Mitch Williams [Sat, 16 Nov 2013 10:00:41 +0000 (10:00 +0000)]
i40e: properly add VF MAC addresses
The code that added new MAC addresses from the VFs did not, in
fact, work. Due to inverse logic, the only addresses that could
be added were addresses that already had been added.
Mitch Williams [Sat, 16 Nov 2013 10:00:38 +0000 (10:00 +0000)]
i40e: allow one more vector for VFs
The number of vectors per VF as reported by the hardware does not
include vector 0, so we need to account for this when checking to
see if the VF is configuring a valid vector.
net: clear local_df when passing skb between namespaces
We must clear local_df when passing the skb between namespaces as the
packet is not local to the new namespace any more and thus may not get
fragmented by local rules. Fred Templin noticed that other namespaces
do fragment IPv6 packets while forwarding. Instead they should have send
back a PTB.
The same problem should be present when forwarding DF-IPv4 packets
between namespaces.
tcp_memcontrol: Cleanup/fix cg_proto->memory_pressure handling.
kill memcg_tcp_enter_memory_pressure. The only function of
memcg_tcp_enter_memory_pressure was to reduce deal with the
unnecessary abstraction that was tcp_memcontrol. Now that struct
tcp_memcontrol is gone remove this unnecessary function, the
unnecessary function pointer, and modify sk_enter_memory_pressure to
set this field directly, just as sk_leave_memory_pressure cleas this
field directly.
This fixes a small bug I intruduced when killing struct tcp_memcontrol
that caused memcg_tcp_enter_memory_pressure to never be called and
thus failed to ever set cg_proto->memory_pressure.
Remove the cg_proto enter_memory_pressure function as it now serves
no useful purpose.
Don't test cg_proto->memory_presser in sk_leave_memory_pressure before
clearing it. The test was originally there to ensure that the pointer
was non-NULL. Now that cg_proto is not a pointer the pointer does not
matter.
Ivan Vecera [Wed, 4 Dec 2013 17:06:51 +0000 (18:06 +0100)]
forcedeth: run loopback test only on chipsets that support it
The driver incorrectly run loopback test on chips that don't support it.
Loopback test is only supported by chips that has DEV_HAS_TEST_EXTENDED
flag and returns 4 (NV_TEST_COUNT_EXTENDED) as test count.
dingtianhong [Wed, 4 Dec 2013 10:59:31 +0000 (18:59 +0800)]
bonding: add arp_ip_target checks when install the module
When I install the bonding with the wrong arp_ip_target,
just like arp_ip_target=500.500.500.500, the arp_ip_target
was transfored to 245.245.245.244 and stored in the ip
target success, it is uncorrect, so I add checks to avoid
adding wrong address.
The in4_pton() will set wrong ip address to 0.0.0.0 and
return 0, also use the micro IS_IP_TARGET_UNUSABLE_ADDRESS
to simplify the code.
Michal Kalderon [Wed, 4 Dec 2013 10:04:54 +0000 (12:04 +0200)]
bnx2x: avoid null pointer dereference when enabling SR-IOV
Fixed NULL pointer dereference when dynamically activating SR-IOV after vf
database failed to be allocated in probe stage (for example due to no ARI
support in pci hub).
wangweidong [Wed, 4 Dec 2013 09:32:39 +0000 (17:32 +0800)]
sctp: disable max_burst when the max_burst is 0
As Michael pointed out that when max_burst is 0, it just disable
max_burst. It declared in rfc6458#section-8.1.24. so add the check
in sctp_transport_burst_limited, when it 0, just do nothing.
Tony Lindgren [Tue, 3 Dec 2013 23:13:02 +0000 (15:13 -0800)]
net: davinci_emac: Fix platform data handling and make usable for am3517
When booted with device tree, we may still have platform data passed
as auxdata. For am3517 this is needed for passing the interrupt_enable
and interrupt_disable callbacks that access the omap system control module
registers. These callback functions will eventually go away when we have
a separate system control module driver.
Some of the things that are currently passed as platform data we don't need
to set up as device tree properties as they are always the same on am3517.
So let's use a new compatible flag for those so we can get those from
the device tree match data.
Also note that we need to fix setting of phy_dev to NULL instead of an empty
string as the code later on uses that to find the first phy on the mdio bus.
This seems to have been caused by 5d69e0076a72 (net: davinci_emac: switch to
new mdio).
Paul Durrant [Tue, 3 Dec 2013 17:39:29 +0000 (17:39 +0000)]
xen-netback: fix fragment detection in checksum setup
The code to detect fragments in checksum_setup() was missing for IPv4 and
too eager for IPv6. (It transpires that Windows seems to send IPv6 packets
with a fragment header even if they are not a fragment - i.e. offset is zero,
and M bit is not set).
This patch also incorporates a fix to callers of maybe_pull_tail() where
skb->network_header was being erroneously added to the length argument.
Kevin Wallace [Tue, 3 Dec 2013 10:55:22 +0000 (02:55 -0800)]
macvlan: Support creating macvtaps from macvlans
When running in a network namespace whose only link to the outside
world is a macvlan device, not being able to create a macvtap off of
it is a real pain.
So modify macvtap creation to automatically forward a creation of a
macvtap on a macvlan to become a creation of a macvtap on the
underlying network device, just like is currently done with
macvlan-on-macvlan devices.
v2: Use netif_is_macvlan and macvlan_dev_real_dev helpers to make it
more clear what we're doing.
David S. Miller [Fri, 6 Dec 2013 00:45:14 +0000 (19:45 -0500)]
Merge branch 'siocghwtstamp' of git://git.kernel.org/pub/scm/linux/kernel/git/bwh/sfc-next
Ben Hutchings says:
====================
SIOCGHWTSTAMP ioctl
1. Add the SIOCGHWTSTAMP ioctl and update the timestamping
documentation.
2. Implement SIOCGHWTSTAMP in most drivers that support SIOCSHWTSTAMP.
3. Add a test program to exercise SIOC{G,S}HWTSTAMP.
====================
David S. Miller [Fri, 6 Dec 2013 00:29:26 +0000 (19:29 -0500)]
Merge branch 'net_sched_actions'
Jamal Hadi Salim says:
====================
net_sched: actions - Add default lookup and walker
This set of patches provide defaults for lookup and walkers for actions.
Users can override when needed.
It also enforces mandatory action methods to be passed.
Thanks to David Miller who paid attention and made this patch set
much better.
====================
David S. Miller [Thu, 5 Dec 2013 21:02:56 +0000 (16:02 -0500)]
Merge branch 'for-davem' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless
John W. Linville says:
====================
Please pull this batch of fixes intende for the 3.13 stream!
For the mac80211 bits, Johannes says:
"For now I have various fixes all over, mostly for issues introduced in
relatively recent patches. There's no real pattern to it. Some of the
issues like go back longer, but still seemed 3.13 material."
And...
"These are just two patches disabling the broken CSA code. Once this
goes into your tree I'll merge it into mac80211-next and revert there
(since we fixed the bugs there)."
For the iwlwifi bits, Emmanuel says:
"I have here a few fixes for BT Coex. One of them is a NULL pointer
dereference. Another one avoids to enable a feature that can make the
firmware unhappy since the firmware isn't ready for it yet. WE also
avoid a WARNING that can be triggered upon association in not-so-bad
cases even if the association succeeded. We add support for new NICs
(not yet on the market) and bump the API so that 3.13 will be able to
work with the new firmware that will be out soon hopefully.
I also have a boundary check from Johannes."
In addition to those...
- Arend van Spriel fixes a brcmfmac problem that could use an
uninitialized variable in an error path.
- Borislav Petkov fixes a Kconfig-based build breakage problem for
brcmsmac.
- Michal Nazarewicz fixes a couple of NULL pointer dereference problems
in ath9k and wcn36xx.
- Sujith Manoharan fixes a couple of ath9k problems related to
incorrect interpretation of EEPROM configuration data.
- Ujjal Roy fixes a memory leak in mwifiex.
====================
As it stands dynamic user regulatory domain support is
only possible for a few programmed regulatory domains as
a few countries do not allow for this.
The existing code however only would take advantage of
the feature if a custom world regulatory domain is used
though as that's when we clear beconing flags. We need
to lift this restriction as otherwise this feature is
pointless.
freq_reg_info() expects KHz and not MHz, fix this. In
this case we'll now be getting the no-ir flags cleared
on channels for any channel when the country IE trusts
that channel.
ath: rename ath_reg_apply_active_scan_flags() to ath_reg_apply_ir_flags()
This also applies the no-ibss flag to the channels or clears it.
The idea here is to clarify no initiated radiation should be
allowed on these channels.
The routine ath_reg_apply_active_scan_flags() can be a bit
hard to read, this cleans it up by adding helpers for the
two cases of clearing IR flags or adding them. This approach
also makes no assumptions on the index of channels 12 and 13
so it should be portable accross different drivers.
ath: fix logic on ath_reg_apply_active_scan_flags()
The existing logic removes the passive scan flag from
channels 12 and 13 when a regulatory hint coming from
something other than a country IE has been passed. This
is incorrect, the original intention was to ensure we
always have passive scan enabled for these two channels
for a specific set of custom world regulatory domains.
Larry Finger [Mon, 25 Nov 2013 16:45:28 +0000 (10:45 -0600)]
rtlwifi: rtl8188ee: Fix typo in code
The static analyser "cppcheck" shows the following typo:
drivers/net/wireless/rtlwifi/rtl8188ee/dm.c:1081]: (style) Same expression on both sides of '!='.
Jingoo Han [Mon, 25 Nov 2013 03:37:24 +0000 (12:37 +0900)]
bcma: pci: remove unnecessary pci_set_drvdata()
The driver core clears the driver data to NULL after device_release
or on probe failure. Thus, it is not needed to manually clear the
device driver data to NULL.
The driver core clears the driver data to NULL after device_release
or on probe failure. Thus, it is not needed to manually clear the
device driver data to NULL.
The driver core clears the driver data to NULL after device_release
or on probe failure. Thus, it is not needed to manually clear the
device driver data to NULL.
The driver core clears the driver data to NULL after device_release
or on probe failure. Thus, it is not needed to manually clear the
device driver data to NULL.
The driver core clears the driver data to NULL after device_release
or on probe failure. Thus, it is not needed to manually clear the
device driver data to NULL.
Hauke Mehrtens [Sun, 24 Nov 2013 21:03:41 +0000 (22:03 +0100)]
brcmsmac: remove dependency on CRC8 and CRC_CCITT
There is no code in brcmsmac that uses a function from CRC8 or
CRC_CCITT any more. Building brcmsmac with these two disabled works
without any problems. This was probably only used by the bus code which
was replaced by bcma some time ago.
Lorenzo Bianconi [Sun, 24 Nov 2013 16:56:51 +0000 (17:56 +0100)]
ath9k: fix retry chain initialization in tx99 code
Initialize first chain attempt counter to 1 in ath9k_build_tx99_skb().
Otherwise multi-retry chain is initialized to {idx,count} = {-1, 0} in
rate_control_fill_sta_table() and tx99 transmission rate is not configured in
rate_control_apply_mask() since first chain idx is set to -1
Chun-Yeow Yeoh [Wed, 20 Nov 2013 09:57:01 +0000 (17:57 +0800)]
wcn36xx: set self STA default HT parameters
These default HT parameters are required for self STA entry.
In example, set the HT capable of self STA entry for bss
configuration in mesh allows the MCS rate to be used. Otherwise,
only legacy rate will be used.
Larry Finger [Mon, 18 Nov 2013 17:11:36 +0000 (11:11 -0600)]
rtlwifi: rtl8192c: Update dynamic gain calculations
The vendor driver contained a number of improvements in the gain settings
for the rtl8192c{e,u} devices. This patch implements them in the kernel
driver.
Larry Finger [Mon, 18 Nov 2013 17:11:27 +0000 (11:11 -0600)]
rtlwifi: Redo register save locations
The initial USB driver did not use some register save locations in the
private data storage. To save some memory, a union was used to overlay these
variables with USB I/O components. In an update of the gain-control code,
these register save locations are now needed for USB drivers.
Larry Finger [Mon, 18 Nov 2013 17:11:26 +0000 (11:11 -0600)]
rtlwifi: rtl8192cu: Add new firmware
Vendor driver rtl8188C_8192C_8192D_usb_linux_v3.4.2_3727.20120404 introduced
new firmware for these chips. The code try for the new file, and fall back to
the original firmware if the new file is not available.
Larry Finger [Mon, 18 Nov 2013 17:11:25 +0000 (11:11 -0600)]
rtlwifi: rtl8192c: Prevent reconnect attempts if not connected
This driver has a watchdog timer that attempts to reconnect when beacon frames
are not seen for 6 seconds. This patch disables that reconnect whenever the
device has never been connected.
Larry Finger [Tue, 12 Nov 2013 17:02:20 +0000 (11:02 -0600)]
rtlwifi: Remove unused calls to rtl_is_special_data()
When routine rtl_is_special_data() is called with false as its last argument,
and the returned value is not tested, the call is essentially an extended
no-op. Accordingly, these calls may be removed.
Larry Finger [Tue, 29 Oct 2013 22:06:41 +0000 (17:06 -0500)]
rtl8187: Increase RX queue depth
Under heavy load, the relatively small number of RX queue entries are
completely filled. With an increase from 16 to 32 entries, this condition
rarely happens.
freq_reg_info() expects KHz and not MHz, fix this. In
this case we'll now be getting the no-ir flags cleared
on channels for any channel when the country IE trusts
that channel.
freq_reg_info() expects KHz and not MHz, fix this. In
this case we'll now be getting the no-ir flags cleared
on channels for any channel when the country IE trusts
that channel.
Solomon Peachy [Fri, 25 Oct 2013 22:15:43 +0000 (18:15 -0400)]
wireless: cw1200: Use consistent internal locking conventions
The cw1200_irq_handler() function expects the hwbus lock to be held when
it is called. On the SDIO platform, this lock is implemented in terms
of sdio_claim_host/sdio_release_host.
This trivial patch makes it explicit that we are performing the hwbus
lock rather than something SDIO-specific.
We can mess logs if user space try to open device again and again if
RFKILL switch is on. Do not print message and return ERFKILL error
instead to indicate where the problem is.
Note that iwl4965 handle this problem differently, it allows to open
device when radio is disabled.
NeilBrown [Thu, 24 Oct 2013 07:33:33 +0000 (18:33 +1100)]
libertas/sdio: make sure card enters power-off when firmware is loaded.
Since firmware loading became async it is possible that if_sdio_finish_power_of
is called with only one remaining runtime_pm reference, so it isn't safe
to call pm_runtime_put_noidle. We must call pm_runtime_put().
Amitkumar Karwar [Tue, 22 Oct 2013 22:24:47 +0000 (15:24 -0700)]
mwifiex: improvements in mwifiex_ret_tx_power_cfg()
1) Move common code out of switch case handling
2) Return from the function if number of bytes left in response buffer
are less than tlv size
3) Pass pg_tlv_hdr directly instead of txp_cfg to mwifiex_get_power_level()
David S. Miller [Tue, 3 Dec 2013 21:55:49 +0000 (16:55 -0500)]
Merge branch 'cxgb4'
Hariprasad Shenai says:
====================
Fixes T5 adapter init, due to incorrect FW version check
This patch series fixes, Chelsio T5 adapter initialization failure due to
incorrect firmware version check. This patch series modifies the firmware
flashing mechanism for T4/T5 adapter.
The patch series moves chip type from struct adapter to struct adapter_params.
It changes the references of chip type in cxgb4 and cxgb4vf drivers such that
build failure is avoided.
Patch 3/3 is dependent on patch 1/3
Patch 2/3 is also dependent on patch 1/3
We would like to request this patch series to get merged via David Miller's
'net' tree.
We have included all the maintainers of respective drivers. Kindly review the
change and let us know in case of any review comments.
====================
rds: prevent BUG_ON triggered on congestion update to loopback
After congestion update on a local connection, when rds_ib_xmit returns
less bytes than that are there in the message, rds_send_xmit calls
back rds_ib_xmit with an offset that causes BUG_ON(off & RDS_FRAG_SIZE)
to trigger.
For a 4Kb PAGE_SIZE rds_ib_xmit returns min(8240,4096)=4096 when actually
the message contains 8240 bytes. rds_send_xmit thinks there is more to send
and calls rds_ib_xmit again with a data offset "off" of 4096-48(rds header)
=4048 bytes thus hitting the BUG_ON(off & RDS_FRAG_SIZE) [RDS_FRAG_SIZE=4k].
The commit 6094628bfd94323fc1cea05ec2c6affd98c18f7f
"rds: prevent BUG_ON triggering on congestion map updates" introduced
this regression. That change was addressing the triggering of a different
BUG_ON in rds_send_xmit() on PowerPC architecture with 64Kbytes PAGE_SIZE:
BUG_ON(ret != 0 &&
conn->c_xmit_sg == rm->data.op_nents);
This was the sequence it was going through:
(rds_ib_xmit)
/* Do not send cong updates to IB loopback */
if (conn->c_loopback
&& rm->m_inc.i_hdr.h_flags & RDS_FLAG_CONG_BITMAP) {
rds_cong_map_updated(conn->c_fcong, ~(u64) 0);
return sizeof(struct rds_header) + RDS_CONG_MAP_BYTES;
}
rds_ib_xmit returns 8240
rds_send_xmit:
c_xmit_data_off = 0 + 8240 - 48 (rds header accounted only the first time)
= 8192
c_xmit_data_off < 65536 (sg->length), so calls rds_ib_xmit again
rds_ib_xmit returns 8240
rds_send_xmit:
c_xmit_data_off = 8192 + 8240 = 16432, calls rds_ib_xmit again
and so on (c_xmit_data_off 24672,32912,41152,49392,57632)
rds_ib_xmit returns 8240
On this iteration this sequence causes the BUG_ON in rds_send_xmit:
while (ret) {
tmp = min_t(int, ret, sg->length - conn->c_xmit_data_off);
[tmp = 65536 - 57632 = 7904]
conn->c_xmit_data_off += tmp;
[c_xmit_data_off = 57632 + 7904 = 65536]
ret -= tmp;
[ret = 8240 - 7904 = 336]
if (conn->c_xmit_data_off == sg->length) {
conn->c_xmit_data_off = 0;
sg++;
conn->c_xmit_sg++;
BUG_ON(ret != 0 &&
conn->c_xmit_sg == rm->data.op_nents);
[c_xmit_sg = 1, rm->data.op_nents = 1]
What the current fix does:
Since the congestion update over loopback is not actually transmitted
as a message, all that rds_ib_xmit needs to do is let the caller think
the full message has been transmitted and not return partial bytes.
It will return 8240 (RDS_CONG_MAP_BYTES+48) when PAGE_SIZE is 4Kb.
And 64Kb+48 when page size is 64Kb.
Paul Durrant [Tue, 3 Dec 2013 14:06:25 +0000 (14:06 +0000)]
xen-netback: clear vif->task on disconnect
xenvif_start_xmit() relies on checking vif->task for NULL to determine
whether the vif is ready to accept packets. The task thread is stopped in
xenvif_disconnect() but task is not set to NULL. Thus, on a re-connect the
check will give a false positive.
Also since commit ea732dff5cfa10789007bf4a5b935388a0bb2a8f (Handle backend
state transitions in a more robust way) it should not be possible for
xenvif_connect() to be called if the vif is already connected so change the
check of vif->tx_irq to a BUG_ON() and also add a BUG_ON(vif->task).
It causes regressions for people using chips driven by the sungem
driver. Suspicion is that the skb->csum value isn't being adjusted
properly.
The change also has a bug in that if __pskb_trim() fails, we'll leave
a corruped skb->csum value in there. We would really need to revert
it to it's original value in that case.
Kamala R [Mon, 2 Dec 2013 14:25:21 +0000 (19:55 +0530)]
IPv6: Fixed support for blackhole and prohibit routes
The behaviour of blackhole and prohibit routes has been corrected by setting
the input and output pointers of the dst variable appropriately. For
blackhole routes, they are set to dst_discard and to ip6_pkt_discard and
ip6_pkt_discard_out respectively for prohibit routes.
ipv6: ip6_pkt_prohibit(_out) should not depend on
CONFIG_IPV6_MULTIPLE_TABLES
We need ip6_pkt_prohibit(_out) available without
CONFIG_IPV6_MULTIPLE_TABLES
net: fec_main: dma_map() only the length of the skb
On tx submit the driver always dma_map_single() FEC_ENET_TX_FRSIZE (=2048)
bytes. This works because we don't overwrite any memory after the data buffer,
we remove it from cache if it was there. So we hurt performace in case the
mapping of a smaller area makes a difference.
There is also a bug: If the data area starts shortly before the end of
RAM say 0xc7fffa10 and the RAM ends at 0xc8000000 then we have enough
space to fit the data area (according to skb->len) but we would map beyond
end of ram if we are using 2048. In v2.6.31 (against which kernel this patch
made) there is the following check in dma_cache_maint():
Since the area starting at 0xc8000000 is no longer virt_addr_valid() we
BUG() during dma_map_single(). The BUG() statement was removed in v3.5-rc1 as
per 2dc6a016 ("ARM: dma-mapping: use asm-generic/dma-mapping-common.h").
This patch was tested on v2.6.31 and then forward-ported and compile
tested only against the net tree. I think it is still worth fixing
mainline even after the BUG() statement is gone.
Mugunthan V N [Mon, 2 Dec 2013 07:23:39 +0000 (12:53 +0530)]
drivers: net: cpsw: fix dt probe for one port ethernet
When only one port of the two port is pinned out, then dt probe is failing
because second port phy is not found. fixing this by checking the number of
slaves and breaking the loop.
PCI / tg3: Give up chip reset and carrier loss handling if PCI device is not present
Modify tg3_chip_reset() and tg3_close() to check if the PCI network
adapter device is accessible at all in order to skip poking it or
trying to handle a carrier loss in vain when that's not the case.
Introduce a special PCI helper function pci_device_is_present()
for this purpose.
Of course, this uncovers the lack of the appropriate RTNL locking
in tg3_suspend() and tg3_resume(), so add that locking in there
too.
These changes prevent tg3 from burning a CPU at 100% load level for
solid several seconds after the Thunderbolt link is disconnected from
a Matrox DS1 docking station.
Duan Jiong [Tue, 26 Nov 2013 07:46:56 +0000 (15:46 +0800)]
ipv6: judge the accept_ra_defrtr before calling rt6_route_rcv
when dealing with a RA message, if accept_ra_defrtr is false,
the kernel will not add the default route, and then deal with
the following route information options. Unfortunately, those
options maybe contain default route, so let's judge the
accept_ra_defrtr before calling rt6_route_rcv.
projects / linux.git / log