- 18 May, 2017 9 commits
-
-
Jason Wang authored
We used to dequeue one skb during recvmsg() from skb_array, this could be inefficient because of the bad cache utilization and spinlock touching for each packet. This patch tries to batch them by calling batch dequeuing helpers explicitly on the exported skb array and pass the skb back through msg_control for underlayer socket to finish the userspace copying. Batch dequeuing is also the requirement for more batching improvement on receive path. Tests were done by pktgen on tap with XDP1 in guest. Host is Intel(R) Xeon(R) CPU E5-2650 0 @ 2.00GHz. rx batch | pps 0 2.25Mpps 1 2.33Mpps (+3.56%) 4 2.33Mpps (+3.56%) 16 2.35Mpps (+4.44%) 64 2.42Mpps (+7.56%) <- Default rx batching 128 2.40Mpps (+6.67%) 256 2.38Mpps (+5.78%) Signed-off-by:
Jason Wang <jasowang@redhat.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Jason Wang authored
This patch makes tap_recvmsg() can receive from skb from its caller through msg_control. Vhost_net will be the first user. Signed-off-by:
-
Jason Wang authored
This patch makes tun_recvmsg() can receive from skb from its caller through msg_control. Vhost_net will be the first user. Signed-off-by:
-
Jason Wang authored
This patch exports skb_array through tap_get_skb_array(). Caller can then manipulate skb array directly. Signed-off-by:
-
Jason Wang authored
This patch exports skb_array through tun_get_skb_array(). Caller can then manipulate skb array directly. Signed-off-by:
-
Jason Wang authored
Signed-off-by:
-
Jason Wang authored
This patch introduce a batched version of consuming, consumer can dequeue more than one pointers from the ring at a time. We don't care about the reorder of reading here so no need for compiler barrier. Signed-off-by:
-
Jason Wang authored
Signed-off-by:
-
Michael S. Tsirkin authored
Applications that consume a batch of entries in one go can benefit from ability to return some of them back into the ring. Add an API for that - assuming there's space. If there's no space naturally can't do this and have to drop entries, but this implies ring is full so we'd likely drop some anyway. Signed-off-by:
Michael S. Tsirkin <mst@redhat.com> Signed-off-by:
-
- 17 May, 2017 31 commits
-
-
David S. Miller authored
Andrew Lunn says: ==================== net: phy: marvell: Checkpatch cleanup I will be contributing a few new features to the Marvell PHY driver soon. Start by making the code mostly checkpatch clean. There should not be any functional changes. Just comments set into the correct format, missing blank lines, turn some comparisons around, and refactoring to reduce indentation depth. There is still one camel in the code, but it actually makes sense, so leave it in piece. ==================== Signed-off-by: -
Andrew Lunn authored
Fold lines longer than 80 characters Signed-off-by:
Andrew Lunn <andrew@lunn.ch> Signed-off-by:
-
Andrew Lunn authored
Makes the code a bit more readable, and solves quite a few checkpatch warnings of lines longer than 80 characters. Signed-off-by:
-
Andrew Lunn authored
Break big functions up by using a number of smaller helper function. Solves some of the over 80 lines warnings, by reducing the indentation level. Signed-off-by:
-
Andrew Lunn authored
Avoid multiple assignments Comparisons should place the constant on the right side of the test Signed-off-by:
-
Andrew Lunn authored
Remove the extra blank lines, add one in where recommended. Signed-off-by:
-
Andrew Lunn authored
Use net style comment blocks, and wrap one block with long lines. Signed-off-by:
-
David S. Miller authored
Eric Dumazet says: ==================== tcp: TCP TS option use 1 ms clock TCP Timestamps option is defined in RFC 7323 Traditionally on linux, it has been tied to the internal 'jiffy' variable, because it had been a cheap and good enough generator. Unfortunately some distros use HZ=250 or even HZ=100 leading to not very useful TCP timestamps. For TCP flows in the DC, Google has used usec resolution for more than two years with great success [1]. RCVBUF autotuning is more precise. This series converts tp->tcp_mstamp to a plain u64 value storing a 1 usec TCP clock. This choice will allow us to upstream the 1 usec TS option as discussed in IETF 97. Kathleen Nichols [2] and others advocate for 1ms TS clocks for network analysis. (1ms being the lowest value supported by RFC 7323.) [1] https://www.ietf.org/proceedings/97/slides/slides-97-tcpm-tcp-options-for-low-latency-00.pdf [2] http://netseminar.stanford.edu/seminars/02_02_17.pdf ==================== Signed-off-by:
-
Eric Dumazet authored
TCP Timestamps option is defined in RFC 7323 Traditionally on linux, it has been tied to the internal 'jiffies' variable, because it had been a cheap and good enough generator. For TCP flows on the Internet, 1 ms resolution would be much better than 4ms or 10ms (HZ=250 or HZ=100 respectively) For TCP flows in the DC, Google has used usec resolution for more than two years with great success [1] Receive size autotuning (DRS) is indeed more precise and converges faster to optimal window size. This patch converts tp->tcp_mstamp to a plain u64 value storing a 1 usec TCP clock. This choice will allow us to upstream the 1 usec TS option as discussed in IETF 97. [1] https://www.ietf.org/proceedings/97/slides/slides-97-tcpm-tcp-options-for-low-latency-00.pdfSigned-off-by:
Eric Dumazet <edumazet@google.com> Acked-by:
Soheil Hassas Yeganeh <soheil@google.com> Signed-off-by:
-
Eric Dumazet authored
After this patch, all uses of tcp_time_stamp will require a change when we introduce 1 ms and/or 1 us TCP TS option. Signed-off-by:
-
Eric Dumazet authored
tcp_time_stamp will become slightly more expensive soon, cache its value. Signed-off-by:
-
Eric Dumazet authored
This CC does not need 1 ms tcp_time_stamp and can use the jiffy based 'timestamp'. Signed-off-by:
-
Eric Dumazet authored
This place wants to use tcp_jiffies32, this is good enough. Signed-off-by:
-
Eric Dumazet authored
tcp_time_stamp will no longer be tied to jiffies. Signed-off-by:
-
Eric Dumazet authored
Use tcp_jiffies32 instead of tcp_time_stamp, since tcp_time_stamp will soon be only used for TCP TS option. Signed-off-by:
-
Eric Dumazet authored
Use tcp_jiffies32 instead of tcp_time_stamp, since tcp_time_stamp will soon be only used for TCP TS option. Signed-off-by:
-
Eric Dumazet authored
Use tcp_jiffies32 instead of tcp_time_stamp, since tcp_time_stamp will soon be only used for TCP TS option. Signed-off-by:
-
Eric Dumazet authored
Use tcp_jiffies32 instead of tcp_time_stamp, since tcp_time_stamp will soon be only used for TCP TS option. Signed-off-by:
-
Eric Dumazet authored
Use tcp_jiffies32 instead of tcp_time_stamp to feed tp->snd_cwnd_stamp. tcp_time_stamp will soon be a litle bit more expensive than simply reading 'jiffies'. Signed-off-by:
-
Eric Dumazet authored
Use tcp_jiffies32 instead of tcp_time_stamp to feed tp->lsndtime. tcp_time_stamp will soon be a litle bit more expensive than simply reading 'jiffies'. Signed-off-by:
-
Eric Dumazet authored
Use our own macro instead of abusing tcp_time_stamp Signed-off-by:
-
Eric Dumazet authored
We abuse tcp_time_stamp for two different cases : 1) base to generate TCP Timestamp options (RFC 7323) 2) A 32bit version of jiffies since some TCP fields are 32bit wide to save memory. Since we want in the future to have 1ms TCP TS clock, regardless of HZ value, we want to cleanup things. tcp_jiffies32 is the truncated jiffies value, which will be used only in places where we want a 'host' timestamp. Signed-off-by:
-
Eric Dumazet authored
Idea is to later convert tp->tcp_mstamp to a full u64 counter using usec resolution, so that we can later have fine grained TCP TS clock (RFC 7323), regardless of HZ value. We try to refresh tp->tcp_mstamp only when necessary. Signed-off-by:
-
David S. Miller authored
We still need to initialize err to -EINVAL for the case where 'opt' is NULL in dsmark_init(). Fixes: 6529eaba ("net: sched: introduce tcf block infractructure") Signed-off-by:
-
David S. Miller authored
Jiri Pirko says: ==================== net: sched: introduce multichain support for filters Currently, each classful qdisc holds one chain of filters. This chain is traversed and each filter could be matched on, which may lead to execution of list of actions. One of such action could be "reclassify", which would "reset" the processing of the filter chain. So this filter chain could be looked at as a flat table. Sometimes it is convenient for user to configure a hierarchy of tables. Example usecase is encapsulation. Hierarchy of tables is a common way how it is done in HW pipelines. So it is much more convenient to offload this. This patchset contains two major patches: 8/10 - This patch introduces the support for having multiple chains of filters. 10/10 - This patch adds new control action to allow going to specified chain The rest of the patches are smaller or bigger depencies of those 2. Please see individual patch descriptions for details. Corresponding iproute2 patches are appended as a reply to this cover letter. Simple example: $ tc qdisc add dev eth0 ingress $ tc filter add dev eth0 parent ffff: protocol ip pref 33 flower dst_mac 52:54:00:3d:c7:6d action goto chain 11 $ tc filter add dev eth0 parent ffff: protocol ip pref 22 chain 11 flower dst_ip 192.168.40.1 action drop $ tc filter show dev eth0 root filter parent ffff: protocol ip pref 33 flower chain 0 filter parent ffff: protocol ip pref 33 flower chain 0 handle 0x1 dst_mac 52:54:00:3d:c7:6d eth_type ipv4 action order 1: gact action goto chain 11 random type none pass val 0 index 2 ref 1 bind 1 filter parent ffff: protocol ip pref 22 flower chain 11 filter parent ffff: protocol ip pref 22 flower chain 11 handle 0x1 eth_type ipv4 dst_ip 192.168.40.1 action order 1: gact action drop random type none pass val 0 index 3 ref 1 bind 1 ==================== Signed-off-by: -
Jiri Pirko authored
Introduce new type of termination action called "goto_chain". This allows user to specify a chain to be processed. This action type is then processed as a return value in tcf_classify loop in similar way as "reclassify" is, only it does not reset to the first filter in chain but rather reset to the first filter of the desired chain. Signed-off-by:
Jiri Pirko <jiri@mellanox.com> Acked-by:
Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by:
-
Jiri Pirko authored
Tp pointer will be needed by the next patch in order to get the chain. Signed-off-by:
-
Jiri Pirko authored
Instead of having only one filter per block, introduce a list of chains for every block. Create chain 0 by default. UAPI is extended so the user can specify which chain he wants to change. If the new attribute is not specified, chain 0 is used. That allows to maintain backward compatibility. If chain does not exist and user wants to manipulate with it, new chain is created with specified index. Also, when last filter is removed from the chain, the chain is destroyed. Signed-off-by:
-
Jiri Pirko authored
Since there will be multiple chains to dump, push chain dumping code to a separate function. Signed-off-by:
-
Jiri Pirko authored
Introduce struct tcf_chain object and set of helpers around it. Wraps up insertion, deletion and search in the filter chain. Signed-off-by:
-
Jiri Pirko authored
Call the helper from the function rather than to always adjust the return value of the function. Signed-off-by:
-
