17 Apr, 2011
6 commits
-
"next" in dn_rebuild_zone() is set but not actually used,
kill it off.Signed-off-by: David S. Miller
-
"iph" in pktgen_output_ipsec() is set but never actually
used. Kill it off.Signed-off-by: David S. Miller
-
This is just a readback to entire completion of a register
write, keep the readback but kill the unused variable.Signed-off-by: David S. Miller
-
Two cases here:
1) idt77252_rx_raw() really does not make any use of the
extracted PTI field of the atm header.2) idt77252_collect_stat() only uses the register values
in code which has been compiled out by a "NOTDEF" cpp
test for more than 10 years. Just kill this NOTDEF
code entirely, but keep the register reads in case
they have side effects.Signed-off-by: David S. Miller
-
GCC complains in these queue index operations because we
perform operations of the form:x = some_operation(++x);
which is undefined. Replace with:
x = some_operation(x + 1);
which is well defined and provides the intended operation.
Signed-off-by: David S. Miller
-
The variable eni_dev is initialized but never subsequently used in
these two functions.Signed-off-by: David S. Miller
16 Apr, 2011
8 commits
-
Note: I bet that gfar_set_features() don't really need a full reset.
Signed-off-by: Michał Mirosław
Signed-off-by: David S. Miller -
Signed-off-by: Michał Mirosław
Signed-off-by: David S. Miller -
This also fixes a race around np->txrxctl_bits while changing RXCSUM offload.
Signed-off-by: Michał Mirosław
Signed-off-by: David S. Miller -
Signed-off-by: Michał Mirosław
Signed-off-by: David S. Miller -
Signed-off-by: Michał Mirosław
Signed-off-by: David S. Miller -
Inline a small static function that's only ever called from one place.
Signed-off-by: Rob Landley
Reviewed-by: Jiri Pirko
Signed-off-by: David S. Miller -
[ipv6] Add support for RTA_PREFSRC
This patch allows a user to select the preferred source address
for a specific IPv6-Route. It can be set via a netlink message
setting RTA_PREFSRC to a valid IPv6 address which must be
up on the device the route will be bound to.Signed-off-by: Daniel Walter
Signed-off-by: David S. Miller -
[ipv6] Ignore looped-back NAs while in Duplicate Address Detection
If we send an unsolicited NA shortly after bringing up an
IPv6 address, the duplicate address detection algorithm
fails and the ip stays in tentative mode forever.
This is due a missing check if the NA is looped-back to us.Signed-off-by: Daniel Walter
Signed-off-by: David S. Miller
15 Apr, 2011
9 commits
-
My changes in commit 4d42d417be75d750b82798922b6e775915e11bce were
written some time before the introduction of FLAG_POINTTOPOINT, so
didn't include that flag in the new driver_info. Change the new
driver_info to be consistent.Signed-off-by: Ben Hutchings
Signed-off-by: David S. Miller -
This updates the bonding driver to support v2.6.27-rc3 enhancements
(b11f8d8c aka. "ethtool: Expand ethtool_cmd.speed to 32 bits") which
allow to encode the Mbps link speed on 32-bits (Max 4 Pbps) instead of
16 (Max 65536 Mbps).This patch also attempts to compact struct slave by reordering its
fields.Signed-off-by: David Decotigny
Signed-off-by: David S. Miller -
The __get_link_speed() function returns a u16 value which was stored
in a u32 local variable. This patch uses the return value directly,
thus fixing that minor type consistency.The 'duplex' field in struct slave being encoded on 8 bits, to be more
consistent we use a u8 integer (instead of u16) whenever we copy it to
local variables.Signed-off-by: David Decotigny
Signed-off-by: David S. Miller -
This gets rid of minor sparse complaints:
drivers/net/bonding/bond_main.c:4361:4: warning: do-while statement is not a compound statement
drivers/net/bonding/bond_main.c:243:12: warning: symbol 'bond_mode_name' was not declared. Should it be static?Signed-off-by: David Decotigny
Signed-off-by: David S. Miller -
The struct nic_operations is just function pointers and should be
declared const for added security.Signed-off-by: Stephen Hemminger
Signed-off-by: David S. Miller -
The phy, mac, and board information structures should be const.
Since tables contain function pointer this improves security
(at least theoretically).Compile tested only.
Signed-off-by: Stephen Hemminger
Acked-by: Ben Hutchings
Signed-off-by: David S. Miller -
This gets rid of the last spinlock in the Phonet stack proper.
Signed-off-by: Rémi Denis-Courmont
Signed-off-by: David S. Miller -
When physical identification of an adapter is done by toggling the
mechanism on and off through software utilizing the set_phys_id operation,
it is done with a fixed duration for both on and off states. Some drivers
may want to set a custom duration for the on/off intervals. This patch
changes the API so the return code from the driver's entry point when it
is called with ETHTOOL_ID_ACTIVE can specify the frequency at which to
cycle the on/off states, and updates the drivers that have already been
converted to use the new set_phys_id and use the synchronous method for
identifying an adapter.The physical identification frequency set in the updated drivers is based
on how it was done prior to the introduction of set_phys_id.Compile tested only. Also fixes a compiler warning in sfc.
v2: drivers do not return -EINVAL for ETHOOL_ID_ACTIVE
v3: fold patchset into single patch and cleanup per Ben's feedbackSigned-off-by: Bruce Allan
Cc: Ben Hutchings
Cc: Sathya Perla
Cc: Subbu Seetharaman
Cc: Ajit Khaparde
Cc: Michael Chan
Cc: Eilon Greenstein
Cc: Divy Le Ray
Cc: Don Fry
Cc: Jon Mason
Cc: Solarflare linux maintainers
Cc: Steve Hodgson
Cc: Stephen Hemminger
Cc: Matt Carlson
Acked-by: Jon Mason
Acked-by: Ben Hutchings
Signed-off-by: David S. Miller -
fib_select_default() is a complete NOP, and completely pointless
to invoke, when we have no more than 1 default route installed.And this is far and away the common case.
So remember how many prefixlen==0 routes we have in the routing
table, and elide the call when we have no more than one of those.This cuts output route creation time by 157 cycles on Niagara2+.
In order to add the new int to fib_table, we have to correct the type
of ->tb_data[] to unsigned long, otherwise the private area will be
unaligned on 64-bit systems.Signed-off-by: David S. Miller
Reviewed-by: Eric Dumazet
14 Apr, 2011
17 commits
-
Nobody alive seems to recall when they last were useful.
Signed-off-by: Ralf Baechle
Signed-off-by: David S. Miller -
DCB enabled X540 devices are not responding to pause frames
due to a missing register set that was added for these
devices that did not exist in other devices.Signed-off-by: John Fastabend
Tested-by: Ross Brattain
Signed-off-by: Jeff Kirsher -
The X540 device has a smaller packet buffer but the DCB configuration
never took this into account. Under stress this can result in the DMA
engine hanging and TX Unit hang occurring to reset the device. This
patch reworks the packet buffer allocation routine used for DCB on
82599 and X540 devices to account for RX packet buffer sizes.This fixes the immediate hang. We should consolidate the various
hardware specific routines for configuring features into a single
routine. This will make it much harder to miss feature cases like
this.Signed-off-by: John Fastabend
Tested-by: Ross Brattain
Tested-by: Evan Swanson
Signed-off-by: Jeff Kirsher -
x540 has the same device capability word in the EEPROM as 82599.
This patch renames ixgbe_get_device_caps_82599 to
ixgbe_get_device_caps_generic, moves it to ixgbe_common.h and
sets up the function pointer for x540.Signed-off-by: Emil Tantilov
Tested-by: Evan Swanson
Signed-off-by: Jeff Kirsher -
100H is not supported on this HW, but the bit is set on the PHY.
This can result in link at 100F when advertising only 1000F.Signed-off-by: Emil Tantilov
Tested-by: Evan Swanson
Signed-off-by: Jeff Kirsher -
Add new device ID supported by ixgbe.
Signed-off-by: Emil Tantilov
Tested-by: Stephen Ko
Signed-off-by: Jeff Kirsher -
HW can upload EEPROM content from flash while
in a middle of checksum calculation. Take NVM ownership for the whole
process of checksum update.Call ixgbe_read_eerd_generic() and ixgbe_write_eewr_generic() directly to
avoid double take of semaphores which leads to long loading times.Signed-off-by: Emil Tantilov
Tested-by: Stephen Ko
Signed-off-by: Jeff Kirsher -
Since msleep might not sleep for the desired amount when less
than 20ms use usleep_range.Signed-off-by: Don Skidmore
Tested-by: Stephen Ko
Signed-off-by: Jeff Kirsher -
Disable KR to KX4/KX downshift on 82599 backplane devices when
LESM (Link Establishment State Machine) is enabled in FW. Those
features cannot co-exist as they both manipulate the same registers.Signed-off-by: Emil Tantilov
Acked-by: Don Skidmore
Tested-by: Phillip Schmitt
Signed-off-by: Jeff Kirsher -
Relaxed ordering can lead to issues with some chipsets.
This patch makes sure that it is disabled by default and
not only when DCA is on.Signed-off-by: Emil Tantilov
Tested-by: Evan Swanson
Signed-off-by: Jeff Kirsher -
Factored out the common start_hw code into a new function
ixgbe_start_hw_gen2() so that it can be used by x540 and 82599.Signed-off-by: Emil Tantilov
Acked-by: Don Skidmore
Tested-by: Evan Swanson
Signed-off-by: Jeff Kirsher -
Signed-off-by: Jeff Kirsher
Acked-by: Don Skidmore
Tested-by: Stephen Ko -
Set ixgbe_identify_82599() as static
Signed-off-by: Emil Tantilov
Tested-by: Stephen Ko
Signed-off-by: Jeff Kirsher -
Some 82598 parts have LAN0 disabled and LAN1 enabled and the LAN ID bits in
Device Status register report the NIC as having only LAN1 as enabled. This
causes ixgbe_set_lan_id_multi_port_pcie() to set bus->func = 1 which is
incorrect.Force bus->func to 0 when LAN0 is disabled in the EEPROM.
Signed-off-by: Emil Tantilov
Tested-by: Evan Swanson
Signed-off-by: Jeff Kirsher -
The value of status was incorrectly tested. Also whitespace cleanup.
Signed-off-by: Emil Tantilov
Tested-by: Evan Swanson
Signed-off-by: Jeff Kirsher -
With the app data on the kernel dcb_app list we no longer
need to specifically handle them in ixgbe for the CEE case.
So now we can remove app handling logic and check when the
hw is configured if the app data matches the hardware
configuration in set_hw_all().If it does not match then we can reconfigure.
Signed-off-by: John Fastabend
Tested-by: Ross Brattain
Signed-off-by: Jeff Kirsher -
Based on a patch from Naga Chumbalkar :
If ASPM L0s needs to be disabled due to HW errata, do it prior to
"enabling" the device. This way if the kernel ever defaults its
aspm_policy to POLICY_POWERSAVE, then the e1000e driver will get a
chance to disable ASPM on the misbehaving device *prior* to calling
pci_enable_device_mem(). This will be useful in situations
where the BIOS indicates ASPM support on the server by clearing the
ACPI FADT "ASPM Controls" bit.Note:
The kernel (2.6.38) currently uses the BIOS "default" as its aspm_policy.
However, Linux distros can diverge from that and set the default to
"powersave".v2: o cleanup namespace pollution of e1000e_disable_aspm(),
o fix type and initialization of the new aspm_disable_flag in a few
functions, and
o redefine FLAG2_DISABLE_ASPM_L0S to the first unused bit in
adapter->flags2.Signed-off-by: Bruce Allan
Cc: Naga Chumbalkar
Tested-by: Aaron Brown
Signed-off-by: Jeff Kirsher
