On 21/06/2017 22:52, Gustavo A. R. Silva wrote:
Hi Ethan,
Quoting Ethan Zhao :
Gustavo,
The return value of ret_val seems used to check if the access to
PHY/NVM
got its semaphore, generally speaking, it is needed for every PHY
access of this driver.
Reviewed-by: Ethan Zhao
Thank yo
From: Jan Kiszka
Make stmmac_default_data compatible with stmmac_pci_info.setup and use
an info structure for all devices. This allows to make the probing more
regular.
Signed-off-by: Jan Kiszka
Reviewed-by: Andy Shevchenko
---
drivers/net/ethernet/stmicro/stmmac/stmmac_pci.c | 36 +++
Some cleanups of the way we probe DMI platforms in the driver. Reduces
a bit of open-coding and makes the logic easier reusable for any
potential DMI platform != Quark.
Tested on IOT2000 and Galileo Gen2.
Changes in v5:
- fixed a remaining issue in patch 5
- dropped patch 6 for now
Jan
Jan Ki
From: Jan Kiszka
No need to carry this reference in stmmac_pci_info - the Quark-specific
setup handler knows that it needs to use the Quark-specific DMI table.
This also allows to drop the stmmac_pci_info reference from the setup
handler parameter list.
Signed-off-by: Jan Kiszka
Reviewed-by: An
From: Jan Kiszka
Move the special case for the early Galileo firmware into
quark_default_setup. This allows to use stmmac_pci_find_phy_addr for
non-quark cases.
Signed-off-by: Jan Kiszka
Reviewed-by: Andy Shevchenko
---
drivers/net/ethernet/stmicro/stmmac/stmmac_pci.c | 20 +--
From: Jan Kiszka
By removing the PCI device reference from the structure and passing it
as parameters to the interested functions, we can make quark_pci_info
const.
Signed-off-by: Jan Kiszka
Reviewed-by: Andy Shevchenko
---
drivers/net/ethernet/stmicro/stmmac/stmmac_pci.c | 23 +++
From: Jan Kiszka
Avoids reimplementation of DMI matching in stmmac_pci_find_phy_addr.
Signed-off-by: Jan Kiszka
---
drivers/net/ethernet/stmicro/stmmac/stmmac_pci.c | 97
1 file changed, 64 insertions(+), 33 deletions(-)
diff --git a/drivers/net/ethernet/stmicro/stmma
We should probably add a might_sleep() to ioremap() to prevent these
bugs in the future.
This bug is eight years old. You can report it, but it's going to hard
to get anyone to fix it. I sometimes ignore ancient bugs. On the other
hand, netxen is fairly well supported so it doesn't hurt to try.
On Wed, Jun 21, 2017 at 10:07:48AM -0600, Jason Gunthorpe wrote:
> On Wed, Jun 21, 2017 at 09:05:17AM +0300, Leon Romanovsky wrote:
> > From: Leon Romanovsky
> >
> > Provide ability to convert from device name to ib_device for the
> > IB/core users.
>
> FWIW, as a general comment, we really need t
On Wed, Jun 21, 2017 at 02:25:30PM +0100, Colin King wrote:
> From: Colin Ian King
>
> The return value ret is unitialized and garbage is being returned
> for the three different error conditions when setting up the PCIe
> BARs. Fix this by initializing ret to -ENOMEM to indicate that
> the BARs
On Wed, 21 Jun 2017 16:04:46 +0300, Gal Pressman wrote:
> Use the new ethtool link down reason api, and expose troubleshooting
> info regarding the link status.
>
> Signed-off-by: Gal Pressman
> Signed-off-by: Saeed Mahameed
Is my reading correct that in case the reason is not in the
pddr2ethto
On Wed, Jun 21, 2017 at 04:18:54PM +, Tung, Chien Tin wrote:
> > From: Jason Gunthorpe [mailto:jguntho...@obsidianresearch.com]
> > Sent: Wednesday, June 21, 2017 11:12 AM
> > To: Leon Romanovsky
> > Cc: Doug Ledford ; linux-r...@vger.kernel.org; Tung,
> > Chien Tin ; Steve Wise
> > ; Stephen
On Wed, 21 Jun 2017 16:04:43 +0300, Gal Pressman wrote:
> Hi All,
>
> Currently, drivers can only tell whether the link is up/down through
> ETHTOOL_GLINK callback, but no additional information is given in case
> of link down/failure.
>
> This series provides an infrastructure to ethtool that al
Hi, Andrew
On 2017/6/22 11:35, Andrew Lunn wrote:
>> I don't think returning to user space is an option, because it mean
>> ethtool phy self test would fail because if availability of some software
>> resoure, which is not some application would expect. here is what I can
>> think of:
>>
>> driver
> I don't think returning to user space is an option, because it mean
> ethtool phy self test would fail because if availability of some software
> resoure, which is not some application would expect. here is what I can
> think of:
>
> driver/net/ethernet/hisilicon/hns/hns_ethtool.c
> int phy_loop
On Tue, Jun 20, 2017 at 3:48 PM, Scott Branden
wrote:
> Hi Rob,
>
>
> On 17-06-18 07:04 AM, Rob Herring wrote:
>>
>> On Mon, Jun 12, 2017 at 01:26:00PM -0700, Arun Parameswaran wrote:
>>>
>>> Add device tree binding documentation for the Broadcom DTE
>>> PTP clock driver.
>>>
>>> Signed-off-by: Ar
On 6/21/17 5:34 PM, Cong Wang wrote:
> diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
> index 686c923..1d2dbac 100644
> --- a/net/ipv6/addrconf.c
> +++ b/net/ipv6/addrconf.c
> @@ -3369,6 +3369,7 @@ static int addrconf_notify(struct notifier_block *this,
> unsigned long event,
> struc
Hi, Andrew
On 2017/6/21 21:34, Andrew Lunn wrote:
>> drivers/net/ph/marvell.c
>> marvell_set_loopback(struct phy_device *dev, bool enable)
>> {
>> /* do some device specific setting */
>>
>>
>> return genphy_loopback(dev, enable);
>> }
>>
>> I don't know if this makes sense
Add an installation-time flag for requesting that the program
be installed only if it can be offloaded to HW.
Internally new command for ndo_xdp is added, this way we avoid
putting checks into drivers since they all return -EINVAL on
an unknown command.
Signed-off-by: Jakub Kicinski
Acked-by: Da
In preparation of XDP offload flags move the driver setup into
a function. Otherwise the number of conditions in one function
would make it slightly hard to follow. The offload handler may
now be called with NULL prog, even if no offload is currently
active, but that's fine, offload code can hand
DRV_MODE means that user space wants the program to be run in
the driver. Do not try to offload. Only offload if no mode
flags have been specified.
Remember what the mode is when the program is installed and refuse
new setup requests if there is already a program loaded in a
different mode. Thi
Pass XDP flags to the xdp ndo. This will allow drivers to look
at the mode flags and make decisions about offload.
Signed-off-by: Jakub Kicinski
Acked-by: Daniel Borkmann
---
include/linux/netdevice.h | 1 +
net/core/dev.c| 5 +++--
2 files changed, 4 insertions(+), 2 deletions(-)
The xdp_prog member of the adapter's data path structure is used
for XDP in driver mode. In case a XDP program is loaded with in
HW-only mode, we need to store it somewhere else. Add a new XDP
prog pointer in the main structure and use that when we need to
know whether any XDP program is loaded,
Extend the XDP_ATTACHED_* values to include offloaded mode.
Let drivers report whether program is installed in the driver
or the HW by changing the prog_attached field from bool to
u8 (type of the netlink attribute).
Exploit the fact that the value of XDP_ATTACHED_DRV is 1,
therefore since all dri
Make use of just added XDP_ATTACHED_HW.
Signed-off-by: Jakub Kicinski
---
drivers/net/ethernet/netronome/nfp/nfp_net_common.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_common.c
b/drivers/net/ethernet/netronome/nfp/nfp_net_common.c
index c9f
Respect the XDP_FLAGS_HW_MODE. When it's set install the program
on the NIC and skip enabling XDP in the driver.
Signed-off-by: Jakub Kicinski
---
drivers/net/ethernet/netronome/nfp/nfp_net_common.c | 13 ++---
1 file changed, 10 insertions(+), 3 deletions(-)
diff --git a/drivers/net/e
Hi!
While we discuss the representors.. :)
This set adds XDP flag for forcing offload and a attachment mode
for reporting to user space that program has been offloaded. The
nfp driver is modified to make use of the new flags, but also to
adhere to the DRV_MODE flag which should disable the HW of
On Wed, 2017-06-21 at 13:12 -0500, Larry Finger wrote:
> On 06/19/2017 03:18 PM, Joe Perches wrote:
[]
> In general, I approve of your patches. Unfortunately, they will conflict with
> a
> number of changes that I am currently making. If you would like, I can merge
> your patches, fix the confli
On 2017/6/22 2:28, Alexander Duyck wrote:
> On Sun, Jun 18, 2017 at 11:53 PM, Ding Tianhong
> wrote:
>> From: Casey Leedom
>>
>> cxgb4 Ethernet driver now queries PCIe configuration space to determine
>> if it can send TLPs to it with the Relaxed Ordering Attribute set.
>>
>> Signed-off-by: Ca
On 6/21/17, 10:13 AM, "Craig Gallek" wrote:
On Wed, Jun 21, 2017 at 12:51 PM, Lawrence Brakmo wrote:
>
> On 6/20/17, 2:25 PM, "Craig Gallek" wrote:
>
> On Mon, Jun 19, 2017 at 11:00 PM, Lawrence Brakmo
wrote:
> > Added support for calling a subset of socket se
Sorry, it's not dead loop, I realized it but I tried to send to mail to note it
on my phone, it failed.
> -Original Message-
> From: Cong Wang [mailto:xiyou.wangc...@gmail.com]
> Sent: Thursday, June 22, 2017 6:54 AM
> To: yuan linyu
> Cc: Linux Kernel Network Developers; David S . Mille
From: Haiyang Zhang
We simply use rndis_device->link_state in the netdev_dbg. The variable,
link_state from struct netvsc_device_info, is not used anywhere else.
Signed-off-by: Haiyang Zhang
Reviewed-by: Stephen Hemminger
---
drivers/net/hyperv/hyperv_net.h |5 +++--
drivers/net/hyperv/
From: Haiyang Zhang
When the VF NIC is opened, the synthetic NIC's carrier state is set to
off. This tells the host to transitions data path to the VF device. But
if startup script or user manipulates the admin state of the netvsc
device directly for example:
# ifconfig eth0 down
On Thu, 22 Jun 2017 00:16:37 +0200
Arnd Bergmann wrote:
> This structure member is hidden behind CONFIG_SYSFS, and we
> get a build error when that is disabled:
>
> drivers/net/hyperv/netvsc_drv.c: In function 'netvsc_set_channels':
> drivers/net/hyperv/netvsc_drv.c:754:49: error: 'struct net_de
On 21/06/17 23:39, Jay Vosburgh wrote:
Michael J Dilmore wrote:
On 21/06/17 22:56, David Miller wrote:
From: Michael D
Date: Wed, 21 Jun 2017 22:41:07 +0100
I don't think you can stop it being dereferenced... you just need to
prevent an attacker from exploiting the null pointer dereferenc
Wir vergeben Kredite mit einem Zinssatz von jährlich 2%.
Die Bearbeitung des Antrags erfolgt rasch, wir verlangen keine Gebühren, was
sie beantragen werden wir annehmen. Wir bewilligen Kredite von bis zu 40
Millionen Euro und von mindestens 15.000 Euro. Sie können einen geschäftlichen
oder pri
On Wed, Jun 21, 2017 at 7:02 AM, yuan linyu wrote:
> From: yuan linyu
>
> when CONFIG_IP_MULTIPLE_TABLES defined,
> if id == RT_TABLE_LOCAL and !net->ipv4.fib_has_custom_rules,
> fib_new_table() call itself to get RT_TABLE_MAIN table,
> but if RT_TABLE_MAIN table not exist at this point,
> this f
Michael J Dilmore wrote:
>On 21/06/17 22:56, David Miller wrote:
>
>> From: Michael D
>> Date: Wed, 21 Jun 2017 22:41:07 +0100
>>
>>> I don't think you can stop it being dereferenced... you just need to
>>> prevent an attacker from exploiting the null pointer dereference
>>> vulnerability right?
David Miller wrote:
>From: Michael D
>Date: Wed, 21 Jun 2017 22:41:07 +0100
>
>> I don't think you can stop it being dereferenced... you just need to
>> prevent an attacker from exploiting the null pointer dereference
>> vulnerability right? And this is done by returning the function right
>> aw
On 21/06/17 22:56, David Miller wrote:
From: Michael D
Date: Wed, 21 Jun 2017 22:41:07 +0100
I don't think you can stop it being dereferenced... you just need to
prevent an attacker from exploiting the null pointer dereference
vulnerability right? And this is done by returning the function ri
> -Original Message-
> From: Arnd Bergmann [mailto:a...@arndb.de]
> Sent: Wednesday, June 21, 2017 6:17 PM
> To: Stephen Hemminger
> Cc: Arnd Bergmann ; KY Srinivasan ;
> Haiyang Zhang ; David S. Miller
> ; Simon Xiao ;
> de...@linuxdriverproject.org; netdev@vger.kernel.org; linux-
> ker
This structure member is hidden behind CONFIG_SYSFS, and we
get a build error when that is disabled:
drivers/net/hyperv/netvsc_drv.c: In function 'netvsc_set_channels':
drivers/net/hyperv/netvsc_drv.c:754:49: error: 'struct net_device' has no
member named 'num_rx_queues'; did you mean 'num_tx_que
I still see build errors in randconfig builds and have had this
patch for a while to locally work around it:
drivers/built-in.o: In function `xgene_mdio_probe':
mux-core.c:(.text+0x352154): undefined reference to `of_mdiobus_register'
mux-core.c:(.text+0x352168): undefined reference to `mdiobus_fr
The driver attempts to 'select MDIO_DEVICE', but the code
is actually a loadable module when PHYLIB=m:
drivers/phy/broadcom/phy-bcm-ns-usb3.o: In function
`bcm_ns_usb3_mdiodev_phy_write':
phy-bcm-ns-usb3.c:(.text.bcm_ns_usb3_mdiodev_phy_write+0x28): undefined
reference to `mdiobus_write'
drivers
An extraneous semicolon crept in, causing a build failure:
In file included from /git/arm-soc/drivers/net/ethernet/qlogic/qede/qede.h:43:0,
from
/git/arm-soc/drivers/net/ethernet/qlogic/qede/qede_main.c:63:
include/linux/qed/qede_rdma.h:85:1: error: expected identifier or '(' bef
And please stop top posting, thank you.
From: Michael D
Date: Wed, 21 Jun 2017 22:41:07 +0100
> I don't think you can stop it being dereferenced... you just need to
> prevent an attacker from exploiting the null pointer dereference
> vulnerability right? And this is done by returning the function right
> away?
What's all of this about
I don't think you can stop it being dereferenced... you just need to
prevent an attacker from exploiting the null pointer dereference
vulnerability right? And this is done by returning the function right
away?
On 21 June 2017 at 22:36, David Miller wrote:
> From: Michael D
> Date: Wed, 21 Ju
From: Wei Wang
In __ip6_datagram_connect(), reset sk->sk_v6_daddr and inet->dport if
error occurs so that udp_v6_early_demux() won't be able to do exact match on
this socket and hence, won't consider this socket as a valid candidate for
early demux.
Signed-off-by: Wei Wang
Acked-by: Maciej Żenc
From: Michael D
Date: Wed, 21 Jun 2017 22:35:56 +0100
> How do we actually stop a null pointer being dereferenced here?
Maybe that's why it's a BUG_ON(), there is no reasonable way
to continue if the pointer is NULL.
How do we actually stop a null pointer being dereferenced here? Other
examples I've seen check for null and then immediately return the
function with an error code so that it cannot be referenced again.
Something like:
if (WARN_ON(!new_active)
return 1
This behaviour should be OK for this f
On 06/21/2017 11:29 PM, Martin KaFai Lau wrote:
This patch adds support to the newly added IFLA_XDP_PROG_ID.
./ip link show dev eth0
3: eth0: mtu 1500 xdpgeneric/id:2 qdisc [...]
Signed-off-by: Martin KaFai Lau
Acked-by: Daniel Borkmann
The per netns loopback_dev->ip6_ptr is unregistered and set to
NULL when its mtu is set to smaller than IPV6_MIN_MTU, this
leads to that we could set rt->rt6i_idev NULL after a
rt6_uncached_list_flush_dev() and then crash after another
call.
In this case we should just bring its inet6_dev down, ra
This patch adds support to the newly added IFLA_XDP_PROG_ID.
./ip link show dev eth0
3: eth0: mtu 1500 xdpgeneric/id:2 qdisc [...]
Signed-off-by: Martin KaFai Lau
---
include/linux/if_link.h | 1 +
ip/iplink_xdp.c | 12 +---
2 files changed, 10 insertions(+), 3 deletions(-)
d
From: Shannon Nelson
Date: Wed, 21 Jun 2017 13:59:45 -0700
> On 6/21/2017 12:06 PM, David Miller wrote:
>> From: Shannon Nelson
>> Date: Wed, 21 Jun 2017 09:09:54 -0700
>>
>>> Add the get_settings callback so that both the ldmvsw and sunvnet
>>> drivers will give a little more information when
From: Shannon Nelson
Date: Wed, 21 Jun 2017 13:59:35 -0700
> On 6/21/2017 12:05 PM, David Miller wrote:
>> Also, this is not at all an appropriate place to expose this. The
>> "driver info" ethtool command is not a place to post per-connection
>> or per-link information.
>
> I knew it was a bit
From: Willem de Bruijn
Introduce zerocopy socket send flag MSG_ZEROCOPY. This extends the
shared page support (SKBTX_SHARED_FRAG) from sendpage to sendmsg.
Implement the feature for TCP, UDP, RAW and packet sockets.
On a send call with MSG_ZEROCOPY, the kernel pins user pages and
links these dir
From: Willem de Bruijn
Add sock_omalloc and sock_ofree to be able to allocate control skbs,
for instance for looping errors onto sk_error_queue.
The transmit budget (sk_wmem_alloc) is involved in transmit skb
shaping, most notably in TCP Small Queues. Using this budget for
control packets would
From: Willem de Bruijn
The send call ignores unknown flags. Legacy applications may already
unwittingly pass MSG_ZEROCOPY. Continue to ignore this flag unless a
socket opts in to zerocopy.
Introduce socket option SO_ZEROCOPY to enable MSG_ZEROCOPY processing.
Processes can also query this socket
From: Willem de Bruijn
The kernel supports zerocopy sendmsg in virtio and tap. Expand the
infrastructure to support other socket types. Introduce a completion
notification channel over the socket error queue. Notifications are
returned with ee_origin SO_EE_ORIGIN_ZEROCOPY. ee_errno is 0 to avoid
From: Willem de Bruijn
A send request with MSG_ZEROCOPY falls back to copying when a request
cannot be completed in zerocopy mode.
Notify processes when the call reverted to copying through a new bit
in the completion notification, so that they can optionally decline
passing the flag on subseque
From: Willem de Bruijn
Add MSG_ZEROCOPY support to INET(6). This includes UDP, but also
RAW sockets that do not take the raw_send_hdrinc() path.
Zerocopy is only effective when payload is not touched at all. Limit
it to paths that support both checksum offload and scatter-gather.
When a caller
From: Willem de Bruijn
In the simple case, each sendmsg() call generates data and eventually
a zerocopy ready notification N, where N indicates the Nth successful
invocation of sendmsg() with the MSG_ZEROCOPY flag on this socket.
TCP and corked sockets can cause send() calls to append new data t
From: Willem de Bruijn
Refine skb_copy_ubufs to support compound pages. With upcoming TCP
and UDP zerocopy sendmsg, such fragments may appear.
The existing code replaces each page one for one. Splitting each
compound page into an independent number of regular pages can result
in exceeding limit
From: Willem de Bruijn
Prepare the datapath for refcounted ubuf_info. Clone ubuf_info with
skb_zerocopy_clone() wherever needed due to skb split, merge, resize
or clone.
Split skb_orphan_frags into two variants. The split, merge, .. paths
support reference counted zerocopy buffers, so do not do
From: Willem de Bruijn
Enable support for MSG_ZEROCOPY to the TCP stack. TSO and GSO are
both supported. Only data sent to remote destinations is sent without
copying. Packets looped onto a local destination have their payload
copied to avoid unbounded latency.
Tested:
A 10x TCP_STREAM between
From: Willem de Bruijn
Introduce regression test for msg_zerocopy feature. Send traffic from
one process to another with and without zerocopy.
Evaluate tcp, udp, raw and packet sockets, including variants
- udp: corking and corking with mixed copy/zerocopy calls
- raw: with and without hdrincl
-
From: Willem de Bruijn
Support MSG_ZEROCOPY on PF_PACKET transmission.
Tested:
msg_zerocopy.sh 4 packet:
without zerocopy
tx=121543 (7588 MB) txc=0 zc=n
rx=121543 (7584 MB)
with zerocopy
tx=338252 (21119 MB) txc=338252 zc=y
rx=338252 (21108 MB)
msg_zerocopy.sh 4 packe
From: Willem de Bruijn
Bound the number of pages that a user may pin.
Follow the lead of perf tools to maintain a per-user bound on memory
locked pages commit 789f90fcf6b0 ("perf_counter: per user mlock gift")
Signed-off-by: Willem de Bruijn
---
include/linux/sched/user.h | 3 ++-
include/li
From: Willem de Bruijn
Zerocopy support for udp also enables it for some raw sockets. Only
raw sockets that have hdrinc set take a different path. Add zerocopy
support for this variant.
Tested:
msg_zerocopy.sh 4 raw_hdrincl:
without zerocopy
tx=150438 (9390 MB) txc=0 zc=n
rx=150438
On Wed, Jun 21, 2017 at 08:16:11PM +0200, Daniel Borkmann wrote:
> In order to be able to retrieve the attached programs from cls_bpf
> and act_bpf, we need to expose the prog ids via netlink so that
> an application can later on get an fd based on the id through the
> BPF_PROG_GET_FD_BY_ID command
Hi Dave,
On Mon, 19 Jun 2017 13:37:43 -0400 (EDT)
David Miller wrote:
> From: Martin Schwidefsky
> Date: Mon, 19 Jun 2017 17:34:25 +0200
>
> > We (as in the s390 guys) tend to add __packed to hardware and hypervisor
> > structures even if the attribute is not strictly necessary. Most of the
>
On Wed, Jun 21, 2017 at 1:08 PM, Cong Wang wrote:
> On Wed, Jun 21, 2017 at 10:53 AM, Andrey Konovalov
> wrote:
>> On Wed, Jun 21, 2017 at 3:09 PM, Andrey Konovalov
>> wrote:
>>> On Wed, Jun 21, 2017 at 2:08 PM, Andrey Konovalov
>>> wrote:
Hi,
I've got the following error repor
On 6/21/2017 12:06 PM, David Miller wrote:
From: Shannon Nelson
Date: Wed, 21 Jun 2017 09:09:54 -0700
Add the get_settings callback so that both the ldmvsw and sunvnet
drivers will give a little more information when asked for its
basic settings. These aren't necessarily very useful, but they
On 6/21/2017 12:05 PM, David Miller wrote:
From: Shannon Nelson
Date: Wed, 21 Jun 2017 09:09:53 -0700
In the ethtool -i output print the vio version and the remote-mac
of the ldom that the vif device is serving as this vif info is
not exposed elsewhere. The remote-mac address is most useful f
On Wed, Jun 21, 2017 at 4:36 PM, Willem de Bruijn
wrote:
> From: Willem de Bruijn
>
> The send call ignores unknown flags. Legacy applications may already
> unwittingly pass MSG_ZEROCOPY. Continue to ignore this flag unless a
> socket opts in to zerocopy.
>
> Introduce socket option SO_ZEROCOPY t
tracex5_kern.c build failed with the following error message:
../samples/bpf/tracex5_kern.c:12:10: fatal error: 'syscall_nrs.h' file not
found
#include "syscall_nrs.h"
The generated file syscall_nrs.h is put in build/samples/bpf directory,
but this directory is not in include path, hence build
Patch1 is a bug fix for correct reconnect when a connection
is restarted. Patch 2 accelerates cleanup by setting linger
to 1 and sending a RST to the peer.
Sowmini Varadhan (2):
rds: tcp: send handshake ping-probe from passive endpoint
rds: tcp: set linger to 1 when unloading a rds-tcp
net/r
The RDS handshake ping probe added by commit 5916e2c1554f
("RDS: TCP: Enable multipath RDS for TCP") is sent from rds_sendmsg()
before the first data packet is sent to a peer. If the conversation
is not bidirectional (i.e., one side is always passive and never
invokes rds_sendmsg()) and the passiv
If we are unloading the rds_tcp module, we can set linger to 1
and drop pending packets to accelerate reconnect. The peer will
end up resetting the connection based on new generation numbers
of the new incarnation, so hanging on to unsent TCP packets via
linger is mostly pointless in this case.
Si
The update to ibmvnic_init to allow an EAGAIN return code broke
the calling of ibmvnic_init from ibmvnic_probe. The code now
will return from this point in the probe routine if anything
other than EAGAIN is returned. The check should be to see if rc
is non-zero and not equal to EAGAIN.
Without thi
From: Willem de Bruijn
The kernel supports zerocopy sendmsg in virtio and tap. Expand the
infrastructure to support other socket types. Introduce a completion
notification channel over the socket error queue. Notifications are
returned with ee_origin SO_EE_ORIGIN_ZEROCOPY. ee_errno is 0 to avoid
From: Willem de Bruijn
The send call ignores unknown flags. Legacy applications may already
unwittingly pass MSG_ZEROCOPY. Continue to ignore this flag unless a
socket opts in to zerocopy.
Introduce socket option SO_ZEROCOPY to enable MSG_ZEROCOPY processing.
Processes can also query this socket
> -Original Message-
> From: Yossi Kuperman
> Sent: Wednesday, June 21, 2017 7:00 PM
> To: 'Steffen Klassert'
> Cc: netdev@vger.kernel.org; Herbert Xu ;
> Yevgeny Kliteynik ; Boris Pismenny
> ; Ilan Tayari
> Subject: RE: [PATCH net 1/2] xfrm6: Fix IPv6 payload_len in
> xfrm6_transport_f
From: Willem de Bruijn
Introduce zerocopy socket send flag MSG_ZEROCOPY. This extends the
shared page support (SKBTX_SHARED_FRAG) from sendpage to sendmsg.
Implement the feature for TCP, UDP, RAW and packet sockets.
On a send call with MSG_ZEROCOPY, the kernel pins user pages and
links these dir
From: Willem de Bruijn
A send request with MSG_ZEROCOPY falls back to copying when a request
cannot be completed in zerocopy mode.
Notify processes when the call reverted to copying through a new bit
in the completion notification, so that they can optionally decline
passing the flag on subseque
From: Willem de Bruijn
Prepare the datapath for refcounted ubuf_info. Clone ubuf_info with
skb_zerocopy_clone() wherever needed due to skb split, merge, resize
or clone.
Split skb_orphan_frags into two variants. The split, merge, .. paths
support reference counted zerocopy buffers, so do not do
From: Willem de Bruijn
Bound the number of pages that a user may pin.
Follow the lead of perf tools to maintain a per-user bound on memory
locked pages commit 789f90fcf6b0 ("perf_counter: per user mlock gift")
Signed-off-by: Willem de Bruijn
---
include/linux/sched/user.h | 3 ++-
include/li
From: Willem de Bruijn
Add sock_omalloc and sock_ofree to be able to allocate control skbs,
for instance for looping errors onto sk_error_queue.
The transmit budget (sk_wmem_alloc) is involved in transmit skb
shaping, most notably in TCP Small Queues. Using this budget for
control packets would
From: Willem de Bruijn
Zerocopy support for udp also enables it for some raw sockets. Only
raw sockets that have hdrinc set take a different path. Add zerocopy
support for this variant.
Tested:
msg_zerocopy.sh 4 raw_hdrincl:
without zerocopy
tx=150438 (9390 MB) txc=0 zc=n
rx=150438
From: Willem de Bruijn
Enable support for MSG_ZEROCOPY to the TCP stack. TSO and GSO are
both supported. Only data sent to remote destinations is sent without
copying. Packets looped onto a local destination have their payload
copied to avoid unbounded latency.
Tested:
A 10x TCP_STREAM between
From: Willem de Bruijn
Add MSG_ZEROCOPY support to INET(6). This includes UDP, but also
RAW sockets that do not take the raw_send_hdrinc() path.
Zerocopy is only effective when payload is not touched at all. Limit
it to paths that support both checksum offload and scatter-gather.
When a caller
From: Willem de Bruijn
Introduce regression test for msg_zerocopy feature. Send traffic from
one process to another with and without zerocopy.
Evaluate tcp, udp, raw and packet sockets, including variants
- udp: corking and corking with mixed copy/zerocopy calls
- raw: with and without hdrincl
-
From: Willem de Bruijn
In the simple case, each sendmsg() call generates data and eventually
a zerocopy ready notification N, where N indicates the Nth successful
invocation of sendmsg() with the MSG_ZEROCOPY flag on this socket.
TCP and corked sockets can cause send() calls to append new data t
From: Willem de Bruijn
Support MSG_ZEROCOPY on PF_PACKET transmission.
Tested:
msg_zerocopy.sh 4 packet:
without zerocopy
tx=121543 (7588 MB) txc=0 zc=n
rx=121543 (7584 MB)
with zerocopy
tx=338252 (21119 MB) txc=338252 zc=y
rx=338252 (21108 MB)
msg_zerocopy.sh 4 packe
From: Willem de Bruijn
Refine skb_copy_ubufs to support compound pages. With upcoming TCP
and UDP zerocopy sendmsg, such fragments may appear.
The existing code replaces each page one for one. Splitting each
compound page into an independent number of regular pages can result
in exceeding limit
On Wed, Jun 21, 2017 at 10:53 AM, Andrey Konovalov
wrote:
> On Wed, Jun 21, 2017 at 3:09 PM, Andrey Konovalov
> wrote:
>> On Wed, Jun 21, 2017 at 2:08 PM, Andrey Konovalov
>> wrote:
>>> Hi,
>>>
>>> I've got the following error report while fuzzing the kernel with syzkaller.
>>>
>>> On commit 9
This is quite counter-intuitive when using the 'preference' keyword.
Signed-off-by: Lukas Braun
---
man/man8/ip-route.8.in | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/man/man8/ip-route.8.in b/man/man8/ip-route.8.in
index c8eb38a..d74e2a8 100644
--- a/man/man8/ip-route.8.i
Hi,
On Wed, Jun 21, 2017 at 2:22 AM, Dison River wrote:
> Hi:
> I've got the following error report while fuzzing the kernel with
> syzkaller on v4.1.40
>
>
> Syzkaller hit 'possible deadlock in rtnl_lock' bug on commit .
>
> The guilty file is: /home/river/git_new/linux-stable/net/core/rtnet
On 06/21/2017 09:37 PM, Shubham Bansal wrote:
Hi Daniel,
Good news. Got the CALL to work.
[ 145.670882] test_bpf: Summary: 316 PASSED, 0 FAILED, [287/308 JIT'ed]
Awesome. Do you think with this implementation, the patch could get
accepted? If you think so, then I will send the patch in couple
1 - 100 of 259 matches
Mail list logo
