December 2014 - kernelnewbies@kernelnewbies.org

Re: Help needed to implement parameter in SCHED_DEADLINE policy
by Chanchal Paul Dec. 27, 2014

Dec. 27, 2014

Hi, Let me explain what I am thinking. I am only concerned about EDF scheduling policy. The scheduler determines the order in which it should schedule tasks according to their DEADLINE. I dont want to disturb that order. If a scheduler picks up a task which is computationally intensive and lets it run then due to its thermal characteristic it may heat up processor so much that eventually a hardware DTM or a frequency scaling occurs and the system slows down as a effect. Where as if the processor would have waited a little bit (only that much due to which the task would not miss its deadline) before letting the task run, and eventually lets the other computationally non intensive tasks run (from the other scheduling classes) then may be that DTM would have not occurred and may be the system would not have slowed down. So I want to solve this problem. or at least see what are the results after implementation. I can think that the scheduler need more information about the EDF tasks about their thermal (or computational) characteristics. Initially I want to supply that information when I submit a EDF task. Thank you. On Sat, Dec 27, 2014 at 3:44 AM, Henrik Austad <henrik(a)austad.us> wrote: > On Fri, Dec 26, 2014 at 09:14:22PM +0530, Chanchal Paul wrote: >> Hi all, > > Hi Paul, > >> I am trying to implement my custom scheduling policy in rt patched >> linux kernel. I am working with version 3.14.25-rt22. I have >> successfully installed the kernel after patching with rt patch of same >> version. >> >> Now the installed kernel also has SCHED_DEADLINE i.e. the EDF scheduling policy. >> >> The problem in my hand is to introduce a parameter in the >> SCHED_DEADLINE policy which will further dynamically change the >> priority according to the supplied value while the task is submitted. > > You do know that SCHED_DEADLINE does not use priorities but _deadlines_, > right? > > care to elaborate a bit? > >> I have already simulated and tested my requirement. I am having >> problem with understanding the hierarchy of scheduler code. > > You mean the order of the scheduling policies? If you look at the > sched_class in the different classes, you'll find the following hierarchy: > > stop_sched_class > dl_sched_class > rt_sched_class > fair_sched_class > idle_sched_class > > So, first the scheduler sees if there's any deadline tasks, then -rt and so > on. > > I (we) need a bit more info about what exactly it is you're struggling with > >> >> Thanks in advance >> -- >> To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in >> the body of a message to majordomo(a)vger.kernel.org >> More majordomo info at http://vger.kernel.org/majordomo-info.html > > -- > Henrik Austad

2 1

Help needed to implement parameter in SCHED_DEADLINE policy
by Chanchal Paul Dec. 26, 2014

Dec. 26, 2014

Hi all, I am trying to implement my custom scheduling policy in rt patched linux kernel. I am working with version 3.14.25-rt22. I have successfully installed the kernel after patching with rt patch of same version. Now the installed kernel also has SCHED_DEADLINE i.e. the EDF scheduling policy. The problem in my hand is to introduce a parameter in the SCHED_DEADLINE policy which will further dynamically change the priority according to the supplied value while the task is submitted. I have already simulated and tested my requirement. I am having problem with understanding the hierarchy of scheduler code. Thanks in advance.

2 1

Problem facing while running checkpatch.pl on linux-next tree
by Rahul Garg Dec. 25, 2014

Dec. 25, 2014

Hi All, I am trying to run checkpatch.pl on linux-next tree's staging folder. But whatever sub directory, I try to run it on. It is giving me the same error for all the files : ERROR: Does not appear to be a unified-diff format patch I thought that it might be a problem with my git. So I clone the git tree again. So can anyone help me, what am I missing ? Regards Rahul

5 4

Re: hi about process_backlog()
by Vignesh Radhakrishnan Dec. 25, 2014

Dec. 25, 2014

Hi, I was tracing the program flow. So basically i figured out that net_rx_action calls process_backlog and i saw that deletion is happening at net_rx_action(). So i used git blame on that file and saw that this commit performed the deletion. Then when i saw what the commit does, i found that it changed the deletion location from process_backlog to here. Thanks and regards, Vignesh Radhakrishnan On Wed, Dec 24, 2014 at 8:22 AM, lx <lxgeek(a)yahoo.com> wrote: > hi : > How to find this commit: > ################################# > Earlier we had deletion in process_backlog itself, but > https://git.kernel.org/cgit/linux/kernel/git/torvalds/ > linux.git/commit/?id=d75b1ade567ffab085e8adbbdacf0092d10cd09c has changed > it. > ################################# > > There are so many commits, and this commit don't show some messages about > process_backlog().The messages is: > ################################# > commit d75b1ade567ffab085e8adbbdacf0092d10cd09c > Author: Eric Dumazet <edumazet(a)google.com> > Date: Sun Nov 2 06:19:33 2014 -0800 > > net: less interrupt masking in NAPI > > net_rx_action() can mask irqs a single time to transfert sd->poll_list > into a private list, for a very short duration. > > Then, napi_complete() can avoid masking irqs again, > and net_rx_action() only needs to mask irq again in slow path. > > This patch removes 2 couples of irq mask/unmask per typical NAPI run, > more if multiple napi were triggered. > > Note this also allows to give control back to caller (do_softirq()) > more often, so that other softirq handlers can be called a bit earlier, > or ksoftirqd can be wakeup earlier under pressure. > > This was developed while testing an alternative to RX interrupt > mitigation to reduce latencies while keeping or improving GRO > aggregation on fast NIC. > > Idea is to test napi->gro_list at the end of a napi->poll() and > reschedule one NAPI poll, but after servicing a full round of > softirqs (timers, TX, rcu, ...). This will be allowed only if softirq > is currently serviced by idle task or ksoftirqd, and resched not > needed. > > Signed-off-by: Eric Dumazet <edumazet(a)google.com> > Cc: Willem de Bruijn <willemb(a)google.com> > Signed-off-by: David S. Miller <davem(a)davemloft.net> > ################################# > > > Thank you. > -- http://vigneshradhakrishnan.blogspot.com/

1 0

Stack register value in the kernel setup code
by Alexander Kuleshov Dec. 24, 2014

Dec. 24, 2014

Hello, I runn kernel with grub in qemu, connected with gdb and put break point to 0x10200 (to this jump - https://github.com/torvalds/linux/blob/master/arch/x86/boot/header.S#L297) Registers value are following: eax 0x1020 4128 ecx 0x0 0 edx 0x100000 1048576 ebx 0x0 0 esp 0xf7f4 0xf7f4 ebp 0x0 0x0 esi 0x1ac7c 109692 edi 0x0 0 eip 0x68 0x68 eflags 0x2 [ ] cs 0x1020 4128 ss 0x1000 4096 ds 0x1000 4096 es 0x1000 4096 fs 0x1000 4096 gs 0x1000 4096 Seems that segment registers are corret, but why esp is 0xf7f4? Thank you.

2 2

Re: delete NIC from poll list in process_backlog()
by lx Dec. 24, 2014

Dec. 24, 2014

hi all: I think I got it. The new kernel just remove the NIC from poll list first, If we should repoll it, kernel will add it again. The codes is: 4560 static void net_rx_action(struct softirq_action *h) 4561 { 4562 struct softnet_data *sd = this_cpu_ptr(&softnet_data); 4563 unsigned long time_limit = jiffies + 2; 4564 int budget = netdev_budget; 4565 LIST_HEAD(list); 4566 LIST_HEAD(repoll); 4567 void *have; 4568 4569 local_irq_disable(); 4570 list_splice_init(&sd->poll_list, &list); 4571 local_irq_enable(); 4572 4573 while (!list_empty(&list)) { 4574 struct napi_struct *n; 4575 int work, weight; 4576 4577 /* If softirq window is exhausted then punt. 4578 * Allow this to run for 2 jiffies since which will allow 4579 * an average latency of 1.5/HZ. 4580 */ 4581 if (unlikely(budget <= 0 || time_after_eq(jiffies, time_limit))) 4582 goto softnet_break; 4583 4584 4585 n = list_first_entry(&list, struct napi_struct, poll_list); 4586 list_del_init(&n->poll_list); 4587 4588 have = netpoll_poll_lock(n); 4589 4590 weight = n->weight; 4591 4592 /* This NAPI_STATE_SCHED test is for avoiding a race 4593 * with netpoll's poll_napi(). Only the entity which 4594 * obtains the lock and sees NAPI_STATE_SCHED set will 4595 * actually make the ->poll() call. Therefore we avoid 4596 * accidentally calling ->poll() when NAPI is not scheduled. 4597 */ 4598 work = 0; 4599 if (test_bit(NAPI_STATE_SCHED, &n->state)) { 4600 work = n->poll(n, weight); 4601 trace_napi_poll(n); 4602 } 4603 4604 WARN_ON_ONCE(work > weight); 4605 4606 budget -= work; 4607 4608 /* Drivers must not modify the NAPI state if they 4609 * consume the entire weight. In such cases this code 4610 * still "owns" the NAPI instance and therefore can 4611 * move the instance around on the list at-will. 4612 */ 4613 if (unlikely(work == weight)) { 4614 if (unlikely(napi_disable_pending(n))) { 4615 napi_complete(n); 4616 } else { 4617 if (n->gro_list) { 4618 /* flush too old packets 4619 * If HZ < 1000, flush all packets. 4620 */ 4621 napi_gro_flush(n, HZ >= 1000); 4622 } 4623 list_add_tail(&n->poll_list, &repoll); 4624 } 4625 } 4626 4627 netpoll_poll_unlock(have); 4628 } 4629 4630 if (!sd_has_rps_ipi_waiting(sd) && 4631 list_empty(&list) && 4632 list_empty(&repoll)) 4633 return; 4634 out: 4635 local_irq_disable(); 4636 4637 list_splice_tail_init(&sd->poll_list, &list); 4638 list_splice_tail(&repoll, &list); 4639 list_splice(&list, &sd->poll_list); 4640 if (!list_empty(&sd->poll_list)) 4641 __raise_softirq_irqoff(NET_RX_SOFTIRQ); 4642 4643 net_rps_action_and_irq_enable(sd); 4644 4645 return; 4646 4647 softnet_break: 4648 sd->time_squeeze++; 4649 goto out; 4650 } line 4585: choose a NIC which will poll. line 4586: remove the NIC from the poll list. line 4623: add the NIC again in repoll list if we nedd poll it. line 4638: splice the poll list and repoll list. So, I think this is not a bug. @valdis.kletnieks I think we should remove NIC form poll list after the list of sd->input_pkt_queue is empty, so I think code should run like this: 4350 rps_lock(sd); 4351 if (skb_queue_empty(&sd->input_pkt_queue)) { 4352 /* 4353 * Inline a custom version of __napi_complete(). 4354 * only current cpu owns and manipulates this napi, 4355 * and NAPI_STATE_SCHED is the only possible flag set 4356 * on backlog. 4357 * We can use a plain write instead of clear_bit(), 4358 * and we dont need an smp_mb() memory barrier. 4359 */ list_del(&napi->poll_list); // I add it. 4360 napi->state = 0; 4361 rps_unlock(sd); 4362 4363 break; 4364 } Thank you 在 2014/12/23 20:13, Vignesh Radhakrishnan 写道: > We do delete it. I am not very thorough with the code , but here is what > i think is happening. > > process_backlog is set as the backlog polling function for softnet data > sd->backlog.poll = process_backlog; > > Now net_rx_action which runs in soft irq context, calls the polling as > per the following code and the deletion as well. > > > Notice where we are deleting it : > > n = list_first_entry(&list, struct napi_struct, poll_list); > list_del_init(&n->poll_list); > --------------------------------------> Here is the deletion. > > have = netpoll_poll_lock(n); > > weight = n->weight; > > /* This NAPI_STATE_SCHED test is for avoiding a race > * with netpoll's poll_napi(). Only the entity which > * obtains the lock and sees NAPI_STATE_SCHED set will > * actually make the ->poll() call. Therefore we avoid > * accidentally calling ->poll() when NAPI is not > scheduled. > */ > work = 0; > if (test_bit(NAPI_STATE_SCHED, &n->state)) { > work = n->poll(n, weight); > -------------------------------------> Process_backlog is called here > trace_napi_poll(n); > } > > Earlier we had deletion in process_backlog itself, but > https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=… > has changed it. > > Thanks and regards, > Vignesh Radhakrishnan > > On Tue, Dec 23, 2014 at 2:29 PM, lx <lxlenovostar(a)gmail.com > <mailto:lxlenovostar@gmail.com>> wrote: > > hi all: > I read the function of process_backlog, and I found it don't > delete the NIC form poll list. the codes is: > > 4321 static int process_backlog(struct napi_struct *napi, int quota) > 4322 { > 4323 int work = 0; > 4324 struct softnet_data *sd = container_of(napi, struct > softnet_data, backlog); > 4325 > 4326 /* Check if we have pending ipi, its better to send them now, > 4327 * not waiting net_rx_action() end. > 4328 */ > 4329 if (sd_has_rps_ipi_waiting(sd)) { > 4330 local_irq_disable(); > 4331 net_rps_action_and_irq_enable(sd); > 4332 } > 4333 > 4334 napi->weight = weight_p; > 4335 local_irq_disable(); > 4336 while (1) { > 4337 struct sk_buff *skb; > 4338 > 4339 while ((skb = __skb_dequeue(&sd->process_queue))) { > 4340 local_irq_enable(); > 4341 __netif_receive_skb(skb); > 4342 local_irq_disable(); > 4343 input_queue_head_incr(sd); > 4344 if (++work >= quota) { > 4345 local_irq_enable(); > 4346 return work; > 4347 } > 4348 } > 4349 > 4350 rps_lock(sd); > 4351 if (skb_queue_empty(&sd->input_pkt_queue)) { > 4352 /* > 4353 * Inline a custom version of __napi_complete(). > 4354 * only current cpu owns and manipulates this napi, > 4355 * and NAPI_STATE_SCHED is the only possible flag set > 4356 * on backlog. > 4357 * We can use a plain write instead of clear_bit(), > 4358 * and we dont need an smp_mb() memory barrier. > 4359 */ > 4360 napi->state = 0; > 4361 rps_unlock(sd); > 4362 > 4363 break; > 4364 } > 4365 > 4366 skb_queue_splice_tail_init(&sd->input_pkt_queue, > 4367 &sd->process_queue); > 4368 rps_unlock(sd); > 4369 } > 4370 local_irq_enable(); > 4371 > 4372 return work; > 4373 } > > I think we should delete the NIC from the poll list by: > list_del(&napi->poll_list); > > and my git repository infromation is: > > [root@localhost linux]# git show > commit aa39477b5692611b91ac9455ae588738852b3f60 > Merge: 48ec833 5164bec > Author: Linus Torvalds <torvalds(a)linux-foundation.org > <mailto:torvalds@linux-foundation.org>> > Date: Mon Dec 22 14:47:17 2014 -0800 > > Merge tag 'dm-3.19-fixes' of > git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm > <http://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm> > Pull device mapper fixes from Mike Snitzer: > "Thre stable fixes and one fix for a regression introduced > during 3.19 > merge: > - Fix inability to discard used space when the thin-pool > target is in > out-of-data-space mode and also transition the thin-pool > back to > write mode once free space is made available. > - Fix DM core bio-based end_io bug that prevented proper > post-processing of the error code returned from the block > layer. > - Fix crash in DM thin-pool due to thin device being added > to the > pool's active_thins list before properly initializing the thin > device's refcount" > * tag 'dm-3.19-fixes' of > git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm > <http://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm>: > dm: fix missed error code if .end_io isn't implemented by > target_type > dm thin: fix crash by initializing thin device's refcount and > completion earlier > dm thin: fix missing out-of-data-space to write mode > transition if blocks are released > dm thin: fix inability to discard blocks when in > out-of-data-space mode > > > Don't handle the list_del is right, or this is a bug? > > Thank you. > > _______________________________________________ > Kernelnewbies mailing list > Kernelnewbies(a)kernelnewbies.org <mailto:Kernelnewbies@kernelnewbies.org> > http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies > > > > > -- > http://vigneshradhakrishnan.blogspot.com/

1 0

delete NIC from poll list in process_backlog()
by lx Dec. 23, 2014

Dec. 23, 2014

hi all: I read the function of process_backlog, and I found it don't delete the NIC form poll list. the codes is: 4321 static int process_backlog(struct napi_struct *napi, int quota) 4322 { 4323 int work = 0; 4324 struct softnet_data *sd = container_of(napi, struct softnet_data, backlog); 4325 4326 /* Check if we have pending ipi, its better to send them now, 4327 * not waiting net_rx_action() end. 4328 */ 4329 if (sd_has_rps_ipi_waiting(sd)) { 4330 local_irq_disable(); 4331 net_rps_action_and_irq_enable(sd); 4332 } 4333 4334 napi->weight = weight_p; 4335 local_irq_disable(); 4336 while (1) { 4337 struct sk_buff *skb; 4338 4339 while ((skb = __skb_dequeue(&sd->process_queue))) { 4340 local_irq_enable(); 4341 __netif_receive_skb(skb); 4342 local_irq_disable(); 4343 input_queue_head_incr(sd); 4344 if (++work >= quota) { 4345 local_irq_enable(); 4346 return work; 4347 } 4348 } 4349 4350 rps_lock(sd); 4351 if (skb_queue_empty(&sd->input_pkt_queue)) { 4352 /* 4353 * Inline a custom version of __napi_complete(). 4354 * only current cpu owns and manipulates this napi, 4355 * and NAPI_STATE_SCHED is the only possible flag set 4356 * on backlog. 4357 * We can use a plain write instead of clear_bit(), 4358 * and we dont need an smp_mb() memory barrier. 4359 */ 4360 napi->state = 0; 4361 rps_unlock(sd); 4362 4363 break; 4364 } 4365 4366 skb_queue_splice_tail_init(&sd->input_pkt_queue, 4367 &sd->process_queue); 4368 rps_unlock(sd); 4369 } 4370 local_irq_enable(); 4371 4372 return work; 4373 } I think we should delete the NIC from the poll list by: list_del(&napi->poll_list); and my git repository infromation is: [root@localhost linux]# git show commit aa39477b5692611b91ac9455ae588738852b3f60 Merge: 48ec833 5164bec Author: Linus Torvalds <torvalds(a)linux-foundation.org> Date: Mon Dec 22 14:47:17 2014 -0800 Merge tag 'dm-3.19-fixes' of git:// git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm Pull device mapper fixes from Mike Snitzer: "Thre stable fixes and one fix for a regression introduced during 3.19 merge: - Fix inability to discard used space when the thin-pool target is in out-of-data-space mode and also transition the thin-pool back to write mode once free space is made available. - Fix DM core bio-based end_io bug that prevented proper post-processing of the error code returned from the block layer. - Fix crash in DM thin-pool due to thin device being added to the pool's active_thins list before properly initializing the thin device's refcount" * tag 'dm-3.19-fixes' of git:// git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm: dm: fix missed error code if .end_io isn't implemented by target_type dm thin: fix crash by initializing thin device's refcount and completion earlier dm thin: fix missing out-of-data-space to write mode transition if blocks are released dm thin: fix inability to discard blocks when in out-of-data-space mode Don't handle the list_del is right, or this is a bug? Thank you.

2 1

Problem in loading Nouveau driver
by Ali Aminian Dec. 21, 2014

Dec. 21, 2014

Hello How should i enable Nouveau driver in a custom kernel. Here is how i compiled my kernel: 1_make localmodconfig 2_make menuconfig and then went to: device drivers --> Graphics support --> Nouveau Support 3_make 4_make modules_install 5_make install and after booting the kernel i tried loading Nouveau by following commands: 1_insmod drm.ko 2_insmod drm_kms_helper.ko 3_insmod ttm.ko 4_insmod nouveau.ko but in the last step it gives me Unknown symbol in module here is the result of "dmesg | tail" after i tried insmod nouvea.ko: drm: module verification failed: signature and/or required key missing - tainting kernel [ 212.330197] [drm] Initialized drm 1.1.0 20060810 [ 255.336591] nouveau: Unknown symbol i2c_bit_add_bus (err 0) [ 255.336610] nouveau: Unknown symbol mxm_wmi_call_mxds (err 0) [ 255.336625] nouveau: Unknown symbol mxm_wmi_supported (err 0) [ 255.336644] nouveau: Unknown symbol i2c_bit_algo (err 0) [ 255.336719] nouveau: Unknown symbol wmi_evaluate_method (err 0) [ 255.336777] nouveau: Unknown symbol wmi_has_guid (err 0) [ 255.336904] nouveau: Unknown symbol mxm_wmi_call_mxmx (err 0) I don't know in what step i did wrong. thank you.

3 5

where do I find the n_gsm module
by robert rottermann Dec. 21, 2014

Dec. 21, 2014

Hi there, this is probably a stupid question .. but where do I fin the N_gsm module. I want to install a utility (https://github.com/Rtone/cmux) that needs the N_gsm module loaded. However neither on raspian nor ubuntu this module is available. thanks robert

2 1

make install error
by Srushti Khadke Dec. 19, 2014

Dec. 19, 2014

os version centOs 6.5 kernel version is 2.6.32-431.el6.x86_64 i followed the procedure 1.make menuconfig (enabled force module unloadin,in kernel hacking changed the debug level from 4 to 7) 2.make 3.make modules 4.make modules_install 5.make install all steps upto 4 gave successful output but the fifth step is giving eror.that error is [root@localhost kvm]# make install sh ./arch/x86/boot/install.sh 3.18.0-rc6+ arch/x86/boot/bzImage \ System.map "/boot" ERROR: modinfo: could not find module aes_generic ERROR: modinfo: could not find module ipt_MASQUERADE ERROR: modinfo: could not find module iptable_nat ERROR: modinfo: could not find module nf_nat ERROR: modinfo: could not find module sco ERROR: modinfo: could not find module l2cap ERROR: modinfo: could not find module freq_table ERROR: modinfo: could not find module mperf ERROR: modinfo: could not find module uvcvideo ERROR: modinfo: could not find module videodev ERROR: modinfo: could not find module v4l2_compat_ioctl32 ERROR: modinfo: could not find module snd_page_alloc ERROR: modinfo: could not find module crc_t10dif ERROR: modinfo: could not find module output

2 1