Himanshu Madhani says:
====================
qlcnic: Multiple Tx queue support and code refactoring
This Patch series contains following changes
o Refactored code to calculate, validate and assign Tx/SDS rings for various modes of driver.
o Enhanced ethtool statistics for multi Tx queue on all supported adapters.
o Enable multiple Tx queue for 83xx and 84xx Series adapters.
o Register netdev for failed device state.
changes from v1 -> v2
o Dropped patch to replace inappropriate usage of kzalloc() with vzalloc().
Please apply to net-next.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
o 83xx and 84xx firmware is capable of multiple Tx queues.
This patch will enable multiple Tx queues for 83xx/84xx
series adapters. Max number of Tx queues supported will be 8.
Signed-off-by: Himanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
o Current driver has duplicate code for validating user input
for changing Tx/SDS rings using set_channel ethtool interface.
This patch removes duplicate code and refactored Tx/SDS ring
validation for 82xx/83xx/84xx series adapter.
o Refactored code now calculates maximum Tx/Rx ring driver can
support based on Default, NPAR and SRIOV PF/VF mode of driver.
Signed-off-by: Himanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
o Enhance ethtool statistics to display multiple Tx queue stats for
all supported adapters.
Signed-off-by: Himanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
o Without failing probe, register netdev when device is in FAILED state.
o Device will come up with minimum functionality and allow diagnostics and
repair of the adapter.
Signed-off-by: Sucheta Chakraborty <sucheta.chakraborty@qlogic.com>
Signed-off-by: Himanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We can safely reduce the number of test cases by a tenth.
There is no particular need to run as many as we're running
now for crc32{,c}_combine, that gives us still ~8000 tests
we're doing if people run kernels with crc selftests enabled
which is perfectly fine.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fengguang reports that when crc32 selftests are running on startup, on
some e.g. 32bit systems, we can get a CPU stall like "INFO: rcu_sched
self-detected stall on CPU { 0} (t=2101 jiffies g=4294967081 c=4294967080
q=41)". As this is not intended, add a cond_resched() at the end of a
test case to fix it. Introduced by efba721f63 ("lib: crc32: add test cases
for crc32{, c}_combine routines").
Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch fixes a build warning in skb_checksum() by wrapping the
csum_partial() usage in skb_checksum(). The problem is that on a few
architectures, csum_partial is used with prefix asmlinkage whereas
on most architectures it's not. So fix this up generically as we did
with csum_block_add_ext() to match the signature. Introduced by
2817a336d4 ("net: skb_checksum: allow custom update/combine for
walking skb").
Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
drivers/net/ethernet/emulex/benet/be.h
drivers/net/netconsole.c
net/bridge/br_private.h
Three mostly trivial conflicts.
The net/bridge/br_private.h conflict was a function signature (argument
addition) change overlapping with the extern removals from Joe Perches.
In drivers/net/netconsole.c we had one change adjusting a printk message
whilst another changed "printk(KERN_INFO" into "pr_info(".
Lastly, the emulex change was a new inline function addition overlapping
with Joe Perches's extern removals.
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull networking fixes from David Miller:
"I'm sending a pull request of these lingering bug fixes for networking
before the normal merge window material because some of this stuff I'd
like to get to -stable ASAP"
1) cxgb3 stopped working on 32-bit machines, fix from Ben Hutchings.
2) Structures passed via netlink for netfilter logging are not fully
initialized. From Mathias Krause.
3) Properly unlink upper openvswitch device during notifications, from
Alexei Starovoitov.
4) Fix race conditions involving access to the IP compression scratch
buffer, from Michal Kubrecek.
5) We don't handle the expiration of MTU information contained in ipv6
routes sometimes, fix from Hannes Frederic Sowa.
6) With Fast Open we can miscompute the TCP SYN/ACK RTT, from Yuchung
Cheng.
7) Don't take TCP RTT sample when an ACK doesn't acknowledge new data,
also from Yuchung Cheng.
8) The decreased IPSEC garbage collection threshold causes problems for
some people, bump it back up. From Steffen Klassert.
9) Fix skb->truesize calculated by tcp_tso_segment(), from Eric
Dumazet.
10) flow_dissector doesn't validate packet lengths sufficiently, from
Jason Wang
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (41 commits)
net/mlx4_core: Fix call to __mlx4_unregister_mac
net: sctp: do not trigger BUG_ON in sctp_cmd_delete_tcb
net: flow_dissector: fail on evil iph->ihl
xfrm: Fix null pointer dereference when decoding sessions
can: kvaser_usb: fix usb endpoints detection
can: c_can: Fix RX message handling, handle lost message before EOB
doc:net: Fix typo in Documentation/networking
bgmac: don't update slot on skb alloc/dma mapping error
ibm emac: Fix locking for enable/disable eob irq
ibm emac: Don't call napi_complete if napi_reschedule failed
virtio-net: correctly handle cpu hotplug notifier during resuming
bridge: pass correct vlan id to multicast code
net: x25: Fix dead URLs in Kconfig
netfilter: xt_NFQUEUE: fix --queue-bypass regression
xen-netback: use jiffies_64 value to calculate credit timeout
cxgb3: Fix length calculation in write_ofld_wr() on 32-bit architectures
bnx2x: Disable VF access on PF removal
bnx2x: prevent FW assert on low mem during unload
tcp: gso: fix truesize tracking
xfrm: Increase the garbage collector threshold
...
In function mlx4_master_deactivate_admin_state() __mlx4_unregister_mac was
called using the MAC index. It should be called with the value of the MAC itself.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Marc Kleine-Budde says:
====================
I have two late fixes for the v3.12 release:
The first patch fixes a problem in the c_can's RX message handling, which can
lead to an endless interrupt loop under heavy load if messages are lost. The
second patch is by Olivier Sobrie and fixes the endpoint detection of the
kvaser_usb driver, which is needed for some devices.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Introduced in f9e42b8535 ("net: sctp: sideeffect: throw BUG if
primary_path is NULL"), we intended to find a buggy assoc that's
part of the assoc hash table with a primary_path that is NULL.
However, we better remove the BUG_ON for now and find a more
suitable place to assert for these things as Mark reports that
this also triggers the bug when duplication cookie processing
happens, and the assoc is not part of the hash table (so all
good in this case). Such a situation can for example easily be
reproduced by:
tc qdisc add dev eth0 root handle 1: prio bands 2 priomap 1 1 1 1 1 1
tc qdisc add dev eth0 parent 1:2 handle 20: netem loss 20%
tc filter add dev eth0 protocol ip parent 1: prio 2 u32 match ip \
protocol 132 0xff match u8 0x0b 0xff at 32 flowid 1:2
This drops 20% of COOKIE-ACK packets. After some follow-up
discussion with Vlad we came to the conclusion that for now we
should still better remove this BUG_ON() assertion, and come up
with two follow-ups later on, that is, i) find a more suitable
place for this assertion, and possibly ii) have a special
allocator/initializer for such kind of temporary assocs.
Reported-by: Mark Thomas <Mark.Thomas@metaswitch.com>
Signed-off-by: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
High-availability Seamless Redundancy ("HSR") provides instant failover
redundancy for Ethernet networks. It requires a special network topology where
all nodes are connected in a ring (each node having two physical network
interfaces). It is suited for applications that demand high availability and
very short reaction time.
HSR acts on the Ethernet layer, using a registered Ethernet protocol type to
send special HSR frames in both directions over the ring. The driver creates
virtual network interfaces that can be used just like any ordinary Linux
network interface, for IP/TCP/UDP traffic etc. All nodes in the network ring
must be HSR capable.
This code is a "best effort" to comply with the HSR standard as described in
IEC 62439-3:2010 (HSRv0).
Signed-off-by: Arvid Brodin <arvid.brodin@xdin.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Joby Poriyath provided a xen-netback patch to reduce the size of
xenvif structure as some netdev allocation could fail under
memory pressure/fragmentation.
This patch is handling the problem at the core level, allowing
any netdev structures to use vmalloc() if kmalloc() failed.
As vmalloc() adds overhead on a critical network path, add __GFP_REPEAT
to kzalloc() flags to do this fallback only when really needed.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Joby Poriyath <joby.poriyath@citrix.com>
Cc: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Borkmann says:
====================
SCTP fix/updates
Please see patch 5 for the main description/motivation, the rest just
brings in the needed functionality for that. Although this is actually
a fix, I've based it against net-next as some additional work for
fixing it was needed.
====================
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This fixes an outstanding bug found through IPVS, where SCTP packets
with skb->data_len > 0 (non-linearized) and empty frag_list, but data
accumulated in frags[] member, are forwarded with incorrect checksum
letting SCTP initial handshake fail on some systems. Linearizing each
SCTP skb in IPVS to prevent that would not be a good solution as
this leads to an additional and unnecessary performance penalty on
the load-balancer itself for no good reason (as we actually only want
to update the checksum, and can do that in a different/better way
presented here).
The actual problem is elsewhere, namely, that SCTP's checksumming
in sctp_compute_cksum() does not take frags[] into account like
skb_checksum() does. So while we are fixing this up, we better reuse
the existing code that we have anyway in __skb_checksum() and use it
for walking through the data doing checksumming. This will not only
fix this issue, but also consolidates some SCTP code with core
sk_buff code, bringing it closer together and removing respectively
avoiding reimplementation of skb_checksum() for no good reason.
As crc32c() can use hardware implementation within the crypto layer,
we leave that intact (it wraps around / falls back to e.g. slice-by-8
algorithm in __crc32c_le() otherwise); plus use the __crc32c_le_combine()
combinator for crc32c blocks.
Also, we remove all other SCTP checksumming code, so that we only
have to use sctp_compute_cksum() from now on; for doing that, we need
to transform SCTP checkumming in output path slightly, and can leave
the rest intact.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, skb_checksum walks over 1) linearized, 2) frags[], and
3) frag_list data and calculats the one's complement, a 32 bit
result suitable for feeding into itself or csum_tcpudp_magic(),
but unsuitable for SCTP as we're calculating CRC32c there.
Hence, in order to not re-implement the very same function in
SCTP (and maybe other protocols) over and over again, use an
update() + combine() callback internally to allow for walking
over the skb with different algorithms.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We already have 100 test cases for crcs itself, so split the test
buffer with a-prio known checksums, and test crc of two blocks
against crc of the whole block for the same results.
Output/result with CONFIG_CRC32_SELFTEST=y:
[ 2.687095] crc32: CRC_LE_BITS = 64, CRC_BE BITS = 64
[ 2.687097] crc32: self tests passed, processed 225944 bytes in 278177 nsec
[ 2.687383] crc32c: CRC_LE_BITS = 64
[ 2.687385] crc32c: self tests passed, processed 225944 bytes in 141708 nsec
[ 7.336771] crc32_combine: 113072 self tests passed
[ 12.050479] crc32c_combine: 113072 self tests passed
[ 17.633089] alg: No test for crc32 (crc32-pclmul)
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds a combinator to merge two or more crc32{,c}s
into a new one. This is useful for checksum computations of
fragmented skbs that use crc32/crc32c as checksums.
The arithmetics for combining both in the GF(2) was taken and
slightly modified from zlib. Only passing two crcs is insufficient
as two crcs and the length of the second piece is needed for
merging. The code is made generic, so that only polynomials
need to be passed for crc32_le resp. crc32c_le.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
This is nothing more but a whitepace cleanup, as 80 chars is not a
hard but soft limit, and otherwise makes the test cases array really
look ugly. So fix it up.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
With the intent to dump other accounting data later.
This patch is a cleanup.
Signed-off-by: Holger Eitzenberger <holger@eitzenberger.org>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Encapsulate counters for both directions into nf_conn_acct. During
that process also consistently name pointers to the extend 'acct',
not 'counters'. This patch is a cleanup.
Signed-off-by: Holger Eitzenberger <holger@eitzenberger.org>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Pull MIPS fixes from Ralf Baechle:
"Three fixes across arch/mips with the most complex one being the GIC
interrupt fix - at nine lines still not monster. I'm confident this
are the final MIPS patches even if there should go for an rc8"
* 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus:
MIPS: ralink: fix return value check in rt_timer_probe()
MIPS: malta: Fix GIC interrupt offsets
MIPS: Perf: Fix 74K cache map
Negative message lengths make no sense -- so don't do negative queue
lenghts or identifier counts. Prevent them from getting negative.
Also change the underlying data types to be unsigned to avoid hairy
surprises with sign extensions in cases where those variables get
evaluated in unsigned expressions with bigger data types, e.g size_t.
In case a user still wants to have "unlimited" sizes she could just use
INT_MAX instead.
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.14 (GNU/Linux)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=O8nP
-----END PGP SIGNATURE-----
Merge tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux
Pull ARM kallsyms fix from Rusty Russell:
"Last minute perf unbreakage for ARM modules; spent a day in
linux-next"
* tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux:
scripts/kallsyms: filter symbols not in kernel address space
A vmalloc fault needs to sync up PGD/PTE entry from init_mm to current
task's "active_mm". ARC vmalloc fault handler however was using mm.
A vmalloc fault for non user task context (actually pre-userland, from
init thread's open for /dev/console) caused the handler to deref NULL mm
(for mm->pgd)
The reasons it worked so far is amazing:
1. By default (!SMP), vmalloc fault handler uses a cached value of PGD.
In SMP that MMU register is repurposed hence need for mm pointer deref.
2. In pre-3.12 SMP kernel, the problem triggering vmalloc didn't exist in
pre-userland code path - it was introduced with commit 20bafb3d23
"n_tty: Move buffers into n_tty_data"
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Cc: Gilad Ben-Yossef <gilad@benyossef.com>
Cc: Noam Camus <noamc@ezchip.com>
Cc: stable@vger.kernel.org #3.10 and 3.11
Cc: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We don't validate iph->ihl which may lead a dead loop if we meet a IPIP
skb whose iph->ihl is zero. Fix this by failing immediately when iph->ihl
is evil (less than 5).
This issue were introduced by commit ec5efe7946
(rps: support IPIP encapsulation).
Cc: Eric Dumazet <edumazet@google.com>
Cc: Petr Matousek <pmatouse@redhat.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
net/xfrm/xfrm_policy.c
Minor merge conflict in xfrm_policy.c, consisting of overlapping
changes which were trivial to resolve.
Signed-off-by: David S. Miller <davem@davemloft.net>
There is an extra semi-colon so bond_get_size() doesn't return the
correct value.
Fixes: ec76aa4985 ('bonding: add Netlink support active_slave option')
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Veaceslav Falico <vfalico@redhat.com>
Reviewed-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>
Bjørn Mork says:
====================
cdc_ncm: many small and mostly trivial fixes
This series ended up longer than expected, and it is still not
complete. There is more to come when time allows...
Most changes are trivial. Notable non-trivial changes are
- removed filtering of identical speed notifications
- tx_max calulation is changed to count the pad byte if
necessary, and respect the device limit as an absolute
upper limit even if it is too low according to the spec
- remove the bug preventing SET_MAX_DATAGRAM_SIZE from having
any effect
- drop the pad-to-max if ZLPs are enabled
- the driver specific VERSION is dropped
- dev->hard_mtu is set to tx_max instead of max_datagram_size
causing usbnet to calculate the qlen based on the real max
size of tx skbs
This series has been tested, along with the previously posted
cdc_mbim series, on the NCM and MBIM devices I have:
- Ericsson F5521gw (NCM)
- Huawei E367 (MBIM)
- D-Link DWM-156 A7 (MBIM w/ too low dwNtb{In,Out}MaxSize bug)
- Sierra Wireless MC7710 (MBIM w/ ZLP and CDC Union bugs)
Apart from the D-Link modem dropping a lot less oversized
frames with the fix dedicated to it, there are no end user
noticable functional changes as a result of this series. But
all the non-trivial changes I listed above are of course
detectable by users looking at that specific area (except maybe
the removed speed notification, which requires a device sending
duplicates to be noticable - I don't have any such device).
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
There are MBIM devices out there reporting
dwNtbInMaxSize=2048 dwNtbOutMaxSize=2048
and since the spec require a datagram max size of at least
2048, this means that a full sized datagram will never fit.
Still, sending larger NTBs than the device supports is not
going to help. We do not have any other options than either
a) refusing to bindi, or
b) respect the insanely low value.
Alternative b will at least make these devices work, so go
for it.
Cc: Alexey Orishko <alexey.orishko@gmail.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
Make it a bit easier for users to figure out what goes
wrong when bind fails.
Cc: Alexey Orishko <alexey.orishko@gmail.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
Most setup errors are ignored to ensure maximum firmware
compatibilty. But GET_NTB_PARAMETERS and the functional
descriptors are required. Use proper error codes and
log level if these fail.
Cc: Alexey Orishko <alexey.orishko@gmail.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
Rewriting the "set max datagram" part of dc_ncm_setup to
separate the selection and validatation of the size from
the code which optionally informs the device of this
value. This ensures that we use the correct value
regardless of device support for the get and set commands.
Removing some of the many indent levels while doing this
to make the code more readable.
Cc: Alexey Orishko <alexey.orishko@gmail.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
Converting the constants used in these comparisons at build
time instead of converting the variables for every received
frame at run time.
Cc: Alexey Orishko <alexey.orishko@gmail.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
These signatures are well known bit patterns, mostly made up
of ascii characters. Mentally parsing works best if they
are printed in hex.
Cc: Alexey Orishko <alexey.orishko@gmail.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
Take advantage of standard device name prefixing and
netdevice msglvl control where possible.
Cc: Alexey Orishko <alexey.orishko@gmail.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix cut'n'paste typo. Log the bogus length and not the
irrelevant signature.
Cc: Alexey Orishko <alexey.orishko@gmail.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
