Signed-off-by: Yoshinori Sato
---
Documentation/devicetree/bindings/net/ne2000.txt | 17 +
drivers/net/ethernet/8390/ne.c | 20 +++-
2 files changed, 36 insertions(+), 1 deletion(-)
create mode 100644
Signed-off-by: Yoshinori Sato
---
drivers/net/ethernet/8390/Kconfig | 2 +-
drivers/net/ethernet/8390/ne.c| 3 ++-
2 files changed, 3 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/8390/Kconfig
b/drivers/net/ethernet/8390/Kconfig
index
On Mon, 28 Sep 2015 04:51:30 +0900,
David Miller wrote:
>
> From: Yoshinori Sato
> Date: Sun, 27 Sep 2015 23:19:15 +0900
>
> > @@ -88,7 +88,7 @@ config MCF8390
> > config NE2000
> > tristate "NE2000/NE1000 support"
> > depends on (ISA || (Q40 && m) || M32R
Hi David,
[auto build test results on next-20151001 -- if it's inappropriate base, please
ignore]
config: mips-nlm_xlp_defconfig (attached as .config)
reproduce:
wget
https://git.kernel.org/cgit/linux/kernel/git/wfg/lkp-tests.git/plain/sbin/make.cross
-O ~/bin/make.cross
chmod
On 01. okt. 2015 13:52, Eric Dumazet wrote:
On Thu, Oct 1, 2015 at 4:43 AM, Holger Hoffstätte
wrote:
On 10/01/15 13:29, Eric Dumazet wrote:
commit 83fccfc3940c4a2db90fd7e7079f5b465cd8c6af
Author: Eric Dumazet
Date: Thu Aug 13
On Thu, 1 Oct 2015 17:11:56 -0700, Pravin Shelar wrote:
> I dont see point of adding this code when IPv6 sampling not support by
> the patch series.
It was requested by Jesse:
http://article.gmane.org/gmane.linux.network/380348
I don't mind leaving this and the previous patch out, it's actually
Tidy-up the Data header 32-bit word synchronisation logic in
asix_rx_fixup_internal() by removing redundant logic tests.
The code is looking at the following cases of the Data header
32-bit word that is present before each Ethernet frame:
a) all 32 bits of the Data header word are in the URB
Avoid a loss of synchronisation of the Ethernet Data header 32-bit
word due to a failure to get a netdev socket buffer.
The ASIX RX handling algorithm returned 0 upon a failure to get
an allocation of a netdev socket buffer. This causes the URB
processing to stop which potentially causes a loss
When RX Ethernet frames span multiple URB socket buffers,
the data stream may suffer a discontinuity which will cause
the current Ethernet frame in the netdev socket buffer
to be incomplete. This frame needs to be discarded instead
of appending unrelated data from the current URB socket buffer
to
From: Mark Craske
The ASIX RX handler algorithm is weak on error handling.
There is a design flaw in the ASIX RX handler algorithm because the
implementation for handling RX Ethernet frames for the DUB-E100 C1 can
have Ethernet frames spanning multiple URBs. This means
From: Nikolay Aleksandrov
There's only one user now and we can include the flag directly.
Signed-off-by: Nikolay Aleksandrov
---
net/bridge/br_vlan.c | 6 ++
1 file changed, 2 insertions(+), 4 deletions(-)
diff --git
From: Nikolay Aleksandrov
When I did the conversion to rhashtable I missed the required locking of
one important user of the vlan list - br_get_link_af_size_filtered()
which is called:
br_ifinfo_notify() -> br_nlmsg_size() -> br_get_link_af_size_filtered()
and the
On 10/02/2015 01:07 PM, Pablo Neira Ayuso wrote:
> On Thu, Oct 01, 2015 at 11:07:30PM +0200, Daniel Mack wrote:
> [...]
>> That, however, got rejected because it doesn't work for multicast. This
>> patch set implements one of the things Pablo suggested in his reply.
>
> People are rising valid
On 10/2/2015 1:48 PM, Neil Armstrong wrote:
To simplify and prevent memory leakage when unbinding, use
the devm_ memory allocation calls.
Tested-by: Andrew Lunn
Tested-by: Florian Fainelli
Signed-off-by: Neil Armstrong
---
On 10/02/2015 03:29 PM, Sergei Shtylyov wrote:
> On 10/2/2015 1:48 PM, Neil Armstrong wrote:
>
>> To simplify and prevent memory leakage when unbinding, use
>> the devm_ memory allocation calls.
>>
>> Tested-by: Andrew Lunn
>> Tested-by: Florian Fainelli
>>
The Data header synchronisation is easier to understand
if the variables "remaining" and "size" are renamed.
Therefore, the lifetime of the "remaining" variable exists
outside of asix_rx_fixup_internal() and is used to indicate
any remaining pending bytes of the Ethernet frame that need
to be
The code is checking that the Ethernet frame will fit into a
netdev allocated socket buffer within the constraints of MTU size,
Ethernet header length plus VLAN header length.
The original code was checking rx->remaining each loop of the while
loop that processes multiple Ethernet frames per URB
On 10/2/2015 4:30 PM, Neil Armstrong wrote:
To simplify and prevent memory leakage when unbinding, use
the devm_ memory allocation calls.
Tested-by: Andrew Lunn
Tested-by: Florian Fainelli
Signed-off-by: Neil Armstrong
---
Without this patch, if the only instructions using r_X are of the
BPF_LD | BPF_IND type, r_X would not be reset to 0, using whatever
value was there when entering the jited code. With this patch, r_X
will be correctly marked as used so it will be reset to 0 in the
prologue code.
This fix also
For ARMv7 with UDIV instruction support, generate an UDIV instruction
followed by an MLS instruction.
For other ARM variants, generate code calling a C wrapper similar to
the jit_udiv() function used for BPF_ALU | BPF_DIV instructions.
Some performance numbers reported by the test_bpf module
From: Nikolay Aleksandrov
Hi,
This is the second follow-up set with one fix (patch 01) and more cleanups
(patches 02,03 and 04). These are minor compared to the previous ones and
should be the last before taking on the optimization changes on the
fast-path.
Cheers,
From: Nikolay Aleksandrov
Introduce br_vlan_(get|put)_master which take a reference (or create the
master vlan first if it didn't exist) and drop a reference respectively.
Signed-off-by: Nikolay Aleksandrov
---
There's one slightly
From: Nikolay Aleksandrov
The checks that lead to num_vlans change are always what
br_vlan_should_use checks for, namely if the vlan is only a context or
not and depending on that it's either not counted or counted
as a real/used vlan respectively.
Also give better
The current ongoing effort to dump existing cBPF seccomp filters back
to user space requires to hold the pre-transformed instructions like
we do in case of socket filters from sk_attach_filter() side, so they
can be reloaded in original form at a later point in time by utilities
such as criu.
To
From: Mark Craske
Please ignore the cover letter PATCH v2 as sent in error.
Patches are all v1, (there are no v2 patches yet)
The ASIX RX handler algorithm is weak on error handling.
There is a design flaw in the ASIX RX handler algorithm because the
implementation for
On Fri, 2 Oct 2015 11:41:18 +0200
Jesper Dangaard Brouer wrote:
> On Thu, 1 Oct 2015 15:10:15 -0700
> Andrew Morton wrote:
>
> > On Wed, 30 Sep 2015 13:44:19 +0200 Jesper Dangaard Brouer
> > wrote:
> >
> > > Make it possible
On Wed, Sep 30, 2015 at 01:26:38PM +0200, Arnd Bergmann wrote:
> The __build_packet_message function fills a nfulnl_msg_packet_timestamp
> structure that uses 64-bit seconds and is therefore y2038 safe, but
> it uses an intermediate 'struct timespec' which is not.
>
> This trivially changes the
On 2015-10-02 12:48, Neil Armstrong wrote:
> To simplify and prevent memory leakage when unbinding, use
> the devm_ memory allocation calls.
>
> Tested-by: Andrew Lunn
> Tested-by: Florian Fainelli
> Signed-off-by: Neil Armstrong
I
On Thu, Oct 01, 2015 at 08:25:46PM -0700, Vinson Lee wrote:
> Hi.
>
> I am seeing a netpoll_send_skb_on_dev warning with bnx2. It happens on
> Linux 4.1 and I am able to reproduce the warning with Linux 4.3-rc3.
>
> [ cut here ]
> WARNING: CPU: 11 PID: 3110 at
When sockets have a native eBPF program attached through
setsockopt(sk, SOL_SOCKET, SO_ATTACH_BPF, ...), and then try to
dump these over getsockopt(sk, SOL_SOCKET, SO_GET_FILTER, ...),
the following panic appears:
[49904.178642] BUG: unable to handle kernel NULL pointer dereference at (null)
If no switch were found in dsa_setup_dst, return -ENODEV and
exit the dsa_probe cleanly.
Tested-by: Andrew Lunn
Tested-by: Florian Fainelli
Signed-off-by: Neil Armstrong
---
net/dsa/dsa.c | 19 +++
1 file changed,
Andrey Vagin writes:
> 2015-10-02 0:05 GMT+03:00 Andrey Vagin :
>> From: Andrey Vagin
>>
>> Now send with MSG_PEEK can return data from multiple SKBs.
>>
>> Unfortunately we take into account the peek offset for each skb,
>> that is wrong.
> Support the IPv6 hw checksum for RTL8111C and later chips. Note
> that the hw has the limitation for the transport offset. The
> checksum must be calculated by sw, when the transport offset is
> out of the range which the hw accepts.
It would be better to implement this check in a
2015-10-02 0:05 GMT+03:00 Andrey Vagin :
> From: Andrey Vagin
>
> Now send with MSG_PEEK can return data from multiple SKBs.
>
> Unfortunately we take into account the peek offset for each skb,
> that is wrong. We need to apply the peek offset only once.
>
>
To simplify and prevent memory leakage when unbinding, use
the devm_ memory allocation calls.
Tested-by: Andrew Lunn
Tested-by: Florian Fainelli
Signed-off-by: Neil Armstrong
---
net/dsa/dsa.c | 6 +++---
1 file changed, 3
When unbinding dsa, complete the dsa_switch_destroy to cleanly
destroy and unregister the net and mdio devices.
Tested-by: Andrew Lunn
Tested-by: Florian Fainelli
Signed-off-by: Neil Armstrong
---
net/dsa/dsa.c | 42
In order to cleanly unbind the dsa core, either as a module removal,
or a platform device unbind, switch the allocation the their devm_
counterparts and complete the destroy functions.
The last patch is an experimental way to exit the probe when no
switch is found in the discover process.
The
When openvswitch tries allocate memory from offline numa node 0:
stats = kmem_cache_alloc_node(flow_stats_cache, GFP_KERNEL | __GFP_ZERO, 0)
It catches VM_BUG_ON(nid < 0 || nid >= MAX_NUMNODES || !node_online(nid))
[ replaced with VM_WARN_ON(!node_online(nid)) recently ] in linux/gfp.h
This patch
On Fri, Oct 02, 2015 at 12:44:03AM +0100, Ian Morris wrote:
>
> On 30/09/15 23:17, Pablo Neira Ayuso wrote:
> >On Thu, Sep 24, 2015 at 11:10:06AM +0100, Ian Morris wrote:
> >>Fixes various whitespace issues detected by checkpatch (e.g. adding spaces
> >>between operations, replacing spaces with
On Thu, Oct 01, 2015 at 11:07:30PM +0200, Daniel Mack wrote:
[...]
> That, however, got rejected because it doesn't work for multicast. This
> patch set implements one of the things Pablo suggested in his reply.
People are rising valid concerns here, so far we got a RFC where you
say that you
From: Richard Alpe
This patch adds man pages for the TIPC tool. There is one main page
and one page for each top level sub-command. These pages mainly aims
to help a user of the tipc tool. In addition to this they describe
a bit about what TIPC is and some of its
> -Original Message-
> From: Wood Scott-B07421
> Sent: Friday, October 02, 2015 4:01 AM
>
> On Thu, Sep 24, 2015 at 06:00:12PM +0300, Madalin Bucur wrote:
> > Introduce managed counterparts for alloc_percpu() and free_percpu().
> > Add devm_alloc_percpu() and devm_free_percpu() into the
Instead of using a custom approach change the code to use %*ph format
specifier.
Signed-off-by: Andy Shevchenko
---
drivers/net/ethernet/intel/i40e/i40e_common.c | 22 --
drivers/net/ethernet/intel/i40evf/i40e_common.c | 22
On Tue, 2015-09-29 at 21:56 -0700, David Miller wrote:
> From: Eric Dumazet
> Date: Tue, 29 Sep 2015 18:52:25 -0700
>
> > From: Eric Dumazet
> >
> > Greg reported crashes hitting the following check in __sk_backlog_rcv()
> >
> >
Support quoting strings with " or ' in an iproute2 batch file.
Enables to configure empty crypto keys (for ESP-null) or keys with
spaces:
xfrm state add src 1.1.1.1 dst 2.2.2.2 proto ah spi 0x1 \
mode tunnel auth hmac(sha1) "r4ezR/@kd6'749f2 6zf$"
xfrm state add src 5.5.5.5 dst
On 10/01/2015 06:32 PM, Andrew Lunn wrote:
> On Thu, Oct 01, 2015 at 05:27:32PM +0200, Neil Armstrong wrote:
>> On 09/30/2015 10:21 AM, Neil Armstrong wrote:
>>> If no switch were found in dsa_setup_dst, return -ENODEV and
>>> exit the dsa_probe cleanly.
>
> ...
>
>> Couldn't we use the probe
On Fri, 2 Oct 2015, Jesper Dangaard Brouer wrote:
> Thus, I need introducing new code like this patch and at the same time
> have to reduce the number of instruction-cache misses/usage. In this
> case we solve the problem by kmem_cache_free_bulk() not getting called
> too often. Thus, +17 bytes
On Fri, 2 Oct 2015 05:10:02 -0500 (CDT)
Christoph Lameter wrote:
> On Fri, 2 Oct 2015, Jesper Dangaard Brouer wrote:
>
> > Thus, I need introducing new code like this patch and at the same time
> > have to reduce the number of instruction-cache misses/usage. In this
> > case we
On Thu, 1 Oct 2015 15:10:15 -0700
Andrew Morton wrote:
> On Wed, 30 Sep 2015 13:44:19 +0200 Jesper Dangaard Brouer
> wrote:
>
> > Make it possible to free a freelist with several objects by adjusting
> > API of slab_free() and __slab_free() to
On 10/02/2015 12:47 PM, Neil Armstrong wrote:
> In order to cleanly unbind the dsa core, either as a module removal,
> or a platform device unbind, switch the allocation the their devm_
> counterparts and complete the destroy functions.
>
> The last patch is an experimental way to exit the probe
On 10/2/15 3:06 AM, Daniel Borkmann wrote:
However, sk_get_filter() wasn't updated to test for this at the
time when eBPF could be attached. Just throw an error to the user
to indicate that eBPF cannot be dumped over this interface.
That way, it can also be known that a program_is_ attached (as
On 10/02/2015 05:09 PM, Alexei Starovoitov wrote:
...
I agree that adding flag to bpf_prog_create_from_user() is cleaner
than exposing static bpf_prog_store_orig_filter(), so
There's also another reason as mentioned, i.e. that the progs
are ro-locked, so doing bpf_prog_store_orig_filter()
Hi all,
Here's v5 of the seccomp filter c/r set. The individual patch notes have
changes, but two highlights are:
* This series is now based on http://patchwork.ozlabs.org/patch/525492/ and
will need to be built with that patch applied. This gets rid of two incorrect
patches in the previous
This patch introduces the concept of a seccomp fd, with a similar interface
and usage to ebpf fds. Initially, one is allowed to create, install, and
dump these fds. Any manipulation of seccomp fds requires users to be root
in their own user namespace, matching the checks done for
This command allows for comparing the filters pointed to by two seccomp
fds. This is useful e.g. to find out if a seccomp filter is inherited,
since struct seccomp_filter are unique across tasks and are the
private_data seccomp fds.
v2: switch to KCMP_SECCOMP_FD instead of KCMP_FILE_PRIVATE_DATA
On Fri, Oct 02, 2015 at 04:37:51PM +0200, Nicolas Schichan wrote:
> @@ -125,7 +125,7 @@ static u64 jit_get_skb_w(struct sk_buff *skb, int offset)
> }
>
> /*
> - * Wrapper that handles both OABI and EABI and assures Thumb2 interworking
> + * Wrappers that handles both OABI and EABI and assures
On Fri, Oct 02, 2015 at 05:06:47PM +0200, Nicolas Schichan wrote:
> For ARMv7 with UDIV instruction support, generate an UDIV instruction
> followed by an MLS instruction.
>
> For other ARM variants, generate code calling a C wrapper similar to
> the jit_udiv() function used for BPF_ALU | BPF_DIV
I just picked 40 for the constant out of thin air, but there may be a more
appropriate value for this. Also, we return EINVAL when there is no filter
for the index the user requested, but ptrace also returns EINVAL for
invalid commands, making it slightly awkward to test whether or not the
kernel
For ARMv7 with UDIV instruction support, generate an UDIV instruction
followed by an MLS instruction.
For other ARM variants, generate code calling a C wrapper similar to
the jit_udiv() function used for BPF_ALU | BPF_DIV instructions.
Some performance numbers reported by the test_bpf module
On 10/02/2015 05:06 PM, Tycho Andersen wrote:
...
Cc: Pavel Emelyanov
Cc: Kees Cook
Cc: Andy Lutomirski
Cc: Alexei Starovoitov
---
This is in realtion to Tycho's latest patch set under [1]. The BPF
On 10/2/15 6:17 AM, Daniel Borkmann wrote:
The current ongoing effort to dump existing cBPF seccomp filters back
to user space requires to hold the pre-transformed instructions like
we do in case of socket filters from sk_attach_filter() side, so they
can be reloaded in original form at a later
Hi Daniel,
On Fri, Oct 02, 2015 at 03:17:33PM +0200, Daniel Borkmann wrote:
> The current ongoing effort to dump existing cBPF seccomp filters back
> to user space requires to hold the pre-transformed instructions like
> we do in case of socket filters from sk_attach_filter() side, so they
> can
On Tue, Sep 22, 2015 at 9:40 PM, Roopa Prabhu wrote:
> + err = fib_nl_fill_rule(skb, rule, NETLINK_CB(cb->skb).portid,
> + cb->nlh->nlmsg_seq, RTM_NEWRULE,
> + NLM_F_MULTI, ops);
> +
On 10/2/15, 10:18 AM, Roland Dreier wrote:
> On Tue, Sep 22, 2015 at 9:40 PM, Roopa Prabhu
> wrote:
>> + err = fib_nl_fill_rule(skb, rule, NETLINK_CB(cb->skb).portid,
>> + cb->nlh->nlmsg_seq, RTM_NEWRULE,
>> +
TCP listener refactoring : this is becoming interesting !
This patch series takes the steps to use normal TCP/DCCP ehash
table to store SYN_RECV requests, instead of the private per-listener
hash table we had until now.
SYNACK skb are now attached to their syn_recv request socket,
so that we no
long term plan is to remove struct listen_sock when its hash
table is no longer there.
Signed-off-by: Eric Dumazet
---
include/net/request_sock.h | 2 +-
net/ipv4/tcp_input.c | 7 +++
2 files changed, 4 insertions(+), 5 deletions(-)
diff --git
This control variable was set at first listen(fd, backlog)
call, but not updated if application tried to increase or decrease
backlog. It made sense at the time listener had a non resizeable
hash table.
Also rounding to powers of two was not very friendly.
Signed-off-by: Eric Dumazet
If a listen backlog is very big (to avoid syncookies), then
the listener sk->sk_wmem_alloc is the main source of false
sharing, as we need to touch it twice per SYNACK re-transmit
and TX completion.
(One SYN packet takes listener lock once, but up to 6 SYNACK
are generated)
By attaching the skb
It is enough to check listener sk_state, no need for an extra
condition.
max_qlen_log can be moved into struct request_sock_queue
We can remove syn_wait_lock and the alignment it enforced.
Signed-off-by: Eric Dumazet
---
include/net/request_sock.h | 26
Commit ea317b267e9d ("bpf: Add new bpf map type to store the pointer
to struct perf_event") added perf_event.h to the main eBPF header, so
it gets included for all users. perf_event.h is actually only needed
from array map side, so lets sanitize this a bit.
Signed-off-by: Daniel Borkmann
David Ahern writes:
> Add support for filtering neighbor dumps by master device by adding
> the NDA_MASTER attribute to the dump request. A new netlink flag,
> NLM_F_DUMP_FILTERED, is added to indicate the kernel supports the
> request and output is filtered as
On Wed, Sep 30, 2015 at 8:32 PM, Thomas F Herbert
wrote:
> Although the Open Flow specification specified support for 802.1AD (qinq)
> as well as push and pop vlan headers, So far Open vSwitch has only
> supported a single tag header. This patch implements 802.1AD in
Add support for filtering neighbor dumps by master device. Kernel side
support provided by commit 21fdd092acc7. Since the feature is not
available in older kernels the user is given a warning message if the
kernel does not support the request.
Signed-off-by: David Ahern
inet6_csk_search_req() and inet6_csk_reqsk_queue_hash_add()
no longer exist.
Signed-off-by: Eric Dumazet
---
include/net/inet6_connection_sock.h | 9 -
1 file changed, 9 deletions(-)
diff --git a/include/net/inet6_connection_sock.h
On 10/2/15 9:42 AM, Daniel Borkmann wrote:
Commit ea317b267e9d ("bpf: Add new bpf map type to store the pointer
to struct perf_event") added perf_event.h to the main eBPF header, so
it gets included for all users. perf_event.h is actually only needed
from array map side, so lets sanitize this a
On Thu, Oct 01, 2015 at 08:29:19AM -0400, Jon Ringle wrote:
> On Thu, 1 Oct 2015, Mark Brown wrote:
> > This completely bypasses and therefore breaks the cache infrastructure.
> Right after sending the v2 patch, I realized that calling the
> custom reg_update_bits would only be applicable for
On 02/10/15 05:10, Neil Armstrong wrote:
> On 10/01/2015 06:32 PM, Andrew Lunn wrote:
>> On Thu, Oct 01, 2015 at 05:27:32PM +0200, Neil Armstrong wrote:
>>> On 09/30/2015 10:21 AM, Neil Armstrong wrote:
If no switch were found in dsa_setup_dst, return -ENODEV and
exit the dsa_probe
On Fri, Oct 02, 2015 at 11:01:45AM +0300, Denys Fedoryshchenko wrote:
> Here is similar panic after patch applied (it might be different bug), got
> over netconsole:
>
> [126348.617115] CPU: 0 PID: 5254 Comm: accel-pppd Not tainted
> 4.2.2-build-0087 #2
> [126348.617632] Hardware name: Intel
struct request_sock_queue fields are currently protected
by the listener 'lock' (not a real spinlock)
We need to add a private spinlock instead, so that softirq handlers
creating children do not have to worry with backlog notion
that the listener 'lock' carries.
Signed-off-by: Eric Dumazet
Everything should now be ready to finally allow SYN
packets processing without holding listener lock.
Tested:
3.5 Mpps SYNFLOOD. Plenty of cpu cycles available.
Next bottleneck is the refcount taken on listener,
that could be avoided if we remove SLAB_DESTROY_BY_RCU
strict semantic for
If a listener with thousands of children in accept queue
is dismantled, it can take a while to close all of them.
Signed-off-by: Eric Dumazet
---
net/ipv4/inet_connection_sock.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/net/ipv4/inet_connection_sock.c
From: Jason Baron
The unix_dgram_poll() routine calls sock_poll_wait() not only for the wait
queue associated with the socket s that we've called poll() on, but it also
calls sock_poll_wait() for a remote peer socket's wait queue, if it's connected.
Thus, if we call
On Thu, Oct 1, 2015 at 11:00 PM, Jiri Benc wrote:
> On Thu, 1 Oct 2015 17:11:56 -0700, Pravin Shelar wrote:
>> I dont see point of adding this code when IPv6 sampling not support by
>> the patch series.
>
> It was requested by Jesse:
>
Rainer Weikusat writes:
> Jason Baron writes:
>> From: Jason Baron
>>
>> The unix_dgram_poll() routine calls sock_poll_wait() not only for the wait
>> queue associated with the socket s that we've called poll() on,
Jason Baron writes:
> On 10/02/2015 03:30 PM, Rainer Weikusat wrote:
>> Jason Baron writes:
>>> From: Jason Baron
>>>
>>> The unix_dgram_poll() routine calls sock_poll_wait() not only for the wait
>>> queue associated with the socket s
In this patch, we insert request sockets into TCP/DCCP
regular ehash table (where ESTABLISHED and TIMEWAIT sockets
are) instead of using the per listener hash table.
ACK packets find SYN_RECV pseudo sockets without having
to find and lock the listener.
In nominal conditions, this halves pressure
Am Freitag, 2. Oktober 2015, 09:17:16 schrieb Holger Hoffstätte:
> On 10/02/15 08:52, Andre Tomt wrote:
> > On 01. okt. 2015 13:52, Eric Dumazet wrote:
> >> On Thu, Oct 1, 2015 at 4:43 AM, Holger Hoffstätte
> >>
> >> wrote:
> >>> On 10/01/15 13:29, Eric Dumazet
Jason Baron writes:
> From: Jason Baron
>
> The unix_dgram_poll() routine calls sock_poll_wait() not only for the wait
> queue associated with the socket s that we've called poll() on, but it also
> calls sock_poll_wait() for a remote peer socket's wait
We'll soon have to call tcp_v[46]_inbound_md5_hash() twice.
Also add const attribute to the socket, as it might be the
unlocked listener for SYN packets.
Signed-off-by: Eric Dumazet
---
net/ipv4/tcp_ipv4.c | 16 ++--
net/ipv6/tcp_ipv6.c | 10 ++
2 files
We no longer use hash_rnd, nr_table_entries and syn_table[]
For a listener with a backlog of 10 millions sockets, this
saves 80 MBytes of vmalloced memory.
Signed-off-by: Eric Dumazet
---
include/net/request_sock.h | 3 ---
net/core/request_sock.c| 14 +++---
On 10/02/2015 03:30 PM, Rainer Weikusat wrote:
> Jason Baron writes:
>> From: Jason Baron
>>
>> The unix_dgram_poll() routine calls sock_poll_wait() not only for the wait
>> queue associated with the socket s that we've called poll() on, but it also
>> calls
On 10/2/15 11:18 AM, Eric W. Biederman wrote:
What is the thinking here because it sure looks like you are busily
adding layer two functionality you swore you did not want.
Interfaces are enslaved to a VRF device, but neighbor entries are
installed with a reference to the actual interface not
On Thu, 1 Oct 2015, Eric Dumazet wrote:
> On Thu, Oct 1, 2015 at 4:43 AM, Holger Hoffstätte
> wrote:
> > On 10/01/15 13:29, Eric Dumazet wrote:
>
> >> commit 83fccfc3940c4a2db90fd7e7079f5b465cd8c6af
> >> Author: Eric Dumazet
> >> Date:
When request sockets are no longer in a per listener hash table
but on regular TCP ehash, we need to access listener uid
through req->rsk_listener
get_openreq6() also gets a const for its request socket argument.
Signed-off-by: Eric Dumazet
---
include/net/tcp.h | 1 -
Once listener is lockless, its sk_state can change anytime.
Signed-off-by: Eric Dumazet
---
net/ipv4/tcp_minisocks.c | 2 --
1 file changed, 2 deletions(-)
diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c
index 897e34273ba3..9adf1e2c3170 100644
---
qlen_inc & young_inc were protected by listener lock,
while qlen_dec & young_dec were atomic fields.
Everything needs to be atomic for upcoming lockless listener.
Also move qlen/young in request_sock_queue as we'll get rid
of struct listen_sock eventually.
Signed-off-by: Eric Dumazet
We plan to use generic functions to insert request sockets
into ehash table.
sk_prot needs to be set (to retrieve sk_prot->h.hashinfo)
sk_node needs to be cleared.
Signed-off-by: Eric Dumazet
---
include/net/request_sock.h | 22 --
1 file changed, 12
This is no longer used.
Signed-off-by: Eric Dumazet
---
include/net/inet_connection_sock.h | 3 +--
net/ipv4/inet_connection_sock.c| 2 +-
net/ipv6/inet6_connection_sock.c | 2 +-
3 files changed, 3 insertions(+), 4 deletions(-)
diff --git
This fixes a typo : We want to store the NAPI id on child socket.
Presumably nobody really uses busy polling, on short lived flows.
Fixes: 3d97379a67486 ("tcp: move sk_mark_napi_id() at the right place")
Signed-off-by: Eric Dumazet
---
net/ipv4/tcp_ipv4.c | 2 +-
Hi,
These patches are against mainline, I can re-base to net-next, just
let me know.
They have been tested against: https://lkml.org/lkml/2015/9/13/195,
which causes the use-after-free quite quickly and here:
https://lkml.org/lkml/2015/10/2/693.
Thanks,
-Jason
Jason Baron (3):
unix: fix
1 - 100 of 135 matches
Mail list logo