December 2014 - kernelnewbies@kernelnewbies.org

I'm confused with Implementation of RPS and RFS.
by lx Dec. 19, 2014

Dec. 19, 2014

hi all: I'm read the source code of RPS and RFS, I think the key function of RPS and RFS is get_rps_cpu(), and It's codes is: ############################################################### 3079 static int get_rps_cpu(struct net_device *dev, struct sk_buff *skb, 3080 struct rps_dev_flow **rflowp) 3081 { 3082 struct netdev_rx_queue *rxqueue; 3083 struct rps_map *map; 3084 struct rps_dev_flow_table *flow_table; 3085 struct rps_sock_flow_table *sock_flow_table; 3086 int cpu = -1; 3087 u16 tcpu; 3088 u32 hash; 3089 3090 if (skb_rx_queue_recorded(skb)) { 3091 u16 index = skb_get_rx_queue(skb); 3092 if (unlikely(index >= dev->real_num_rx_queues)) { 3093 WARN_ONCE(dev->real_num_rx_queues > 1, 3094 "%s received packet on queue %u, but number " 3095 "of RX queues is %u\n", 3096 dev->name, index, dev->real_num_rx_queues); 3097 goto done; 3098 } 3099 rxqueue = dev->_rx + index; 3100 } else 3101 rxqueue = dev->_rx; 3102 3103 map = rcu_dereference(rxqueue->rps_map); 3104 if (map) { 3105 if (map->len == 1 && 3106 !rcu_access_pointer(rxqueue->rps_flow_table)) { 3107 tcpu = map->cpus[0]; 3108 if (cpu_online(tcpu)) 3109 cpu = tcpu; 3110 goto done; 3111 } 3112 } else if (!rcu_access_pointer(rxqueue->rps_flow_table)) { 3113 goto done; 3114 } 3115 3116 skb_reset_network_header(skb); 3117 hash = skb_get_hash(skb); 3118 if (!hash) 3119 goto done; 3120 3121 flow_table = rcu_dereference(rxqueue->rps_flow_table); 3122 sock_flow_table = rcu_dereference(rps_sock_flow_table); 3123 if (flow_table && sock_flow_table) { 3124 u16 next_cpu; 3125 struct rps_dev_flow *rflow; 3126 3127 rflow = &flow_table->flows[hash & flow_table->mask]; 3128 tcpu = rflow->cpu; 3129 3130 next_cpu = sock_flow_table->ents[hash & sock_flow_table->mask]; 3131 3132 /* 3133 * If the desired CPU (where last recvmsg was done) is 3134 * different from current CPU (one in the rx-queue flow 3135 * table entry), switch if one of the following holds: 3136 * - Current CPU is unset (equal to RPS_NO_CPU). 3137 * - Current CPU is offline. 3138 * - The current CPU's queue tail has advanced beyond the 3139 * last packet that was enqueued using this table entry. 3140 * This guarantees that all previous packets for the flow 3141 * have been dequeued, thus preserving in order delivery. 3142 */ 3143 if (unlikely(tcpu != next_cpu) && 3144 (tcpu == RPS_NO_CPU || !cpu_online(tcpu) || 3145 ((int)(per_cpu(softnet_data, tcpu).input_queue_head - 3146 rflow->last_qtail)) >= 0)) { 3147 tcpu = next_cpu; 3148 rflow = set_rps_cpu(dev, skb, rflow, next_cpu); 3149 } 3150 3151 if (tcpu != RPS_NO_CPU && cpu_online(tcpu)) { 3152 *rflowp = rflow; 3153 cpu = tcpu; 3154 goto done; 3155 } 3156 } 3157 3158 if (map) { 3159 tcpu = map->cpus[reciprocal_scale(hash, map->len)]; 3160 if (cpu_online(tcpu)) { 3161 cpu = tcpu; 3162 goto done; 3163 } 3164 } 3165 3166 done: 3167 return cpu; 3168 } ############################################################### I know the RPS/RFS get the CPU id by this function.I think RPS can't work, when RFS is active, beacuse the implementation of RPS is between 3158 and 3163. and the codes between 3123 and 3156 are used for RFS. I think RPS is work just in this two conditions: 1. When the packet first arrived this host, the value of tcpu and next_cpu are RPS_NO_CPU. 2. When the destination of packet is not this host, the value of tcpu and next_cpu always are RPS_NO_CPU. My question is: If the RFS is enable, the RPS can't work? Thank you.

1 0

How to submit for eudyptula-challenge
by wangyubin Dec. 19, 2014

Dec. 19, 2014

Hello, I have dive into the eudyptula-challenge, but when I submit my result for task 01. I received a reply below: > Please print to the kernel debug log level. I have sended module(c file), Makefile, and dmesg output for load/unload the module. Could anyone to tell me what does this message mean? Thanks a lot! -- Best Regards! wangyubin

15 20

Asus N53 USB WiFi Driver porting
by Abhishek Sharma Dec. 18, 2014

Dec. 18, 2014

Hi folks, I have a board with 2.6.27.5 kernel and I want to use Asus N53 USB WiFi dongle on it in direct mode. I have used rt5572sta driver ver 2.6.0.1 and the dongle is able to work in infra and ad hoc mode, but direct mode is not working. Only ra0 interface is made when the module is inserted, interface p2p0 needed for direct mode does not comes up. I can not change my kernel version as all my running applications can not be ported to newer kernels. Thanks in advance for your help! Regards, A Sharma

6 8

Cpu Freq Merge with Scheduler
by nick Dec. 18, 2014

Dec. 18, 2014

Greetings Fellow Developers, I am curious about the work going into making the kernel scheduler more CPU power efficient. I have done some googling on this and am curious about what how is going into the ideas/patches for this work. Nick

2 2

printk from x86-64 assembly code
by John de la Garza Dec. 18, 2014

Dec. 18, 2014

I want to call printk from arch/x86/lib/copy_user_64.S. The only example in assembly that I can find for calling something like printk is from: arch/x86/kernel/head_64.S (code follows) here is what I tried: ENTRY(jj_print) CFI_STARTPROC ASM_STAC leaq jj_buf(%rip),%rdi mov $0, %rsi call printk xorl %eax,%eax ASM_CLAC ret CFI_ENDPROC ENDPROC(jj_print) jj_buf: .ascii "hello asm\n\0" Then I added this to where I wanted to run the code: ALTERNATIVE_JUMP X86_FEATURE_REP_GOOD,X86_FEATURE_ERMS, \ jj_print, jj_print, jj_print I wasn't sure how to make the call to jj_print so I just used the ALTERNATIVE_JUMP without trying to understand it, but from what I can tell, it seems like it should work (I know it is not ideal and lame) sample workig code which I based my jj_print func on: #ifdef CONFIG_EARLY_PRINTK GET_CR2_INTO(%r9) # can clobber any volatile register if pv movl 80(%rsp),%r8d # error code movl 72(%rsp),%esi # vector number movl 96(%rsp),%edx # %cs movq 88(%rsp),%rcx # %rip xorl %eax,%eax leaq early_idt_msg(%rip),%rdi call early_printk cmpl $2,early_recursion_flag(%rip) jz 1f call dump_stack #ifdef CONFIG_KALLSYMS leaq early_idt_ripmsg(%rip),%rdi movq 40(%rsp),%rsi # %rip again call __print_symbol #endif #endif /* EARLY_PRINTK */

2 1

Solving checkpatch error with lex
by karthik nayak Dec. 18, 2014

Dec. 18, 2014

Hello, was trying to fix the checkpatch error of not using c99 comments in the file drivers/staging/rtl8192u/r8192U_dm.c so I wrote a lex program to do so since there were way too many to do it manually. Could anyone have a look at my lex code and tell me if I need to change it. I have attached the lex code. As per my testing it works perfectly, would love a second opinion. Thanks, Karthik

4 8

kobject sample code
by John de la Garza Dec. 17, 2014

Dec. 17, 2014

I'm looking at samples/kobject/kobject-example.c at line 39 foo is read: sscanf(buf, "%du", &foo); foo is an int so why is it read using %du not %d?

3 4

Synchronize_rcu() and synchronize_sched() in _cpu_down path
by Vignesh Radhakrishnan Dec. 17, 2014

Dec. 17, 2014

Hi, I had a query regarding calling synchronize_rcu and synchronize_sched in _cpu_down path : Here is the snippet below : /* * By now we've cleared cpu_active_mask, wait for all preempt-disabled * and RCU users of this state to go away such that all new such users * will observe it. * * For CONFIG_PREEMPT we have preemptible RCU and its sync_rcu() might * not imply sync_sched(), so explicitly call both. * * Do sync before park smpboot threads to take care the rcu boost case. */ #ifdef CONFIG_PREEMPT synchronize_sched(); #endif synchronize_rcu(); My query is that : During cpu_down path, we do migrate tasks in this CPU to another CPU As per LWN article here - http://lwn.net/Articles/253651/ : *For example, suppose that a task calls rcu_read_lock() on one CPU, is preempted, resumes on another CPU, and then calls rcu_read_unlock(). The first CPU's counter will then be +1 and the second CPU's counter will be -1, however, they will still sum to zero. Regardless of possible preemption, when the sum of the old counter elements does go to zero, it is safe to move to the next grace-period stage, as shown on the right-hand side of the above figure. * >From what i understand : Since we migrate the tasks to another CPU, the read side critical sections should get completed from other CPU on which this task is going to run. Therefore why call synchronize_rcu()/synchronize_sched() and wait in the cpu_down path while there is no write operations happening here in this code? Thanks and regards, Vignesh Radhakrishnan

1 0

Re: Kernelnewbies Digest, Vol 49, Issue 30
by Dean Michael Ancajas Dec. 17, 2014

Dec. 17, 2014

> ---------------------------------------------------------------------- > > Message: 1 > Date: Mon, 15 Dec 2014 14:56:59 -0600 > From: Jessica Ross <jessica.r.ross(a)gmail.com> > Subject: Re: [eudyptula] has anyone received a gift after completing > the challenge ? > To: Markus B?hme <markus.boehme(a)outlook.com> > Cc: kernelnewbies <kernelnewbies(a)kernelnewbies.org> > Message-ID: > <CA%2Bzc98Qn07iynMmHPJfAaV_tv6OQbbW%3Do3t6x5S23LL-%3D3Z09g(a)mail.gmail.com> > Content-Type: text/plain; charset="utf-8" > > I got it, I was impatient. > I need to find a real machine for task 2 now. :) > oh man, I have been waiting for 5 days now since I submitted task 03. I can't believe how slow this script is if all it's doing is parsing the results. I don't mean this in a disrespectful way but I'd happily pay 20$ to run this script on a VPS for one month so I could finish the tasks promptly. I am that desperate to finish it. I am itching to contribute but at this pace, I don't think it's far fetched to say you'll finish all 20 tasks in at least a year. Could we get the eudyptula guys to participate in the discussion please? Thanks, Dean

5 6

Proprietary drivers
by Ali Aminian Dec. 16, 2014

Dec. 16, 2014

Hello I am wondering how should i install proprietary drivers on mainline kernel. I have NVIDIA Corporation G98M [GeForce 9300M GS] graphic chipset and Broadcom Corporation BCM4312 network adapter. in Broadcom case the driver source is available although the compilation process is not very clear to me but nvidia source code is not available and i don't know if there is a way to use drivers available in ubuntu's repository. My linux distribution is Ubuntu 14.04 and when i install mainline kernel these two hardware stop working. And also how can i find linux certified laptop. i mean a laptop that has all its driver in linux source tree. i have seen ubuntu certified laptops that again use nvidia chipsets. thank you

3 2