This fixes a number of warnings such as:
CC drivers/net/ethernet/ti/davinci_cpdma.o
drivers/net/ethernet/ti/davinci_cpdma.c:279:1: warning: data definition
has no type or storage class
drivers/net/ethernet/ti/davinci_cpdma.c:279:1: warning: type defaults to
‘int’ in declaration of ‘EXPORT_SYMBOL_GPL’
drivers/net/ethernet/ti/davinci_cpdma.c:279:1: warning: parameter names
(without types) in function declaration
Signed-off-by: Daniel Mack <zonque@gmail.com>
Cc: Vaibhav Hiremath <hvaibhav@ti.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Christian Riesch <christian.riesch@omicron.at>
Signed-off-by: David S. Miller <davem@davemloft.net>
The correct behavior is to program the interrupt coalescing regs
(RXICr/TXICr) in accordance with the Rx/Tx Q's "rx/txcoalescing"
flag. That is, if the coalescing flag is 0 for a given Rx/Tx queue
then the corresponding coalescing register should be cleared.
This behavior is correctly implemented for the single-queue mode
(SQ_SG_MODE), but not for the multi-queue mode (MQ_MG_MODE).
This fixes the later case.
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
FCoE target mode was experiencing issues due to the fact that we were
sending up data frames that were padded to 60 bytes after the DDP logic had
already stripped the frame down to 52 or 56 depending on the use of VLANs.
This was resulting in the FCoE DDP logic having issues since it thought the
frame still had data in it due to the padding.
To resolve this, adding code so that we do not pad FCoE frames prior to
handling them to the stack.
CC: <stable@vger.kernel.org>
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add a missing resource release in ring cleanup.
Not doing this leaves a range of QPs that are being reserved,
and no one can use them.
Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix a crash at the error flow of NOP command which caused the driver to try and use
a completion vector which wasn't allocated.
Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>
Set valid port parameters: MTU and flow control configuration when
configuring the port during HW device initialization,
prior to the net device open() being called.
Using invalid parameters (such as all zeros)
could lead to bad firmware behavior.
Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>
The following commit couldn't work if the RMCR is not set to 1.
"net: sh_eth: fix the rxdesc pointer when rx descriptor empty happens"
commit id 79fba9f517
If RMCR is not set, the controller will clear the EDRRR after it received
a frame. In this case, the driver doesn't need to fix the value of
cur_rx/dirty_rx. The driver only needs it when the controll detects
receive descriptors are empty.
Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Tested-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
The 8168evl (RTL_GIGA_MAC_VER_34) based Gigabyte GA-990FXA motherboards
are very prone to NETDEV watchdog problems without this change. See
https://bugzilla.kernel.org/show_bug.cgi?id=42899 for instance.
I don't know why it *works*. It's depressingly effective though.
For the record:
- the problem may go along IOMMU (AMD-Vi) errors but it really looks
like a red herring.
- the patch sets the RX_MULTI_EN bit. If the 8168c doc is any guide,
the chipset now fetches several Rx descriptors at a time.
- long ago the driver ignored the RX_MULTI_EN bit.
e542a2269f changed the RxConfig
settings. Whatever the problem it's now labeled a regression.
- Realtek's own driver can identify two different 8168evl devices
(CFG_METHOD_16 and CFG_METHOD_17) where the r8169 driver only
sees one. It sucks.
Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch fixes a memory leak that was introduced in the 3.4 kernel. The
leak occurred when FCoE was enabled and traffic was passed over the FCoE
rings reserved for FCoE. The memory leak was due to us not populating the
compound page information on the order 1 pages needed for FCoE.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Fix Kconfig file to make sure that PTP and IGB/IXGBE are both either
in-kernel or modules, not mixed. Having the build status mixed causes
compile errors.
Signed-off-by: Carolyn Wyborny <carolyn.wyborny@intel.com>
Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
i210/i211 device has only 16 RAR address filters like 82575, instead of
32 like i350. This patch removes the entries for i210/i211 in the
get_invariants function which was setting them for 32. This ensures that
they will get the default value which is the correct one.
Signed-off-by: Carolyn Wyborny <carolyn.wyborny@intel.com>
Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Increasing the hardware statistics structure to accomodate statistics for skyhawk.
Signed-off-by: Vasundhara Volam <vasundhara.volam@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Modify IOCTL error message to print subsystem also.
Signed-off-by: Vasundhara Volam <vasundhara.volam@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The maximum size of packet that can be handled by controller including ethernet
header is 65535. Reducing gso_max_size accordingly.
Signed-off-by: Sarveshwar Bandi <sarveshwar.bandi@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch fixes a potential hole when configuring the cycle counter used to
generate the nanosecond time clock. This clock is based off of the SYSTIME
registers along with the TIMINCA registers. The TIMINCA register determines
the increment to be added to the SYSTIME registers every DMA clock tick. This
register needs to be reconfigured whenever the link-speed changes. However,
the value calculated stays the same when link is down and when link is up.
Misconfiguration can occur if the link status changes due to a reset, which
causes the TIMINCA register to be reset. This reset puts the device in an
unstable state where the SYSTIME registers stop incrementing and the PTP
protocol does not function.
The solution is to double check the TIMINCA value and always reset the value
if the register is zero. This prevents a misconfiguration bug that halts the
PHC.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Acked-by: Don Skidmore <donald.c.skidmore@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When I2C is not responding it's usually due to a previous
unexpected reset during I2C operation. We release it by
powering down and up the SFP+ module.
Signed-off-by: Yaniv Rosner <yaniv.rosner@broadcom.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The adapter->npars[] array has QLCNIC_MAX_PCI_FUNC elements. We
allocate it that way a few lines earlier in the function. So this test
is off by one.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Anirban Chakraborty <anirban.chakraborty@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There is a off by one error in the minimal number of BD in
bnx2x_start_xmit() and bnx2x_tx_int() before stopping/resuming tx queue.
A full size GSO packet, with data included in skb->head really needs
(MAX_SKB_FRAGS + 4) BDs, because of bnx2x_tx_split()
This error triggers if BQL is disabled and heavy TCP transmit traffic
occurs.
bnx2x_tx_split() definitely can be called, remove a wrong comment.
Reported-by: Tomas Hruby <thruby@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Eilon Greenstein <eilong@broadcom.com>
Cc: Yaniv Rosner <yanivr@broadcom.com>
Cc: Merav Sicron <meravs@broadcom.com>
Cc: Tom Herbert <therbert@google.com>
Cc: Robert Evans <evansr@google.com>
Cc: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull networking fixes from David S. Miller:
This has the fix for the wireless issues I ran into the other week as
well as:
1) Fix CAN c_can driver transmit handling resulting in BUG check
triggers, from AnilKumar Ch.
2) Fix packet drop monitor sleeping in atomic context, from Eric
Dumazet.
3) Fix mv643xx_eth driver build regression, from Andrew Lunn.
4) Inetpeer freeing needs an RCU grace period in order to avoid races
during tree invalidation. From Eric Dumazet.
5) Fix endianness bugs in xt_HMARK netfilter module, from Hans
Schillstrom.
6) Add proper module refcounting to l2tp_eth to avoid crash on module
unload, from Eric Dumazet.
7) Fix truncation of neighbour entry dumps due to logic errors in
neigh_dump_info() and friends, from Eric Dumazet.
8) The conversion of fib6_age() to dst_neigh_lookup() accidently
reversed the logic of a flags test, fix from Thomas Graf.
9) Fix checksum configuration in newer sky2 chips, from Stephen
Hemminger.
10) Revert BQL support in NIU driver, doesn't work.
11) l2tp_ip_sendmsg() illegally uses a route without a proper reference.
From Eric Dumazet.
12) be2net driver references an SKB after it's potentially been freed,
also from Eric Dumazet.
13) Fix RCU stalls in dummy net driver init. Also from Eric Dumazet.
14) lpc_eth has several bugs in it's transmit engine leading to packet
leaks and improper queue wakes, from Eric Dumazet.
15) Apply short DMA workaround to more tg3 chips, from Matt Carlson.
16) Add tilegx network driver.
17) Bonding queue mapping for a packet can get corrupted, fix from Eric
Dumazet.
18) Fix bug in netpoll_send_udp() SKB management that can leave garbage
in the payload in certain situations. From Eric Dumazet.
19) bnx2x driver interprets chip RX checksum offload incorrectly in
encapsulation situations. Fix from Eric Dumazet.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (75 commits)
bnx2x: fix checksum validation
netpoll: fix netpoll_send_udp() bugs
bonding: Fix corrupted queue_mapping
bonding:record primary when modify it via sysfs
tilegx network driver: initial support
tg3: Apply short DMA frag workaround to 5906
net: stmmac: Fix clock en-/disable calls
lpc_eth: fix tx completion
lpc_eth: add missing ndo_change_mtu()
dummy: fix rcu_sched self-detected stalls
net: Reorder initialization in ip_route_output to fix gcc warning
virtio-net: fix a race on 32bit arches
r8169: avoid NAPI scheduling delay.
net: Make linux/tcp.h C++ friendly (trivial)
netdev: fix drivers/net/phy/ kernel-doc warnings
net/core: fix kernel-doc warnings
be2net: fix a race in be_xmit()
l2tp: fix a race in l2tp_ip_sendmsg()
mac80211: add back channel change flag
NFC: Fix possible NULL ptr deref when getting the name of a socket
...
bnx2x driver incorrectly sets ip_summed to CHECKSUM_UNNECESSARY on
encapsulated segments. TCP stack happily accepts frames with bad
checksums, if they are inside a GRE or IPIP encapsulation.
Our understanding is that if no IP or L4 csum validation was done by the
hardware, we should leave ip_summed as is (CHECKSUM_NONE), since
hardware doesn't provide CHECKSUM_COMPLETE support in its cqe.
Then, if IP/L4 checksumming was done by the hardware, set
CHECKSUM_UNNECESSARY if no error was flagged.
Patch based on findings and analysis from Robert Evans
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Eilon Greenstein <eilong@broadcom.com>
Cc: Yaniv Rosner <yanivr@broadcom.com>
Cc: Merav Sicron <meravs@broadcom.com>
Cc: Tom Herbert <therbert@google.com>
Cc: Robert Evans <evansr@google.com>
Cc: Willem de Bruijn <willemb@google.com>
Acked-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This change adds support for the tilegx network driver based on the
GXIO IORPC support in the tilegx software stack, using the on-chip
mPIPE packet processing engine.
Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5906 devices also need the short DMA fragment workaround. This patch
makes the necessary change.
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Tested-by: Christian Kujau <lists@nerdbynature.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
clk_{un}prepare is mandatory for platforms using common clock framework.
Since these drivers are used by SPEAr platform, which supports common
clock framework, add clk_{un}prepare() support for them. Otherwise
the clocks are not correctly en-/disabled and ethernet support doesn't
work.
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Viresh Kumar <viresh.linux@gmail.com>
Cc: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
__lpc_handle_xmit() has two bugs :
1) It can leak skbs in case TXSTATUS_ERROR is set
2) It can wake up txqueue while no slot was freed.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Roland Stigge <stigge@antcom.de>
Tested-by: Roland Stigge <stigge@antcom.de>
Cc: Kevin Wells <kevin.wells@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
lpc_eth does a copy of transmitted skbs to DMA area, without checking
skb lengths, so can trigger buffer overflows :
memcpy(pldat->tx_buff_v + txidx * ENET_MAXF_SIZE, skb->data, len);
One way to get bigger skbs is to allow MTU changes above the 1500 limit.
Calling eth_change_mtu() in ndo_change_mtu() makes sure this cannot
happen.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Roland Stigge <stigge@antcom.de>
Cc: Kevin Wells <kevin.wells@nxp.com>
Acked-by: Roland Stigge <stigge@antcom.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
While reworking the r8169 driver a few months ago to perform the
smallest amount of work in the irq handler, I took care of avoiding
any irq mask register operation in the slow work dedicated user
context thread. The slow work thread scheduled an extra round of NAPI
work which would ultimately set the irq mask register as required,
thus keeping such irq mask operations in the NAPI handler.
It would eventually race with the irq handler and delay NAPI execution
for - assuming no further irq - a whole ksoftirqd period. Mildly a
problem for rare link changes or corner case PCI events.
The race was always lost after the last bh disabling lock had been
removed from the work thread and people started wondering where those
pesky "NOHZ: local_softirq_pending 08" messages came from.
Actually the irq mask register _can_ be set up directly in the slow
work thread.
Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Reported-by: Dave Jones <davej@redhat.com>
Tested-by: Marc Dionne <marc.c.dionne@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
As soon as hardware is notified of a transmit, we no longer can assume
skb can be dereferenced, as TX completion might have freed the packet.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This reverts commit efa230f2c6.
BQL doesn't work with how this driver currently only takes TX
interrupts every 1/4 of the TX ring. That behavior needs to be fixed,
but that's a larger non-trivial task and for now we have to revert
BQL support as this makes the device currently completely unusable.
Signed-off-by: David S. Miller <davem@davemloft.net>
The commit ba27ec66ff fixes the Kconfig of the
driver when built as module allowing to select/unselect
the PCI and Platform modules that are not anymore mutually
exclusive. This patch fixes and guarantees that the driver
builds on all the platforms w/ w/o PCI and when select/unselect
the two stmmac supports. In case of there are some problems
on both the configuration and the pci/pltf registration the
module_init will fail.
v2: set the CONFIG_STMMAC_PLATFORM enabled by default.
I've just noticed that this can actually help on
some configurations that don't enable any STMMAC
options by default (e.g. SPEAr).
v3: change printk level when do not register the driver.
Reported-by: Fengguang Wu <wfg@linux.intel.com>
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The newer flavors of Yukon II use a different method for receive
checksum offload. This is indicated in the driver by the SKY2_HW_NEW_LE
flag. On these newer chips, the BMU_ENA_RX_CHKSUM should not be set.
The driver would get incorrectly toggle the bit, enabling the old
checksum logic on these chips and cause a BUG_ON() assertion. If
receive checksum was toggled via ethtool.
Reported-by: Kirill Smelkov <kirr@mns.spb.ru>
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 452503ebc (ARM: Orion: Eth: Add clk/clkdev support.) broke
the building of the driver on architectures which don't have clk
support. In particular PPC32 Pegasos which uses this driver.
Add #ifdef around the clk API usage.
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 096335b3f9 ("mlx4_core: Allow dynamic MTU configuration for
IB ports") modifies the port VL setting. This exposes a bug in
mlx4_common_set_port(), where the VL cap value passed in (inside the
command mailbox) is incorrectly zeroed-out:
mlx4_SET_PORT modifies the VL_cap field (byte 3 of the mailbox).
Since the SET_PORT command is paravirtualized on the master as well as
on the slaves, mlx4_SET_PORT_wrapper() is invoked on the master. This
calls mlx4_common_set_port() where mailbox byte 3 gets overwritten by
code which should only set a single bit in that byte (for the reset
qkey counter flag) -- but instead overwrites the entire byte.
The result is that when running in SR-IOV mode, the VL_cap will be set
to zero -- fix this.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This patches fixes the driver when built as dynamic module.
In fact, the platform part cannot be built and the probe fails
(thanks to Bob Liu that reported this bug).
v2: as D. Miller suggested, it is not necessary to make the
pci and the platform code mutually exclusive.
Having both could also help, at built time ,to verify that
all the code is validated and compiles fine.
v3: removed wrong Reviewed-by from the patch
Reported-by: Bob Liu <lliubbo@gmail.com>
cc: Rayagond Kokatanur <rayagond@vayavyalabs.com>
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The hardware bit IXGBE_RXD_STAT_VP appears to be set even when Rx
stripping is disabled. This results in passing frames up the stack
which do not have the 802.1Q tag stripped but have the tci bits
set as if it was.
Working around this with a check for the feature flag bit. I
would welcome any better ideas or a pointer to exactly which
bits in the hardware register need to be cleared to get the
IXGBE_RXD_STAT_VP bit to be set per data sheet.
Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Acked-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Marcus Dennis <marcusx.e.dennis@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
DCB can be used independent of if RX VLAN stripping is enabled
or disabled so remove erroneous check.
Also enable or disable VLAN stripping when features are applied so
hardware and feature flags are in sync.
CC: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Tested-by: Marcus Dennis <marcusx.e.dennis@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
commit 44abd5c127 introduced NULL pointer
dereferences when attempting to access the check_reset_block function
pointer on 8257x and 80003es2lan non-copper devices.
This fix should be applied back through 3.4.
Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
skb_defer_rx_timestamp was called with a freshly allocated skb but must
be called with rskb instead.
Signed-off-by: Stephan Gatzka <stephan@gatzka.org>
Cc: stable <stable@vger.kernel.org>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The definition of I217_PROXY_CTRL must use the BM_PHY_REG() macro instead
of the PHY_REG() macro for PHY page 800 register 70 since it is for a PHY
register greater than the maximum allowed by the latter macro, and fix a
typo setting the I217_MEMPWR register in e1000_suspend_workarounds_ich8lan.
Also for clarity, rename a few defines as bit definitions instead of masks.
Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This is another fixup where the data is not transfered into buffer
addressed by skb->data but into a page.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
when register_netdev fails, the init'ed NAPIs by netif_napi_add must be
deleted with netif_napi_del, and also when driver unloads, it should
delete the NAPI before unregistering netdevice using unregister_netdev.
Signed-off-by: Devendra Naga <devendra.aaru@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, we terminate the eeprom access through clearing the CS by:
RTL_W8 (Cfg9346, ~EE_CS); or writeb (~EE_CS, ee_addr);
This would left the eeprom into "Config. Register Write Enable:"
state which is not expcted as the highest two bits were set to
0x11 ( expected is the "Normal" mode (0x00)). Solving this by write
0x0 instead of ~EE_CS when terminating the eeprom access.
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, we enable the receiver before setting the ring address which could
lead the card DMA into unexpected areas. Solving this by set the ring address
before enabling the receiver.
btw. I find and test this in qemu as I didn't have a 8139cp card in hand. please
review it carefully.
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The "!mlx4_is_slave" is totally confusing. Fix with
constant MLX4_CMD_NATIVE, which is the intended behavior.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The range check was performed after using the port number.
Reverse this to prevent a potential array overflow.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
- pass the following parameters:
- firmware version (added QUERY_FW paravirtualization for that)
- disable Blueflame on slaves. KVM disables write combining on guests,
and we get better performance without BF in this case. (This requires
QUERY_DEV_CAP paravirtualization, also in this commit)
- max qp rdma as destination
- get rid of a chunk of "if (0)" dead code
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Port is used as an array index before we know if that is proper.
For example, in the catas event case, port is zero; however,
the port index should lie in the range (1..2).
Fix this by using 'port' only in the events where it is of interest.
Test for port out of range in the default (unhandled event) case,
and do not output a message if it is not an ethernet port.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In SRIOV mode, the number of EQs used when computing the total ICM size
was incorrect.
To fix this, we do the following:
1. We add a new structure to mlx4_dev, mlx4_phys_caps, to contain physical HCA
capabilities. The PPF uses the phys capabilities when it computes things
like ICM size.
The dev_caps structure will then contain the paravirtualized values, making
bookkeeping much easier in SRIOV mode. We add a structure rather than a
single parameter because there will be other fields in the phys_caps.
The first field we add to the mlx4_phys_caps structure is num_phys_eqs.
2. In INIT_HCA, when running in SRIOV mode, the "log_num_eqs" parameter
passed to the FW is the number of EQs per VF/PF; each function (PF or VF)
has this number of EQs available.
However, the total number of EQs which must be allowed for in the ICM is
(1 << log_num_eqs) * (#VFs + #PFs). Rather than compute this quantity,
we allocate ICM space for 1024 EQs (which is the device maximum
number of EQs, and which is the value we place in the mlx4_phys_caps structure).
For INIT_HCA, however, we use the per-function number of EQs as described
above.
Signed-off-by: Marcel Apfelbaum <marcela@dev.mellanox.co.il>
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Mack