This way, modules can decide to do different things based on having
all the ancillary data available. In particular, the UNIX module will
now post only one message header for all the FDs, even if they came
from multiple sets of ancillary data.
This should fix "Message needs unreceived descriptors" from the Chromium
IPC code (which is used by Firefox).
_BEOS_R5_COMPATIBLE_ was defined in ArchitectureRules while
__HAIKU_BEOS_COMPATIBLE is defined in HaikuConfig.h (which is
in the include path for sys/types, SupportDefs, and other
base headers.)
This paves the way for a variety of more interesting interactions
between drivers and the stack which are currently not possible
(e.g. checksum offload, #18744). The main advantage for the moment
is that we will save a memcpy of the buffer on each send/receive.
Adapt the virtio_net driver so that at least one driver is using
the new interface. Network still seems to work OK with it.
Change-Id: Ic5832e4865e3e1bed7462583ca1ffd16418d7cab
Reviewed-on: https://review.haiku-os.org/c/haiku/+/7912
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
Does not actually break ABI since net_buffer had 3 bytes of padding
remaining unused.
Add the first two constants for this field: L3/L4_CHECKSUM_VALID.
These will indicate when the checksum of the packet is known to be
valid (good/correct).
Change-Id: I155b3ea51d1093e229677cb788a007560ddbd428
Reviewed-on: https://review.haiku-os.org/c/haiku/+/7916
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
It contains only the MSG_* flags, not any other kind of flags.
Change-Id: Ia4590d87a1638fcdb848ef2b816b047b72ca2836
Reviewed-on: https://review.haiku-os.org/c/haiku/+/7915
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
Reviewed-by: Jérôme Duval <jerome.duval@gmail.com>
* Rename the "tun" network device to "tunnel". FreeBSD calls theirs
"tuntap" but speaks of both TUN and TAP devices as interfaces for
tunnels. The other BSDs seem to do likewise.
* Fold the "tun" driver into the "tunnel" network device. The
network device now publishes entries in devfs when interfaces
are created, and unpublishes them when interfaces are destroyed.
This removes the need for the driver and device to communicate
through a file descriptor, and thus allows the receive queue
to be totally eliminated, massively simplifying that logic.
* Use standard net-stack FIFOs instead of TCP BufferQueue, which is
specialized to TCP's needs in far too many ways. Thanks to the
previous commit adding support for interrupting semaphore waits,
we can use the FIFO wait mechanisms, too.
* Restructure the TAP logic, and generate MAC addresses more like
Linux does.
* Actually set type = IFT_TUN, and use the "loopback" frame handler
instead of the "ethernet" frame handler. This allows significant
cleanup of the header handling logic.
* In TUN mode, reject packets that don't look like IP packets.
* Delete "tunconfig"; it was mostly stubs and is now unnecessary.
TUN mode tested and confirmed as working by kallisti5 with OpenVPN.
TAP mode partially tested, but not yet confirmed as working.
Fixes #18673.
Change-Id: Ibd803139474e8db556a4f567901da15ee4083621
Reviewed-on: https://review.haiku-os.org/c/haiku/+/7143
Reviewed-by: Alex von Gluck IV <kallisti5@unixzen.com>
These have not been used in years (maybe since the redesign of the
network stack?). Only socket_writev had an implementation, which was
mostly just an incomplete version of socket_send.
Scatter/gather I/O is already supported via msghdr's msg_iov/len fields,
so this is redundant anyway.
Change-Id: If41c4f4ee021856f6db49c7cb95422a9c1aa7700
Reviewed-on: https://review.haiku-os.org/c/haiku/+/7127
Reviewed-by: waddlesplash <waddlesplash@gmail.com>
Tested-by: Commit checker robot <no-reply+buildbot@haiku-os.org>
Reviewed-by: Adrien Destugues <pulkomandy@pulkomandy.tk>
Reviewed-by: Jérôme Duval <jerome.duval@gmail.com>
Change-Id: Idf7ab6d2aa2b4f8cb4893053b531d2eae7418427
Reviewed-on: https://review.haiku-os.org/c/haiku/+/6608
Reviewed-by: Jérôme Duval <jerome.duval@gmail.com>
Reviewed-by: Alex von Gluck IV <kallisti5@unixzen.com>
The device is what actually controls the MTU, and it has its own
field for this, so having a second one just meant the MTU never
got updated after startup.
Remove the "mtu" field from the interface, use the "device->mtu" directly,
and then actually invoke device->module->set_mtu when updating.
These classes have been moved to the public API too soon, and they need some
more time to mature before they can be declared stable.
Change-Id: I9c52a8e6cc103922abde7a6b911fe0c3e6bf5700
Reviewed-on: https://review.haiku-os.org/c/haiku/+/3665
Reviewed-by: Adrien Destugues <pulkomandy@gmail.com>
- Messages that expect a reply are now tagged with a unique ID field to
indicate that expectation to the receiving socket messenger.
- The messenger now maintains a map of received reply IDs and their
corresponding messages, along with a message queue of other unsolicited
replies.
- After successfully connecting, the messenger now spawns a thread
whose sole responsibility is receiving and parsing all incoming messages,
and consequently sorting them into the aforementioned data structures based
on the presence of the reply ID. Callers who are awaiting either replies or
other messages are signalled appropriately via a semaphore. This allows
multiplexing of both types of messages on the same socket.
- Introduces new network API class BSocketMessenger, allowing one to send
and receive BMessages across a network socket in a BMessenger-like
fashion. Still very much WIP, hence currently not exposed via public headers.
Based partly on previous work by Axel.
The BNetworkRoute class manages a route_entry and the sockaddr's
associated with it. It replaces the direct use of route_entry in the
BNetworkInterface API.
Using route_entry is fragile and inconvenient as it only holds pointers
to the sockaddr's. When getting a list of routes from the kernel, each
route_entry is set up so that its pointers point into the single flat
buffer that is passed around. Creating a copy of the route_entry and
then deleting the flat buffer makes the pointers in the copy stale.
Returning these route entries therefore always lead to a use-after-free
when they were eventually used.
BNetworkRoute also takes over the code and functionallity of getting
routes from RouteSupport. The corresponding method in BNetworkRoster is
replaced by a static method in BNetworkRoute.
Also distinguish between the default route and gateway of an interface.
GetDefaultRoute() now gets the default BNetworkRoute for the interface
while GetDefaultGateway() gets the associated gateway address within
that default route. Adjust network preferences panel to this change.
Note that we currently only seem to have per interface default routes
and not an actual global default route. This was already the case before
these changes and I did not further investigate what this means.
Some websites set cookies expiring in the (not so) far future, after year 2038.
So, using time_t to store the cookie expiration date won't do. Use the
BDateTime class instead.
This makes goodsearch.com login work again (#10460).
* If we have a configured network, then we always try to connect to it
as soon as the interface has been brought up.
* If we don't have a configured network and are auto configuring, we
use the AutoconfigLooper to also do initial auto joins.
* Before issuing auto joins we need to wait for scan results to come
in, so we watch for corresponding messages.
For now auto joining is a one shot attempt as the infrastructure to
properly tell reasons for scans apart is not yet there.
* Various compilation fixes.
* Fixes to the FreeBSD compatibility layer (from comparing the x86-
specific bits with the equivalent amd64 sources in FreeBSD).
* Compile all the Ethernet drivers except for sis900 and wb840, these
require a bit more work to fix (will file a ticket soon). Tested
ipro1000 and rtl81xx, no issues.
BNetworkRoster::{Count|GetNext|Add|Remove}PersistentNetwork() as it fits
better (thanks Philippe for the heads up).
* Implement the backend for these functions in the net_server and also move
conversion of the wireless_network based format into the settings based format
there.
* Implement removal of a network from the settings and make adding a new network
with the same name replace the old one instead of just adding multiple ones.
Might need to change this in the future depending on how we want to handle
multiple networks with the same name (i.e. distinguish based on BSSID or
similar).
* Fix apparent oversight that caused configured networks _not_ to be used in the
auto join attempt.
* Remove auto joining open networks. I've been bitten by that more than once now
because we happen to have an open network in the neighbourhood that I now
accidentally used to transfer quite a bit of (unencrypted) stuff before
noticing... In the future, one will instead have to explicitly join an open
network once and store that config. Note that the driver will actually still
auto-associate with open networks due to how things are set up currently.
Note also that the auto join will fire join requests whenever there's a
disassociation event, so you might see spurious join dialogs when the
wpa_supplicant actually just re-establishes the connection.
* Make join requests async again. Instead of waiting for a synchronous reply of
the wpa_supplicant we instead return success when the request has been sent.
While the API call might still be made synchronous again in the future, the
net_server should really not block on an external application. In the case of
the wpa_supplicant we would otherwise deadlock when using the new
*PersistentNetwork() API after a successful join, and in other cases we might
just unacceptably delay other calls.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42816 a95241bf-73f2-0310-859d-f6bbb57e9c96
sdl_e_type in network byte order, causing such a socket not to receive
anything.
When working around that by not using htons() on bind, sending would then fail
on such a socket because the byte order is actually required to be swapped
there.
* Extend the comment for the B_NET_FRAME_TYPE macro to document that the input
types are supposed to be in host byte order to avoid future confusion.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42357 a95241bf-73f2-0310-859d-f6bbb57e9c96
