BMC/Intel-BMC/linux.git/arch/um/drivers, branch dev-4.17

um: Fix raw interface options

2018-07-03T09:27:05+00:00

commit 5ec9121195a4f1cecd0fa592636c5f81eb03dc8c upstream. Raw interface initialization needs QDISC_BYPASS. Otherwise it sees its own packets when transmitting. Fixes: 49da7e64f33e ("High Performance UML Vector Network Driver") Cc: Signed-off-by: Anton Ivanov Signed-off-by: Richard Weinberger Signed-off-by: Greg Kroah-Hartman

um: Fix initialization of vector queues

2018-07-03T09:27:05+00:00

commit 4579a1ba692af81da7ea6ce197f8169ddc0c327f upstream. UML vector drivers could derefence uninitialized memory when cleaning up after a queue allocation failure. Fixes: 49da7e64f33e ("High Performance UML Vector Network Driver") Cc: Reported-by: Dan Capenter Signed-off-by: Anton Ivanov Signed-off-by: Richard Weinberger Signed-off-by: Greg Kroah-Hartman

net: in virtio_net_hdr only add VLAN_HLEN to csum_start if payload holds vlan

2018-06-25T23:51:20+00:00

[ Upstream commit fd3a88625844907151737fc3b4201676effa6d27 ] Tun, tap, virtio, packet and uml vector all use struct virtio_net_hdr to communicate packet metadata to userspace. For skbuffs with vlan, the first two return the packet as it may have existed on the wire, inserting the VLAN tag in the user buffer. Then virtio_net_hdr.csum_start needs to be adjusted by VLAN_HLEN bytes. Commit f09e2249c4f5 ("macvtap: restore vlan header on user read") added this feature to macvtap. Commit 3ce9b20f1971 ("macvtap: Fix csum_start when VLAN tags are present") then fixed up csum_start. Virtio, packet and uml do not insert the vlan header in the user buffer. When introducing virtio_net_hdr_from_skb to deduplicate filling in the virtio_net_hdr, the variant from macvtap which adds VLAN_HLEN was applied uniformly, breaking csum offset for packets with vlan on virtio and packet. Make insertion of VLAN_HLEN optional. Convert the callers to pass it when needed. Fixes: e858fae2b0b8f4 ("virtio_net: use common code for virtio_net_hdr and skb GSO conversion") Fixes: 1276f24eeef2 ("packet: use common code for virtio_net_hdr and skb GSO conversion") Signed-off-by: Willem de Bruijn Signed-off-by: David S. Miller Signed-off-by: Greg Kroah-Hartman

Fix vector raw inintialization logic

2018-03-29T20:18:02+00:00

Vector RAW in UML needs to BPF filter its own MAC only if QDISC_BYPASS has failed. If QDISC_BYPASS is successful, the frames originated locally are not visible to readers on the raw socket. Signed-off-by: Anton Ivanov Signed-off-by: Richard Weinberger

Migrate vector timers to new timer API

2018-03-29T20:16:34+00:00

The patches for the UML vector drivers were in-flight when the timer changes happened and were not covered by them. This change migrates vector_kern.c to use the new timer API. Signed-off-by: Anton Ivanov Signed-off-by: Richard Weinberger

um: vector: Fix an error handling path in 'vector_parse()'

2018-02-19T18:38:51+00:00

If 'find_device()' finds something, we set '*error_out' and we should return an error. However, 'err' is known to be 0 at this point. Explicitly return -EINVAL instead. While at it, remove the initialization of 'err' at the beginning of the function and also explicitly return an error code if the first check fails. Fixes: ad1f62ab2bd4 ("High Performance UML Vector Network Driver") Signed-off-by: Christophe JAILLET Signed-off-by: Richard Weinberger

um: vector: Fix a memory allocation check

2018-02-19T18:38:51+00:00

Checking the result of the previous 'kstrdup()' call is expected here, so we should test 'params' and not 'str' (which is known to be non-NULL at this point) Fixes: ad1f62ab2bd4 ("High Performance UML Vector Network Driver") Signed-off-by: Christophe JAILLET Signed-off-by: Richard Weinberger

um: vector: fix missing unlock on error in vector_net_open()

2018-02-19T18:38:51+00:00

Add the missing unlock before return from function vector_net_open() in the error handling case. Fixes: ad1f62ab2bd4 ("High Performance UML Vector Network Driver") Signed-off-by: Wei Yongjun Signed-off-by: Richard Weinberger

High Performance UML Vector Network Driver

2018-02-19T18:38:51+00:00

1. Provides infrastructure for vector IO using recvmmsg/sendmmsg. 1.1. Multi-message read. 1.2. Multi-message write. 1.3. Optimized queue support for multi-packet enqueue/dequeue. 1.4. BQL/DQL support. 2. Implements transports for several transports as well support for direct wiring of PWEs to NIC. Allows direct connection of VMs to host, other VMs and network devices with no switch in use. 2.1. Raw socket >4 times higher PPS and 10 times higher tcp RX than existing pcap based transport (> 4Gbit) 2.2. New tap transport using socket RX and tap xmit. Similar performance improvements (>4Gbit) 2.3. GRE transport - direct wiring to GRE PWE 2.4. L2TPv3 transport - direct wiring to L2TPv3 PWE 3. Tuning, performance and offload related setting support via ethtool. 4. Initial BPF support - used in tap/raw to avoid software looping 5. Scatter Gather support. 6. VNET and checksum offload support for raw socket transport. 7. TSO/GSO support where applicable or available 8. Migrates all error messages to netdevice_*() and rate limits them where needed. Signed-off-by: Anton Ivanov Signed-off-by: Richard Weinberger

Epoll based IRQ controller

2018-02-19T18:38:51+00:00

1. Removes the need to walk the IRQ/Device list to determine who triggered the IRQ. 2. Improves scalability (up to several times performance improvement for cases with 10s of devices). 3. Improves UML baseline IO performance for one disk + one NIC use case by up to 10%. 4. Introduces write poll triggered IRQs. 5. Prerequisite for introducing high performance mmesg family of functions in network IO. 6. Fixes RNG shutdown which was leaking a file descriptor Signed-off-by: Anton Ivanov Signed-off-by: Richard Weinberger