This patch adds scatter/gather support to the sunvnet driver.
Signed-off-by: David L Stevens <david.stevens@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds support for VIO v1.7 (extended descriptor format)
and v1.8 (receive-side checksumming) to the sunvnet driver.
Signed-off-by: David L Stevens <david.stevens@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch changes the name of vnet_port_alloc_tx_bufs to
vnet_port_alloc_tx_ring, since there are no buffer allocations after
transmit zero copy support was added. This patch also moves the ring
allocation to after VIO version negotiation to allow for
different-sized descriptors in later VIO versions.
Signed-off-by: David L Stevens <david.stevens@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In recent testing I had disabled CONFIG_IP_MULTIPLE_TABLES and as a result
when I ran "cat /proc/net/fib_trie" the main trie was displayed multiple
times. I found that the problem line of code was in the function
fib_trie_seq_next. Specifically the line below caused the indexes to go in
the opposite direction of our traversal:
h = tb->tb_id & (FIB_TABLE_HASHSZ - 1);
This issue was that the RT tables are defined such that RT_TABLE_LOCAL is ID
255, while it is located at TABLE_LOCAL_INDEX of 0, and RT_TABLE_MAIN is 254
with a TABLE_MAIN_INDEX of 1. This means that the above line will return 1
for the local table and 0 for main. The result is that fib_trie_seq_next
will return NULL at the end of the local table, fib_trie_seq_start will
return the start of the main table, and then fib_trie_seq_next will loop on
main forever as h will always return 0.
The fix for this is to reverse the ordering of the two tables. It has the
advantage of making it so that the tables now print in the same order
regardless of if multiple tables are enabled or not. In order to make the
definition consistent with the multiple tables case I simply masked the to
RT_TABLE_XXX values by (FIB_TABLE_HASHSZ - 1). This way the two table
layouts should always stay consistent.
Fixes: 93456b6 ("[IPV4]: Unify access to the routing tables")
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Amir Vadai says:
====================
ethtool, net/mlx4_en: RSS hash function selection
This patchset by Eyal adds support in set/get of RSS hash function. Current
supported functions are Toeplitz and XOR. The API is design to enable adding
new hash functions without breaking backward compatibility.
Userspace patch will be sent after API is available in kernel.
The patchset was applied and tested over commit cd4c910 ("netpoll: delete
defconfig references to obsolete NETPOLL_TRAP")
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
The ConnectX HW is capable of using one of the following hash functions:
Toeplitz and an XOR hash function. This patch extends the implementation
of the mlx4_en driver set/get_rxfh callbacks to support getting and
setting the RSS hash function used by the device.
Signed-off-by: Eyal Perry <eyalpe@mellanox.com>
Signed-off-by: Amir Vadai <amirv@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch extends the set/get_rxfh ethtool-options for getting or
setting the RSS hash function.
It modifies drivers implementation of set/get_rxfh accordingly.
This change also delegates the responsibility of checking whether a
modification to a certain RX flow hash parameter is supported to the
driver implementation of set_rxfh.
User-kernel API is done through the new hfunc bitmask field in the
ethtool_rxfh struct. A bit set in the hfunc field is corresponding to an
index in the new string-set ETH_SS_RSS_HASH_FUNCS.
Got approval from most of the relevant driver maintainers that their
driver is using Toeplitz, and for the few that didn't answered, also
assumed it is Toeplitz.
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Cc: Ariel Elior <ariel.elior@qlogic.com>
Cc: Prashant Sreedharan <prashant@broadcom.com>
Cc: Michael Chan <mchan@broadcom.com>
Cc: Hariprasad S <hariprasad@chelsio.com>
Cc: Sathya Perla <sathya.perla@emulex.com>
Cc: Subbu Seetharaman <subbu.seetharaman@emulex.com>
Cc: Ajit Khaparde <ajit.khaparde@emulex.com>
Cc: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Cc: Jesse Brandeburg <jesse.brandeburg@intel.com>
Cc: Bruce Allan <bruce.w.allan@intel.com>
Cc: Carolyn Wyborny <carolyn.wyborny@intel.com>
Cc: Don Skidmore <donald.c.skidmore@intel.com>
Cc: Greg Rose <gregory.v.rose@intel.com>
Cc: Matthew Vick <matthew.vick@intel.com>
Cc: John Ronciak <john.ronciak@intel.com>
Cc: Mitch Williams <mitch.a.williams@intel.com>
Cc: Amir Vadai <amirv@mellanox.com>
Cc: Solarflare linux maintainers <linux-net-drivers@solarflare.com>
Cc: Shradha Shah <sshah@solarflare.com>
Cc: Shreyas Bhatewara <sbhatewara@vmware.com>
Cc: "VMware, Inc." <pv-drivers@vmware.com>
Cc: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Eyal Perry <eyalpe@mellanox.com>
Signed-off-by: Amir Vadai <amirv@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
mvneta_tx() dereferences skb to get skb->len too late,
as hardware might have completed the transmit and TX completion
could have freed the skb from another cpu.
Fixes: 71f6d1b31f ("net: mvneta: replace Tx timer with a real interrupt")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
since head->handle == handle (checked before), just assign handle.
Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
rcu variant is not correct here. The code is called by updater (rtnl
lock is held), not by reader (no rcu_read_lock is held).
Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>
rcu variant is not correct here. The code is called by updater (rtnl
lock is held), not by reader (no rcu_read_lock is held).
Signed-off-by: Jiri Pirko <jiri@resnulli.us>
ACKed-by: Jamal Hadi Salim <jhs@mojatatu.com>
Acked-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Acked-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Jeff Kirsher says:
====================
Intel Wired LAN Driver Updates 2014-12-06
This series contains updates to i40e and i40evf.
Shannon provides several patches to cleanup and fix i40e. First removes
an unneeded break statement in i40e_vsi_link_event(). Then removes
some debug messages that really do not give any useful information and
ends up getting printed every service_task loop, which fills the logfile
with noise when AQ tracing is enabled. Updates the aq_cmd arguments to
use %i which is much more forgiving and user friendly than the more
restrictive %x, or %d. Fixes the netdev_stat macro, where the old
xxx_NETDEV_STAT() macro was defined long before the newer
rtnl_link_stats64 came into being, and just never got updated.
Getting the pf_id from the function number had an issue when
when the PF was setup in passthru mode, the PCI bus/device/function
was virtualized and the number in the VM is different from the number in
the bare metal. This caused HW configuration issues when the wrong pf_id
was used to set up the HMC and other structures. The PF_FUNC_RID register
has the real bus/device/function information as configured by the BIOS,
so use that for a better number.
Carolyn adds additional text description for the base pf0 and flow
director generated interrupts, since these interrupts are difficult
to distinguish per port on a multi-function device.
Jacob resolves an issue related to images with multiple PFs per
physical port. We cannot fully support 1588 PTP features, since only
one port should control (i.e. write) the registers at a time. Doing
so can cause interference of functionality.
Anjali provides several updates to i40e, first adds the Virtual Channel
OP event opcode for CONFIG_RSS, so that the Virtual Channel state
machine can properly decipher status change events. Then updates the
driver to add (and use) i40e_is_vf macro for future expansion when new
VF MAC types get added. Adds new update VSI flow to accommodate a
firmware dix with VSI loopback mode. All VSIs on a VEB should either
have loopback enabled or disabled, a mixed mode is not supported for a
VEB. Since our driver supports multiple VSIs per PF that need to talk to
each other make sure to enable Loopback for the PF and FDIR VSI as well.
Mitch provides a couple of i40e and i40evf patches. First updates
i40evf init code more adept at handling when multiple VFs attempt
to initialize simultaneously.
Joe Perches provides a i40e patch which resolves a compile warning
about about frame size being larger than 2048 bytes by reducing the
stack use by using kmemdup and not using a very large struct on the
stack.
v2:
- Dropped patch 13 & 14 while Mitch reworks the patches based on
feedback from Ben Hutchings, probably the tryptophan in the turkey
is to blame for the delay...
- Added Joe Perches patch which resolves a compile warning about frame
size
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Alexander Duyck says:
====================
net: Add helper for padding short Ethernet frames
This patch series adds a pair of helpers to pad short Ethernet frames. The
general idea is to clean up a number of code paths that were all writing
their own versions of the same or similar function.
An added advantage is that this will help to discourage introducing new
bugs as in at least one case I found the skb->len had been updated, but the
tail pointer update was overlooked.
v2: Added skb_put_padto for cases where length is not ETH_ZLEN
Updated intel drivers and emulex driver to use skb_put_padto
Updated eth_skb_pad to use skb_put_padto
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Replace rtl_skb_pad with eth_skb_pad since they do the same thing.
Cc: Realtek linux nic maintainers <nic_swsd@realtek.com>
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Update myri10ge to use eth_skb_pad helper. This also corrects a minor
issue as the driver was updating length without updating the tail pointer.
Cc: Hyong-Youb Kim <hykim@myri.com>
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Replace the standard layout for padding an ethernet frame with the
eth_skb_pad call.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Update the Intel Ethernet drivers to use eth_skb_pad() and skb_put_padto
instead of doing their own implementations of the function.
Also this cleans up two other spots where skb_pad was called but the length
and tail pointers were being manipulated directly instead of just having
the padding length added via __skb_put.
Cc: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Acked-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds two new helper functions skb_put_padto and eth_skb_pad.
These functions deviate from the standard skb_pad or skb_padto in that they
will also update the length and tail pointers so that they reflect the
padding added to the frame.
The eth_skb_pad helper is meant to be used with Ethernet devices to update
either Rx or Tx frames so that they report the correct size. The
skb_put_padto helper is meant to be used primarily in the transmit path for
network devices that need frames to be padded up to some minimum size and
don't wish to simply update the length somewhere external to the frame.
The motivation behind this is that there are a number of implementations
throughout the network device drivers that are all doing the same thing,
but each a little bit differently and as a result several implementations
contain bugs such as updating the length without updating the tail offset
and other similar issues.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Eli Cohen says:
====================
mlx5 driver updates
The following series contains some fixes to mlx5 as well as update to the list
of supported devices.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
If mlx5_core_create_mkey fails, decrease the pending counter to undo the
previous increment.
Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
1. Add required __acquire/__release statements to balance spinlock usage.
2. Change the index parameter of begin_wqe() to be unsigned to match supplied
argument type.
Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add ConnectX-4LX to the list of supported devices as well as their virtual
functions.
Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The outbox should be cleared before executing the command.
Signed-off-by: Majd Dibbiny <majd@mellanox.com>
Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Command queue descriptor page size is 4KB and not the page size used by the
kernel.
Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
mlx5 requires at least one interrupt vector for completions so fix the minvec
argument to pci_enable_msix_range() accordingly.
Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Call request module on mlx5_ib so it will be available for applications
requiring it, such as installers that require boot over IB.
Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Chunhao Lin says:
====================
r8169:change hardware setting
This patch series contains two hardware setting modification to prevent
hardware become abnormal.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Cmac engine is the bridge between driver and dash firmware.
Other os may not disable cmac when leave. And r8169 did not allocate any
resources for cmac engine. Disable it to prevent abnormal system behavior.
Signed-off-by: Chunhao Lin <hau@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For RTL8168G/GU/H/EP and RTL8411B remove enable tx/rx from its own hw_start
function. This will prevent enable tx/rx before complete hardware tx/rx
setting.
Tx/Rx will be enabled in the end of function rtl_hw_start_8168.
Signed-off-by: Chunhao Lin <hau@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The mvneta driver sets the amount of Tx coalesce packets to 16 by
default. Normally that does not cause any trouble since the driver
uses a much larger Tx ring size (532 packets). But some sockets
might run with very small buffers, much smaller than the equivalent
of 16 packets. This is what ping is doing for example, by setting
SNDBUF to 324 bytes rounded up to 2kB by the kernel.
The problem is that there is no documented method to force a specific
packet to emit an interrupt (eg: the last of the ring) nor is it
possible to make the NIC emit an interrupt after a given delay.
In this case, it causes trouble, because when ping sends packets over
its raw socket, the few first packets leave the system, and the first
15 packets will be emitted without an IRQ being generated, so without
the skbs being freed. And since the socket's buffer is small, there's
no way to reach that amount of packets, and the ping ends up with
"send: no buffer available" after sending 6 packets. Running with 3
instances of ping in parallel is enough to hide the problem, because
with 6 packets per instance, that's 18 packets total, which is enough
to grant a Tx interrupt before all are sent.
The original driver in the LSP kernel worked around this design flaw
by using a software timer to clean up the Tx descriptors. This timer
was slow and caused terrible network performance on some Tx-bound
workloads (such as routing) but was enough to make tools like ping
work correctly.
Instead here, we simply set the packet counts before interrupt to 1.
This ensures that each packet sent will produce an interrupt. NAPI
takes care of coalescing interrupts since the interrupt is disabled
once generated.
No measurable performance impact nor CPU usage were observed on small
nor large packets, including when saturating the link on Tx, and this
fixes tools like ping which rely on too small a send buffer. If one
wants to increase this value for certain workloads where it is safe
to do so, "ethtool -C $dev tx-frames" will override this default
setting.
This fix needs to be applied to stable kernels starting with 3.10.
Tested-By: Maggie Mae Roxas <maggie.mae.roxas@gmail.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Ying Xue says:
====================
tipc: convert name table read-write lock to RCU
Now TIPC name table is statically allocated and is protected with a
Read-Write lock. To enhance the performance of TIPC name table lookup,
we are going to involve RCU lock to protect the name table. As a
consequence, it becomes lockless to concurrently look up name table on
read side. However, before the conversion can be successfully made,
the following two things must be first done:
- change allocation way of name table from static to dynamic
- fix several incorrect locking policy issues
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Convert tipc name table read-write lock to RCU. After this change,
a new spin lock is used to protect name table on write side while
RCU is applied on read side.
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Reviewed-by: Erik Hugne <erik.hugne@ericsson.com>
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Tested-by: Erik Hugne <erik.hugne@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When a list_head variable is seen as a new entry to be added to a
list head, it's unnecessary to be initialized with INIT_LIST_HEAD().
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Reviewed-by: Erik Hugne <erik.hugne@ericsson.com>
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Tested-by: Erik Hugne <erik.hugne@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When tipc name sequence is published, name table lock is released
before name sequence buffer is delivered to remote nodes through its
underlying unicast links. However, when name sequence is withdrawn,
the name table lock is held until the transmission of the removal
message of name sequence is finished. During the process, node lock
is nested in name table lock. To prevent node lock from being nested
in name table lock, while withdrawing name, we should adopt the same
locking policy of publishing name sequence: name table lock should
be released before message is sent.
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Reviewed-by: Erik Hugne <erik.hugne@ericsson.com>
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Tested-by: Erik Hugne <erik.hugne@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
As tipc_nametbl_lock is used to protect name_table structure, the lock
must be held while all members of name_table structure are accessed.
However, the lock is not obtained while a member of name_table
structure - local_publ_count is read in tipc_nametbl_publish(), as
a consequence, an inconsistent value of local_publ_count might be got.
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Reviewed-by: Erik Hugne <erik.hugne@ericsson.com>
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Tested-by: Erik Hugne <erik.hugne@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
TIPC internally created a name table which is used to store name
sequences. Now there is a read-write lock - tipc_nametbl_lock to
protect the table, and each name sequence saved in the table is
protected with its private lock. When a name sequence is inserted
or removed to or from the table, its members might need to change.
Therefore, in normal case, the two locks must be held while TIPC
operates the table. However, there are still several places where
we only hold tipc_nametbl_lock without proprerly obtaining name
sequence lock, which might cause the corruption of name sequence.
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Reviewed-by: Erik Hugne <erik.hugne@ericsson.com>
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Tested-by: Erik Hugne <erik.hugne@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
As TIPC subscriber server is terminated before name table, no user
depends on subscription list of name sequence when name table is
stopped. Therefore, all name sequences stored in name table should
be released whatever their subscriptions lists are empty or not,
otherwise, memory leak might happen.
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Reviewed-by: Erik Hugne <erik.hugne@ericsson.com>
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Tested-by: Erik Hugne <erik.hugne@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Name table locking policy is going to be adjusted from read-write
lock protection to RCU lock protection in the future commits. But
its essential precondition is to convert the allocation way of name
table from static to dynamic mode.
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Reviewed-by: Erik Hugne <erik.hugne@ericsson.com>
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Tested-by: Erik Hugne <erik.hugne@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The size variable is introduced in publ_list struct to help us exactly
calculate SKB buffer sizes needed by publications when all publications
in name table are delivered in bulk in named_distribute(). But if
publication SKB buffer size is assumed to MTU, the size variable in
publ_list struct can be completely eliminated at the cost of wasting
a bit memory space for last SKB.
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Tero Aho <tero.aho@coriant.com>
Reviewed-by: Erik Hugne <erik.hugne@ericsson.com>
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Tested-by: Erik Hugne <erik.hugne@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The compute_score functions are a bit difficult to read.
Neaten them a bit to reduce object sizes and make them a
bit more intelligible.
Return early to avoid indentation and avoid unnecessary
initializations.
(allyesconfig, but w/ -O2 and no profiling)
$ size net/ipv[46]/udp.o.*
text data bss dec hex filename
28680 1184 25 29889 74c1 net/ipv4/udp.o.new
28756 1184 25 29965 750d net/ipv4/udp.o.old
17600 1010 2 18612 48b4 net/ipv6/udp.o.new
17632 1010 2 18644 48d4 net/ipv6/udp.o.old
Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Modify bcmgenet driver so that it can be used on Broadcom 7xxx
MIPS-based STB platforms without a device tree.
Signed-off-by: Petri Gynther <pgynther@google.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
David L Stevens