When many parallel flows are present and being received on the same
RX queue, GRO processing can become expensive because each incoming
frame must traverse the per-NAPI GRO list at each protocol layer
of GRO receive (eth --> ipv{4,6} --> tcp).
Use the already computed hash to chain these SKBs in
Manage pending per-NAPI GRO packets via list_head.
Return an SKB pointer from the GRO receive handlers. When GRO receive
handlers return non-NULL, it means that this SKB needs to be completed
at this time and removed from the NAPI queue.
Several operations are greatly simplified by this transf
Improve the performance of GRO receive by splitting flows into
multiple hash chains.
Suggested-by: Eric Dumazet
Signed-off-by: David S. Miller
---
include/linux/netdevice.h | 3 +-
net/core/dev.c| 105 --
2 files changed, 81 insertions(+), 27
From: Ganesh Goudar
Date: Sat, 23 Jun 2018 20:28:26 +0530
> When we are disabling DCB, store "0" in txq->dcb_prio
> since that's used for future TX Work Request "OVLAN_IDX"
> values. Setting non zero priority upon disabling DCB
> would halt the traffic.
>
> Reported-by: AMG Zollner Robert
> CC:
From: "Jason A. Donenfeld"
Date: Sat, 23 Jun 2018 17:59:30 +0200
> Two rules with different values of suppress_prefix or suppress_ifgroup
> are not the same. This fixes an -EEXIST when running:
>
>$ ip -4 rule add table main suppress_prefixlength 0
>
> Signed-off-by: Jason A. Donenfeld
> F
From: Fabio Estevam
When platform_device_register_simple() fails we can return
the error immediately instead of jumping to the 'err_pdev'
label.
This makes the error path a bit simpler.
Signed-off-by: Fabio Estevam
---
drivers/net/phy/fixed_phy.c | 7 ++-
1 file changed, 2 insertions(+),
On Sat, Jun 23, 2018 at 1:47 PM, Nishanth Devarajan wrote:
> net/sched: add skbprio scheduler
>
> Skbprio (SKB Priority Queue) is a queueing discipline that prioritizes packets
> according to their skb->priority field. Although Skbprio can be employed in
> any
> scenario in which a higher skb->pr
On Sat, Jun 23, 2018 at 1:47 PM, Nishanth Devarajan wrote:
> diff --git a/include/uapi/linux/pkt_sched.h b/include/uapi/linux/pkt_sched.h
> index 37b5096..6fd07e8 100644
> --- a/include/uapi/linux/pkt_sched.h
> +++ b/include/uapi/linux/pkt_sched.h
...
> +#define SKBPRIO_MAX_PRIORITY 64
> +
> +stru
net/sched: add skbprio scheduler
Skbprio (SKB Priority Queue) is a queueing discipline that prioritizes packets
according to their skb->priority field. Although Skbprio can be employed in any
scenario in which a higher skb->priority field means a higher priority packet,
Skbprio was concieved as a
After commit a09ceb0e0814 ("sched: remove qdisc->drop"),
it is no longer used.
Cc: Florian Westphal
Signed-off-by: Cong Wang
---
net/sched/sch_htb.c | 13 -
1 file changed, 13 deletions(-)
diff --git a/net/sched/sch_htb.c b/net/sched/sch_htb.c
index 2a4ab7caf553..43c4bfe625a9 10064
Two rules with different values of suppress_prefix or suppress_ifgroup
are not the same. This fixes an -EEXIST when running:
$ ip -4 rule add table main suppress_prefixlength 0
Signed-off-by: Jason A. Donenfeld
Fixes: f9d4b0c1e969 ("fib_rules: move common handling of newrule delrule msgs
int
Hey Roopa,
On a kernel with a minimal networking config,
CONFIG_IP_MULTIPLE_TABLES appears to be broken for certain rules after
f9d4b0c1e9695e3de7af3768205bacc27312320c.
Try, for example, running:
$ ip -4 rule add table main suppress_prefixlength 0
It returns with EEXIST.
Perhaps the reason is
On 6/22/2018 9:07 PM, Jakub Kicinski wrote:
On Fri, 22 Jun 2018 17:31:37 -0700, Shannon Nelson wrote:
Implement the IPsec/XFRM offload API for testing.
Signed-off-by: Shannon Nelson
Thanks for the patch! Just a number of stylistic nit picks.
Thanks for the comments, I'll do a v2 in a coup
When we are disabling DCB, store "0" in txq->dcb_prio
since that's used for future TX Work Request "OVLAN_IDX"
values. Setting non zero priority upon disabling DCB
would halt the traffic.
Reported-by: AMG Zollner Robert
CC: David Ahern
Signed-off-by: Casey Leedom
Signed-off-by: Ganesh Goudar
-
From: Heiner Kallweit
Date: Sat, 23 Jun 2018 09:49:37 +0200
> This patch series enables ASPM for the RTL8168E-VL and aligns ASPM entry
> latency handling with the vendor driver before.
Looks good to me, series applied, thank you.
From: Paul Burton
Date: Fri, 22 Jun 2018 20:17:39 -0700
> This series begins the process of cleaning up the pch_gbe network
> driver. Whilst my ultimate goal is to add support for using this driver
> on the MIPS Boston development board, this series sets that aside in
> favor of making some more
Merge window is over so new bpf-next development round begins. Due to travel
this whole week till this Sun mostly offline, but feel free to push your
patches out already so they land in patchwork meanwhile.
The r8168 vendor driver always uses value 0x27. In r8169 we have few
chips where 0x17 is used. So far this didn't matter because ASPM was
disabled anyway. Now that ASPM was re-enabled let's also use 0x27 only.
One of the chips affected by this change is RTL8168E-VL, on my system
with this chip val
Let's enable ASPM also on the RTL8168E-VL (chip version 34).
Works fine on my Zotac Mini PC with this chip. Temperature when
being idle is significantly lower than before due to reaching
deeper PC states.
Signed-off-by: Heiner Kallweit
---
drivers/net/ethernet/realtek/r8169.c | 2 ++
1 file chan
This patch series enables ASPM for the RTL8168E-VL and aligns ASPM entry
latency handling with the vendor driver before.
Heiner Kallweit (2):
r8169: align ASPM entry latency setting with vendor driver
r8169: enable ASPM on RTL8168E-VL
drivers/net/ethernet/realtek/r8169.c | 45 +--
On Fri, Jun 22, 2018 at 08:21:44AM -0700, John Fastabend wrote:
> First in tcp_close, reduce scope of sk_callback_lock() the lock is
> only needed for protecting maps list the ingress and cork
> lists are protected by sock lock. Having the lock in wider scope is
> harmless but may confuse the reade
On Fri, Jun 22, 2018 at 08:21:39AM -0700, John Fastabend wrote:
> If a hashmap is free'd with open socks it removes the reference to
> the hash entry from the psock. If that is the last reference to the
> psock then it will also be free'd by the reference counting logic.
> However the current logic
On Fri, Jun 22, 2018 at 08:21:34AM -0700, John Fastabend wrote:
> This fixes a crash where we assign tcp_prot to IPv6 sockets instead
> of tcpv6_prot.
>
> Previously we overwrote the sk->prot field with tcp_prot even in the
> AF_INET6 case. This patch ensures the correct tcp_prot and tcpv6_prot
>
23 matches
Mail list logo
