From: Taehee Yoo
Date: Sun, 8 Jul 2018 11:55:51 +0900
> @@ -1143,13 +1143,14 @@ void rhashtable_free_and_destroy(struct rhashtable
> *ht,
>void (*free_fn)(void *ptr, void *arg),
>void *arg)
> {
> - struct bucket_table *tbl;
>
From: Alexei Starovoitov
Date: Sat, 7 Jul 2018 20:58:26 -0700
> The following pull-request contains BPF updates for your *net* tree.
>
> Plenty of fixes for different components:
>
> 1) A set of critical fixes for sockmap and sockhash, from John Fastabend.
>
> 2) fixes for several race
From: Paolo Abeni
Date: Fri, 6 Jul 2018 12:30:20 +0200
> Setting the low threshold to 0 has no effect on frags allocation,
> we need to clear high_thresh instead.
>
> The code was pre-existent to commit 648700f76b03 ("inet: frags:
> use rhashtables for reassembly units"), but before the above,
From: Li RongQing
Date: Fri, 6 Jul 2018 17:33:36 +0800
> The return of num_possible_cpus() is same as nr_cpu_ids, but
> nr_cpu_ids can reduce cpu computation
>
> Signed-off-by: Li RongQing
These two numbers are not the same, in fact num_possible_cpus() can be
and often is smaller than
Hi David,
The following pull-request contains BPF updates for your *net* tree.
Plenty of fixes for different components:
1) A set of critical fixes for sockmap and sockhash, from John Fastabend.
2) fixes for several race conditions in af_xdp, from Magnus Karlsson.
3) hash map refcnt fix, from
From: Vlad Buslov
Date: Thu, 5 Jul 2018 17:24:22 +0300
> Currently, all netlink protocol handlers for updating rules, actions and
> qdiscs are protected with single global rtnl lock which removes any
> possibility for parallelism. This patch set is a first step to remove
> rtnl lock dependency
rhashtable_free_and_destroy() cancels re-hash deferred work
then walks and destroys elements. at this moment, some elements can be
still in future_tbl. that elements are not destroyed.
test case:
nft_rhash_destroy() calls rhashtable_free_and_destroy() to destroy
all elements of sets before
From: Yifeng Sun
Date: Mon, 2 Jul 2018 08:18:03 -0700
> Add 'clone' action to kernel datapath by using existing functions.
> When actions within clone don't modify the current flow, the flow
> key is not cloned before executing clone actions.
>
> This is a follow up patch for this incomplete
From: Randy Dunlap
Date: Sat, 7 Jul 2018 08:31:15 -0700
> From: Randy Dunlap
>
> Fix build warnings in drivers/isdn/capi/ when CONFIG_PROC_FS is not
> enabled by marking the unused functions as __maybe_unused.
>
> ../drivers/isdn/capi/capi.c:1324:12: warning: 'capi20_proc_show' defined but
>
From: Randy Dunlap
Date: Sat, 7 Jul 2018 08:27:53 -0700
> From: Randy Dunlap
>
> Fix a build warning in connector.c when CONFIG_PROC_FS is not enabled
> by marking the unused function as __maybe_unused.
>
> ../drivers/connector/connector.c:242:12: warning: 'cn_proc_show' defined but
> not
From: Lorenzo Colitti
Date: Sat, 7 Jul 2018 16:31:40 +0900
> When tcp_diag_destroy closes a TCP_NEW_SYN_RECV socket, it first
> frees it by calling inet_csk_reqsk_queue_drop_and_and_put in
> tcp_abort, and then frees it again by calling sock_gen_put.
>
> Since tcp_abort only has one caller,
From: dsah...@kernel.org
Date: Sat, 7 Jul 2018 16:15:26 -0700
> From: David Ahern
>
> Xin reported that icmp replies may not use the address on the device the
> echo request is received if the destination address is broadcast. Instead
> a route lookup is done without considering VRF context.
From: David Ahern
Date: Sat, 7 Jul 2018 19:23:16 -0600
> On 7/7/18 7:14 AM, Ido Schimmel wrote:
>> On Sat, Jul 07, 2018 at 08:11:16PM +0900, David Miller wrote:
>>> Chas, it seems to me that you add the new notifier by not even one
>>> driver is listening for the event.
>>>
>>> Either it is
From: Eric Dumazet
Date: Sat, 7 Jul 2018 17:22:01 -0700
> Maybe gro_count should be replaced by a bitmask, so that we can
> speed up napi_gro_flush(), since it now has to use 3 cache lines
> (gro_hash[] size is 192 bytes)
Something like this?
diff --git a/include/linux/netdevice.h
On 7/7/18 7:14 AM, Ido Schimmel wrote:
> On Sat, Jul 07, 2018 at 08:11:16PM +0900, David Miller wrote:
>> Chas, it seems to me that you add the new notifier by not even one
>> driver is listening for the event.
>>
>> Either it is necessary, and you should show at least one example
>> use case, or
On 07/03/2018 03:42 PM, Jesus Sanchez-Palencia wrote:
> From: Richard Cochran
>
> This patch introduces SO_TXTIME. User space enables this option in
> order to pass a desired future transmit time in a CMSG when calling
> sendmsg(2). The argument to this socket option is a 8-bytes long struct
On 07/05/2018 03:20 AM, David Miller wrote:
> From: Li RongQing
> Date: Thu, 5 Jul 2018 14:34:32 +0800
>
>> After commit 07d78363dcff ("net: Convert NAPI gro list into a small hash
>> table.")' there is 8 hash buckets, which allows more flows to be held for
>> merging. but MAX_GRO_SKBS, the
From: David Ahern
Xin reported that icmp replies may not use the address on the device the
echo request is received if the destination address is broadcast. Instead
a route lookup is done without considering VRF context. Fix by setting
oif in flow struct to the master device if it is enslaved.
On Fri, Jul 06, 2018 at 11:49:00AM +0900, Toshiaki Makita wrote:
> Otherwise we end up with attempting to send packets from down devices
> or to send oversized packets, which may cause unexpected driver/device
> behaviour. Generic XDP has already done this check, so reuse the logic
> in native
On Thu, Jul 05, 2018 at 08:49:54AM -0700, John Fastabend wrote:
> First three patches resolve issues found while testing sockhash and
> reviewing code. Syzbot also found them about the same time as I was
> working on fixes. The main issue is in the sockhash path we reduced
> the scope of
On Thu, Jul 05, 2018 at 08:05:51AM -0700, John Fastabend wrote:
> I missed fixing the error path in the sockhash code to align with
> supporting socks in multiple maps. Simply checking if the psock is
> present does not mean we can decrement the reference count because
> it could be part of
On 7/7/18 1:31 AM, Lorenzo Colitti wrote:
> When tcp_diag_destroy closes a TCP_NEW_SYN_RECV socket, it first
> frees it by calling inet_csk_reqsk_queue_drop_and_and_put in
> tcp_abort, and then frees it again by calling sock_gen_put.
>
> Since tcp_abort only has one caller, and all the other
On Wed, 4 Jul 2018 12:32:33 +0900
fumihiko kakuma wrote:
> A commandline like 'tc -d class show dev dev-name' does not
> display value of prio and quantum option when we use htb qdisc.
> This patch fixes the bug.
>
> Signed-off-by: Fumihiko Kakuma
Applied
On Tue, 3 Jul 2018 15:54:32 +0300
Roi Dayan wrote:
> Example output is of tos and ttl.
> Befoe this fix the format used %x caused output of the pointer
> instead of the intended string created in the out variable.
>
> Fixes: e28b88a464c4 ("tc: jsonify flower filter")
> Signed-off-by: Roi Dayan
On Thu, 5 Jul 2018 08:20:09 -0700
Jesus Sanchez-Palencia wrote:
> Fix 2 typos on the man page of the CBS qdisc.
>
> Signed-off-by: Jesus Sanchez-Palencia
> Reviewed-by: Simon Horman
Applied
On Sat, 7 Jul 2018 12:40:01 -0400
Qiaobin Fu wrote:
> The new action inheritdsfield copies the field DS of
> IPv4 and IPv6 packets into skb->priority. This enables
> later classification of packets based on the DS field.
>
> v2:
> *Use optional flags, so that it won't break old versions of tc.
The new action inheritdsfield copies the field DS of
IPv4 and IPv6 packets into skb->priority. This enables
later classification of packets based on the DS field.
v2:
*Use optional flags, so that it won't break old versions of tc.
Original idea by Jamal Hadi Salim
Signed-off-by: Qiaobin Fu
On Sat, Jul 7, 2018 at 4:23 AM, David Miller wrote:
> From: Roopa Prabhu
> Date: Wed, 4 Jul 2018 16:46:28 -0700
>
>> From: Roopa Prabhu
>>
>> Problem:
>> In vxlan_newlink, a default fdb entry is added before register_netdev.
>> The default fdb creation function notifies user-space of the
>>
On Wed, Jul 04, 2018 at 08:33:20PM +0200, Linus Walleij wrote:
> The code to calculate the hardware register enumerator
> for the maximum L3 length isn't entirely simple to read.
> Use the existing defines and rewrite the function into a
> table look-up.
A matter of habit. ;-)
I think that if you
From: Randy Dunlap
Fix build warnings in drivers/isdn/capi/ when CONFIG_PROC_FS is not
enabled by marking the unused functions as __maybe_unused.
../drivers/isdn/capi/capi.c:1324:12: warning: 'capi20_proc_show' defined but
not used [-Wunused-function]
../drivers/isdn/capi/capi.c:1347:12:
From: Randy Dunlap
Fix a build warning in connector.c when CONFIG_PROC_FS is not enabled
by marking the unused function as __maybe_unused.
../drivers/connector/connector.c:242:12: warning: 'cn_proc_show' defined but
not used [-Wunused-function]
Signed-off-by: Randy Dunlap
Cc: Evgeniy
On 7/6/18 3:50 AM, Xin Long wrote:
>>> Try this:
>>>
>>> diff --git a/net/ipv4/fib_frontend.c b/net/ipv4/fib_frontend.c
>>> index b21833651394..e46cdd310e5f 100644
>>> --- a/net/ipv4/fib_frontend.c
>>> +++ b/net/ipv4/fib_frontend.c
>>> @@ -300,6 +300,7 @@ __be32 fib_compute_spec_dst(struct sk_buff
On 07/07/2018 07:16 AM, Eric Dumazet wrote:
>
>
> On 07/07/2018 06:56 AM, David Ahern wrote:
>
>>
>> So I have to either learn how to use packetdrill or install a web server
>> and put load on it. If the Android tests are not publicly available then
>> the reference should be removed from
On 07/07/2018 06:56 AM, David Ahern wrote:
>
> So I have to either learn how to use packetdrill or install a web server
> and put load on it. If the Android tests are not publicly available then
> the reference should be removed from the commit log.
We see many Change-Id tags in changelogs,
On Sat, Jul 7, 2018 at 7:15 AM David Miller wrote:
>
> From: Lawrence Brakmo
> Date: Tue, 3 Jul 2018 09:26:13 -0700
>
> > When have observed high tail latencies when using DCTCP for RPCs as
> > compared to using Cubic. For example, in one setup there are 2 hosts
> > sending to a 3rd one, with
On 07/07/2018 06:29 AM, Eric Dumazet wrote:
>
>
> On 07/07/2018 06:11 AM, David Miller wrote:
>> From: Lorenzo Colitti
>> Date: Sat, 7 Jul 2018 16:31:40 +0900
>>
>>> Tested: passes Android sock_diag_test.py, which exercises this codepath
>>
>> If this Android test case exercises this path,
On 7/7/18 7:51 AM, Eric Dumazet wrote:
>
>
> On 07/07/2018 06:45 AM, Eric Dumazet wrote:
>>
>>
>> On 07/07/2018 06:33 AM, David Ahern wrote:
>>> On 7/7/18 7:11 AM, David Miller wrote:
From: Lorenzo Colitti
Date: Sat, 7 Jul 2018 16:31:40 +0900
> Tested: passes Android
On 07/07/2018 06:45 AM, Eric Dumazet wrote:
>
>
> On 07/07/2018 06:33 AM, David Ahern wrote:
>> On 7/7/18 7:11 AM, David Miller wrote:
>>> From: Lorenzo Colitti
>>> Date: Sat, 7 Jul 2018 16:31:40 +0900
>>>
Tested: passes Android sock_diag_test.py, which exercises this codepath
>>>
>>>
On 07/07/2018 06:33 AM, David Ahern wrote:
> On 7/7/18 7:11 AM, David Miller wrote:
>> From: Lorenzo Colitti
>> Date: Sat, 7 Jul 2018 16:31:40 +0900
>>
>>> Tested: passes Android sock_diag_test.py, which exercises this codepath
>>
>> If this Android test case exercises this path, why didn't
On 7/7/18 7:11 AM, David Miller wrote:
> From: Lorenzo Colitti
> Date: Sat, 7 Jul 2018 16:31:40 +0900
>
>> Tested: passes Android sock_diag_test.py, which exercises this codepath
>
> If this Android test case exercises this path, why didn't it trigger
> the double free and thus cause this bug
On 07/07/2018 06:11 AM, David Miller wrote:
> From: Lorenzo Colitti
> Date: Sat, 7 Jul 2018 16:31:40 +0900
>
>> Tested: passes Android sock_diag_test.py, which exercises this codepath
>
> If this Android test case exercises this path, why didn't it trigger
> the double free and thus cause
On 07/07/2018 12:31 AM, Lorenzo Colitti wrote:
> When tcp_diag_destroy closes a TCP_NEW_SYN_RECV socket, it first
> frees it by calling inet_csk_reqsk_queue_drop_and_and_put in
> tcp_abort, and then frees it again by calling sock_gen_put.
>
> Since tcp_abort only has one caller, and all the
On Sat, Jul 07, 2018 at 08:11:16PM +0900, David Miller wrote:
> Chas, it seems to me that you add the new notifier by not even one
> driver is listening for the event.
>
> Either it is necessary, and you should show at least one example
> use case, or it not necessary and therefore should not be
From: Lorenzo Colitti
Date: Sat, 7 Jul 2018 16:31:40 +0900
> Tested: passes Android sock_diag_test.py, which exercises this codepath
If this Android test case exercises this path, why didn't it trigger
the double free and thus cause this bug to be found much sooner?
Just curious.
From: Davide Caratti
Date: Fri, 6 Jul 2018 21:01:04 +0200
> in a couple of TC actions (i.e. csum and tunnel_key), the control action
> is stored together with the action-specific configuration data.
> This avoids a race condition (see [1]), but it causes a crash when 'goto
> chain' is used with
From: Georg Kohmann
Date: Fri, 6 Jul 2018 14:59:56 +0200
> @@ -92,7 +92,7 @@ static void icmpv6_err(struct sk_buff *skb, struct
> inet6_skb_parm *opt,
> struct net *net = dev_net(skb->dev);
>
> if (type == ICMPV6_PKT_TOOBIG)
> - ip6_update_pmtu(skb, net, info, 0,
From: Jianbo Liu
Date: Fri, 6 Jul 2018 05:38:11 +
> Currently TC flower supports only one vlan tag, it doesn't match on both outer
> and inner vlan headers for QinQ. To do this, we add support to get both outer
> and inner vlan headers for flow dissector, and then TC flower do matching on
>
From: Vlad Buslov
Date: Thu, 5 Jul 2018 17:24:22 +0300
> Currently, all netlink protocol handlers for updating rules, actions and
> qdiscs are protected with single global rtnl lock which removes any
> possibility for parallelism. This patch set is a first step to remove
> rtnl lock dependency
From: Petr Machata
Date: Thu, 05 Jul 2018 21:10:53 +0200
> Some configurations of mirror-to-gretap are impossible for mlxsw to
> offload. Add a test that checks that these out-of-domain conditions are
> handled properly by mlxsw.
>
> In patch #1, fix mirror_gre_lib.sh and mirror_gre_topo_lib.sh
From: Roopa Prabhu
Date: Wed, 4 Jul 2018 16:46:28 -0700
> From: Roopa Prabhu
>
> Problem:
> In vxlan_newlink, a default fdb entry is added before register_netdev.
> The default fdb creation function notifies user-space of the
> fdb entry on the vxlan device which user-space does not know
From: Lawrence Brakmo
Date: Tue, 3 Jul 2018 09:26:13 -0700
> When have observed high tail latencies when using DCTCP for RPCs as
> compared to using Cubic. For example, in one setup there are 2 hosts
> sending to a 3rd one, with each sender having 3 flows (1 stream,
> 1 1MB back-to-back RPCs and
From: Chas Williams <3ch...@gmail.com>
Date: Mon, 2 Jul 2018 18:35:28 -0400
> vlan_changelink silently ignores attempts to change the vlan id
> or protocol id of an existing vlan interface. Implement by adding
> the new vlan id and protocol to the interface's vlan group and then
> removing the
net/sched: add skbprio scheduer
Skbprio (SKB Priority Queue) is a queueing discipline that prioritizes packets
according to their skb->priority field. Under congestion, already-enqueued lower
priority packets will be dropped to make space available for higher priority
packets. Skbprio was
On Fri, Jul 6, 2018 at 7:24 PM Eric Dumazet wrote:
>
> Your patch makes sense to me, please submit it formally with :
>
> Fixes: d7226c7a4dd1 ("net: diag: Fix refcnt leak in error path destroying
> socket")
> Cc: David Ahern
Submitted a patch against net:
When tcp_diag_destroy closes a TCP_NEW_SYN_RECV socket, it first
frees it by calling inet_csk_reqsk_queue_drop_and_and_put in
tcp_abort, and then frees it again by calling sock_gen_put.
Since tcp_abort only has one caller, and all the other codepaths
in tcp_abort don't free the socket, just
From: Casey Leedom
t4_get_flash_params() fails in a fatal fashion if the FLASH part isn't
one of the recognized parts. But this leads to desperate efforts to update
drivers when various FLASH parts which we are using suddenly become
unavailable and we need to substitute new FLASH parts. This
56 matches
Mail list logo
