHCP packet
and tie it to the main loop. Verify IP and UDP headers and checksum.
Creat a new lease structure with using the values supplied in the
DHCP message. Free the lease structure when client is stopped.
Split out socket handling into a creation and a sending part. As a
result modify the test code.
|
|
Require a main loop to be set when creating a DHCP client. Set up
a timer to resend DHCP Discover messages and add a 0-2 second
delay to the timeout value. Move to state Selecting after successful
sending of a Discover message.
|
|
The client is stopped and brought back to its initial state.
|
|
It was noticed by Grant Erickson in ConnMan commit
95e15c09350acf58d4707056ae2614570883ef66 that:
"Certain DHCP servers, such as that implemented in Mac OS X
(< 10.7) for its "Internet Sharing" feature, refuse to issue
a DHCP lease to clients that have not set a non-zero value
in their DISCOVER or REQUEST packets."
|
|
Set a fake MAC address and emulate raw packet sending. When the buffer
containing the Discover message is received, check selected IP and
UDP headers and compute IP header and UDP message checksums. Also
send the DHCP message for option parsing and expect a successful
outcome.
|
|
On starting the client, use the supplied interface mac address and create
a transaction id. Puzzle together an IP/UDP/DHCP Discover message, compute
checksums and send it out as a raw packet.
Create an additional function that constructs default options common to
all DHCP messages.
Set the DHCP Client ID option as noticed by Grant Erickson in ConnMan
commit b18d9798b3a0ae46ed87d6d2be8d5a474bf3ab1e:
"Some Internet gateways and Wi-Fi access points are unhappy when the
DHCPv4 client-id option (61) is missing and will refuse to issue a
DHCP lease."
|
|
Open a packet socket, create a link level header, send packet and
close socket. Adding it to a separate file makes testing of the
DHCP sending much easier, as the test program can supply any socket
to the DHCP client code.
|
|
|
|
|
|
Add checks for invalid lengths and parameters when using the option
appending function. Add also checks for adding options, see to it
that the resulting array is identical to the array of options added.
|
|
Add a structure describing the DHCP file, sname and trailing options
fields. Create a messge holding these fields and call the internal
option parsing function.
In the test callback function verify that only regular options are
passed and figure out which part of the DHCP message is the one that
is being processed. As the test program knows the full contents of
the test options in the test structure, skip all non-regular fields
and verify that the option provided to the callback indeed is the
one expected. Check also if non-regular option fields are to be
ignored in the end of the option field as the callback is not called
again and the final check when the whole message has been processed
needs to be successful.
Add a boolean flag for pretty-printing, anticipate there will be a
nice option to toggle it in the future.
|
|
Create an initial simple test program for these two cases.
|
|
Add functions to append and parse DHCP options. Not all options
are passed to the callback function, the ones not exposed are
pad, end, message type and overload. If indicated by the overload
option, file and sname fields will be examined for more options.
The option functions are internal to DHCP, add a new header files
for interal function prototypes.
|
|
The client test program is the only one to be built so far.
|
|
|
|
Provide functionality for initializing a DHCP client struct, setting
interface index, last used address and additional options to request.
On initialization the most useful options are added by default.
|
|
Create a new directory to host DHCP components.
|
|
|
|
|
|
Needed for socketpair, recv
|
|
flags conversion
|
|
that are closer to kdbus
This turns around DO_NOT_QUEUE into QUEUE which implies a more useful
default. (And negative options are awful anyway.)
|
|
|
|
|
|
If --link-journal=host or --link-journal=guest is used, this totally
cannot work and we exit with an error. If however --link-journal=auto
or --link-journal=no is used, just display a warning.
Having the same machine id can happen if booting from the same
filesystem as the host. Since other things mostly function correctly,
let's allow that.
https://bugs.freedesktop.org/show_bug.cgi?id=68369
|
|
% build/journalctl help
Assertion 'match_is_valid(data, size)' failed at ../src/journal/sd-journal.c:227, function sd_journal_add_match(). Ignoring.
Callers cannot be expect to check all arguments always.
|
|
journalctl help output might run off the screen, so be consistent
as other systemd tools do and pipe it into a pager.
|
|
|
|
containers)
|
|
This is a fairly useless thing to do but it makes the compilers
and analyzers shut up about the use of mktemp.
|
|
to queue/starter
|
|
|
|
|
|
assert_return()
These are not programming errors, so they shouldn't use assert_return()
|
|
kdbus fortunately exposes the container's busses in the host fs, hence
we can access it directly instead of doing the namespacing dance.
|
|
In addition to the same spot within every 1min, every 1s, every 250s
|
|
we close it
|
|
log message
|
|
show messages from host too
|
|
generating them fresh for each log entry
|
|
connection
|
|
This adds the new library call sd_journal_open_container() and a new
"-M" switch to journalctl. Particular care is taken that journalctl's
"-b" switch resolves to the current boot ID of the container, not the
host.
|
|
|
|
|
|
Adds a new call sd_event_set_watchdog() that can be used to hook up the
event loop with the watchdog supervision logic of systemd. If enabled
and $WATCHDOG_USEC is set the event loop will ping the invoking systemd
daemon right after coming back from epoll_wait() but not more often than
$WATCHDOG_USEC/4. The epoll_wait() will sleep no longer than
$WATCHDOG_USEC/4*3, to make sure the service manager is called in time.
This means that setting WatchdogSec= in a .service file and calling
sd_event_set_watchdog() in your daemon is enough to hook it up with the
watchdog logic.
|
|
event source
That way the even source callback is run with the zombie process still
around so that it can access /proc/$PID/ and similar, and so that it can
be sure that the PID has not been reused yet.
|
|
|
|
> [simon@troela server]$ /usr/lib/systemd/systemd-activate -l 9000 main.js
> Assertion 'fd == 3 + count' failed at src/activate/activate.c:115,
> function open_sockets(). Aborting.
> Aborted (core dumped)
> after a bit debuging i found the problem:
> slim appears to leak an fd into all of its children:
> stat /proc/14004/fd/3 (14004 is the pid a random process in my session)
> File: '/proc/14004/fd/3' -> '/var/log/slim.log'
systemd-activate should be robust against the shell (or anything else) leaking
descriptors. Now everything except stdin/stdout/stderr and received sockets
will be closed.
|
|
|