A connect that fails is now echoed back to the guest with a valid RST frame
so that the client user sees "connection refused" not a long timeout and
failure.
Currently, if a ->tx or ->rx callback into the net_dev_operations
encounters an error, it returns -1 to the virtio-net code, which in turn
treats this as an unsigned (size_t) size describing the data available.
The resulting memcpy operation then quickly explodes with a SEGV.
This patch detects the error code from the low-level callbacks and
exits the thread dealing with the erroneous queue.
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Pekka Enberg <penberg@kernel.org>
When fa7226f (kvm tools: init network devices only when the virtio
driver is ready to go) was introduced, a tiny detail was overlooked:
- Initialization of the uip layer is now coming in very late (only
when the guest driver says it is ready).
- In parallel, the rx thread is created quite early (as soon as the
queues are allocated).
This cause the rx thread to call uip_rx, which calls uip_buf_get_used,
which starts to use buf_lock mutex/the buf_used_cond, which haven't
been initialized yet. Tears and devastation follow, not to mention a
certain lack of network connectivity for the unsuspecting guest.
The (not so pretty) fix is to split uip_init:
- uip_static_init: initialize the lists, mutexes and conditions,
called from virtio_net__init_one.
- uip_init: perform the dynamic memory allocations, called from
notify_status.
This allows the network to be safely initialized.
Cc: Sasha Levin <sasha.levin@oracle.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Pekka Enberg <penberg@kernel.org>
Support mergable rx buffers for virtio-net. This helps reduce the amount
of memory the guest kernel has to allocate per rx vq.
Signed-off-by: Sasha Levin <sasha.levin@oracle.com>
Signed-off-by: Pekka Enberg <penberg@kernel.org>
We want to make the size of the virtio net header opaque to
uip.
Signed-off-by: Sasha Levin <sasha.levin@oracle.com>
Signed-off-by: Pekka Enberg <penberg@kernel.org>
We already have something to wrap pthread with mutex_[init,lock,unlock]
calls. This patch creates a new struct mutex abstraction and moves
everything to work with it.
Signed-off-by: Sasha Levin <sasha.levin@oracle.com>
Signed-off-by: Pekka Enberg <penberg@kernel.org>
Give threads a meaningful name. This makes debugging much easier, and
everything else much prettier.
Suggested-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Sasha Levin <sasha.levin@oracle.com>
[ penberg@kernel.org: specify vcpu names ]
Signed-off-by: Pekka Enberg <penberg@kernel.org>
Respect guest tcp window size and stop sending tcp segments to guest
if guest's receive window is closed.
This fixes the TCP hang I'm seeing where guest and host are transferring
big chuck of data.
This problem was not triggered when guest and external host
communicates, probably because guest to external host communication
walks through real network and is much slower than guest and host
communication. Thus, guest's receive window has little chance to be
closed.
v2: use pthread_cond_wait to wait
Signed-off-by: Asias He <asias.hejun@gmail.com>
Signed-off-by: Pekka Enberg <penberg@kernel.org>
This pach makes tcp between 'guest ip' and 'host virtual ip' work in UIP mode.
(The defulat guest ip is 192.168.33.15, host virtual ip is 192.168.33.1)
guest$ wget http://192.168.33.1/file
guest$ ssh 192.168.33.1
Without this patch, user has to figure out the ip address of host's
interface (e.g. eth0) and use that ip address to access host.
With this patch, user can simply use the virtual host ip.
Signed-off-by: Asias He <asias.hejun@gmail.com>
Signed-off-by: Pekka Enberg <penberg@kernel.org>
In uip_dhcp_get_dns() we try to open /etc/resolv.conf. If we fail to
open it we then SEGV trying to fclose() it.
Fix the code to just return directly if we can't open it.
Acked-by: Asias He <asias.hejun@gmail.com>
Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Pekka Enberg <penberg@kernel.org>
This patch removes the manual/usermode dhcp client configuration and instead
uses the DHCP client built within the kernel.
Since this client is tightly integrated with NFS (if NFS config is set), we
will add a specific NFS root addr in our DHCP offer to point it to a non
existent address so that we won't hang trying to poke it for our root.
Acked-by: Asias He <asias.hejun@gmail.com>
Signed-off-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: Pekka Enberg <penberg@kernel.org>
DHCP packages are handled by the hypervisor, they are not sent to remote
servers. In contrast, all other non-DHCP UDP packages are sent to remote
servers.
Signed-off-by: Asias He <asias.hejun@gmail.com>
Signed-off-by: Pekka Enberg <penberg@kernel.org>
Use uip_dhcp_get_dns() in uip_init() to get DNS information from host
for DHCP.
Signed-off-by: Asias He <asias.hejun@gmail.com>
Signed-off-by: Pekka Enberg <penberg@kernel.org>
uip_tx_do_ipv4_udp_dhcp() is used to handle DHCP packages from guest.
Signed-off-by: Asias He <asias.hejun@gmail.com>
Signed-off-by: Pekka Enberg <penberg@kernel.org>
This patch get domain name and nameserver information from host config
file /etc/resolv.conf.
Guest can obtain DNS information through DHCP.
Signed-off-by: Asias He <asias.hejun@gmail.com>
[ penberg@kernel.org: check fscanf() for errors ]
Signed-off-by: Pekka Enberg <penberg@kernel.org>
If DHCP DISCOVER or DHCP REQUEST is found, reply with DHCP OFFER or DHCP
ACK respectively.
Signed-off-by: Asias He <asias.hejun@gmail.com>
Signed-off-by: Pekka Enberg <penberg@kernel.org>
This patch checks:
- sport and dport
- magic cookie
to detemine whether a UDP package is a DHCP package.
Signed-off-by: Asias He <asias.hejun@gmail.com>
Signed-off-by: Pekka Enberg <penberg@kernel.org>
This helper cooks a ethernet package and virtio header for UDP.
This patch also makes uip_udp_socket_thread() shorter.
Signed-off-by: Asias He <asias.hejun@gmail.com>
Signed-off-by: Pekka Enberg <penberg@kernel.org>
Alan