| Message ID | 3760790a-e3c9-73d4-5191-16320f6cdbde@grimberg.me (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 04/09/2018 04:31 PM, Sagi Grimberg wrote: > >>> My device exposes nr_hw_queues which is not higher than num_online_cpus >>> so I want to connect all hctxs with hope that they will be used. >> >> The issue is that CPU online & offline can happen any time, and after >> blk-mq removes CPU hotplug handler, there is no way to remap queue >> when CPU topo is changed. >> >> For example: >> >> 1) after nr_hw_queues is set as num_online_cpus() and hw queues >> are initialized, then some of CPUs become offline, and the issue >> reported by Zhang Yi is triggered, but in this case, we should fail >> the allocation since 1:1 mapping doesn't need to use this inactive >> hw queue. > > Normal cpu offlining is fine, as the hctxs are already connected. When > we reset the controller and re-establish the queues, the issue triggers > because we call blk_mq_alloc_request_hctx. > > The question is, for this particular issue, given that the request > execution is guaranteed to run from an online cpu, will the below work? Hi Sagi Sorry for the late response, bellow patch works, here is the full log: [ 117.370832] nvme nvme0: new ctrl: NQN "nqn.2014-08.org.nvmexpress.discovery", addr 172.31.0.90:4420 [ 117.427385] nvme nvme0: creating 40 I/O queues. [ 117.736806] nvme nvme0: new ctrl: NQN "testnqn", addr 172.31.0.90:4420 [ 122.531891] smpboot: CPU 1 is now offline [ 122.573007] IRQ 37: no longer affine to CPU2 [ 122.577775] IRQ 54: no longer affine to CPU2 [ 122.582532] IRQ 70: no longer affine to CPU2 [ 122.587300] IRQ 98: no longer affine to CPU2 [ 122.592069] IRQ 140: no longer affine to CPU2 [ 122.596930] IRQ 141: no longer affine to CPU2 [ 122.603166] smpboot: CPU 2 is now offline [ 122.840577] smpboot: CPU 3 is now offline [ 125.204901] print_req_error: operation not supported error, dev nvme0n1, sector 143212504 [ 125.204907] print_req_error: operation not supported error, dev nvme0n1, sector 481004984 [ 125.204922] print_req_error: operation not supported error, dev nvme0n1, sector 436594584 [ 125.204924] print_req_error: operation not supported error, dev nvme0n1, sector 461363784 [ 125.204945] print_req_error: operation not supported error, dev nvme0n1, sector 308124792 [ 125.204957] print_req_error: operation not supported error, dev nvme0n1, sector 513395784 [ 125.204959] print_req_error: operation not supported error, dev nvme0n1, sector 432260176 [ 125.204961] print_req_error: operation not supported error, dev nvme0n1, sector 251704096 [ 125.204963] print_req_error: operation not supported error, dev nvme0n1, sector 234819336 [ 125.204966] print_req_error: operation not supported error, dev nvme0n1, sector 181874128 [ 125.938858] nvme nvme0: Reconnecting in 10 seconds... [ 125.938862] Buffer I/O error on dev nvme0n1, logical block 367355, lost async page write [ 125.942587] Buffer I/O error on dev nvme0n1, logical block 586, lost async page write [ 125.942589] Buffer I/O error on dev nvme0n1, logical block 375453, lost async page write [ 125.942591] Buffer I/O error on dev nvme0n1, logical block 587, lost async page write [ 125.942592] Buffer I/O error on dev nvme0n1, logical block 588, lost async page write [ 125.942593] Buffer I/O error on dev nvme0n1, logical block 375454, lost async page write [ 125.942594] Buffer I/O error on dev nvme0n1, logical block 589, lost async page write [ 125.942595] Buffer I/O error on dev nvme0n1, logical block 590, lost async page write [ 125.942596] Buffer I/O error on dev nvme0n1, logical block 591, lost async page write [ 125.942597] Buffer I/O error on dev nvme0n1, logical block 592, lost async page write [ 130.205584] print_req_error: 537000 callbacks suppressed [ 130.205586] print_req_error: I/O error, dev nvme0n1, sector 471135288 [ 130.218763] print_req_error: I/O error, dev nvme0n1, sector 471137240 [ 130.225985] print_req_error: I/O error, dev nvme0n1, sector 471138328 [ 130.233206] print_req_error: I/O error, dev nvme0n1, sector 471140096 [ 130.240433] print_req_error: I/O error, dev nvme0n1, sector 471140184 [ 130.247659] print_req_error: I/O error, dev nvme0n1, sector 471140960 [ 130.254874] print_req_error: I/O error, dev nvme0n1, sector 471141864 [ 130.262095] print_req_error: I/O error, dev nvme0n1, sector 471143296 [ 130.269317] print_req_error: I/O error, dev nvme0n1, sector 471143776 [ 130.276537] print_req_error: I/O error, dev nvme0n1, sector 471144224 [ 132.954315] nvme nvme0: Identify namespace failed [ 132.959698] buffer_io_error: 3801549 callbacks suppressed [ 132.959699] Buffer I/O error on dev nvme0n1, logical block 0, async page read [ 132.974669] Buffer I/O error on dev nvme0n1, logical block 0, async page read [ 132.983078] Buffer I/O error on dev nvme0n1, logical block 0, async page read [ 132.991476] Buffer I/O error on dev nvme0n1, logical block 0, async page read [ 132.999859] Buffer I/O error on dev nvme0n1, logical block 0, async page read [ 133.008217] Buffer I/O error on dev nvme0n1, logical block 0, async page read [ 133.016575] Dev nvme0n1: unable to read RDB block 0 [ 133.022423] Buffer I/O error on dev nvme0n1, logical block 0, async page read [ 133.030800] Buffer I/O error on dev nvme0n1, logical block 0, async page read [ 133.039151] nvme0n1: unable to read partition table [ 133.050221] Buffer I/O error on dev nvme0n1, logical block 65535984, async page read [ 133.060154] Buffer I/O error on dev nvme0n1, logical block 65535984, async page read [ 136.334516] nvme nvme0: creating 37 I/O queues. [ 136.636012] no online cpu for hctx 1 [ 136.640448] no online cpu for hctx 2 [ 136.644832] no online cpu for hctx 3 [ 136.650432] nvme nvme0: Successfully reconnected (1 attempts) [ 184.894584] x86: Booting SMP configuration: [ 184.899694] smpboot: Booting Node 1 Processor 1 APIC 0x20 [ 184.913923] smpboot: Booting Node 0 Processor 2 APIC 0x2 [ 184.929556] smpboot: Booting Node 1 Processor 3 APIC 0x22 And here is the debug ouput: [root@rdma-virt-01 linux (test)]$ gdb vmlinux GNU gdb (GDB) Red Hat Enterprise Linux 7.6.1-110.el7 Copyright (C) 2013 Free Software Foundation, Inc. License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html> This is free software: you are free to change and redistribute it. There is NO WARRANTY, to the extent permitted by law. Type "show copying" and "show warranty" for details. This GDB was configured as "x86_64-redhat-linux-gnu". For bug reporting instructions, please see: <http://www.gnu.org/software/gdb/bugs/>... Reading symbols from /home/test/linux/vmlinux...done. (gdb) l *(blk_mq_get_request+0x23e) 0xffffffff8136bf9e is in blk_mq_get_request (block/blk-mq.c:327). 322 rq->rl = NULL; 323 set_start_time_ns(rq); 324 rq->io_start_time_ns = 0; 325 #endif 326 327 data->ctx->rq_dispatched[op_is_sync(op)]++; 328 return rq; 329 } 330 331 static struct request *blk_mq_get_request(struct request_queue *q, > -- > diff --git a/block/blk-mq.c b/block/blk-mq.c > index 75336848f7a7..81ced3096433 100644 > --- a/block/blk-mq.c > +++ b/block/blk-mq.c > @@ -444,6 +444,10 @@ struct request *blk_mq_alloc_request_hctx(struct > request_queue *q, > return ERR_PTR(-EXDEV); > } > cpu = cpumask_first_and(alloc_data.hctx->cpumask, > cpu_online_mask); > + if (cpu >= nr_cpu_ids) { > + pr_warn("no online cpu for hctx %d\n", hctx_idx); > + cpu = cpumask_first(alloc_data.hctx->cpumask); > + } > alloc_data.ctx = __blk_mq_get_ctx(q, cpu); > > rq = blk_mq_get_request(q, NULL, op, &alloc_data); > -- > >> 2) when nr_hw_queues is set as num_online_cpus(), there may be >> much less online CPUs, so the hw queue number can be initialized as >> much smaller, then performance is degraded much even if some CPUs >> become online later. > > That is correct, when the controller will be reset though, more queues > will be added to the system. I agree it would be good if we can change > stuff dynamically. > >> So the current policy is to map all possible CPUs for handing CPU >> hotplug, and if you want to get 1:1 mapping between hw queue and >> online CPU, the nr_hw_queues can be set as num_possible_cpus. > > Having nr_hw_queues == num_possible_cpus cannot work as it requires > establishing an RDMA queue-pair with a set of HW resources both on > the host side _and_ on the controller side, which are idle. > >> Please see commit 16ccfff28976130 (nvme: pci: pass max vectors as >> num_possible_cpus() to pci_alloc_irq_vectors). > > Yes, I am aware of this patch, however I not sure it'll be a good idea > for nvmf as it takes resources from both the host and the target for > for cpus that may never come online... > >> It will waste some memory resource just like percpu variable, but it >> simplifies the queue mapping logic a lot, and can support both hard >> and soft CPU online/offline without CPU hotplug handler, which may >> cause very complicated queue dependency issue. > > Yes, but these some memory resources are becoming an issue when it > takes HW (RDMA) resources on the local device and on the target device. > >>> I agree we don't want to connect hctx which doesn't have an online >>> cpu, that's redundant, but this is not the case here. >> >> OK, I will explain below, and it can be fixed by the following patch >> too: >> >> https://marc.info/?l=linux-block&m=152318093725257&w=2 >> > > I agree this patch is good! > >>>>>> Or I may understand you wrong, :-) >>>>> >>>>> In the report we connected 40 hctxs (which was exactly the number of >>>>> online cpus), after Yi removed 3 cpus, we tried to connect 37 hctxs. >>>>> I'm not sure why some hctxs are left without any online cpus. >>>> >>>> That is possible after the following two commits: >>>> >>>> 4b855ad37194 ("blk-mq: Create hctx for each present CPU) >>>> 20e4d8139319 (blk-mq: simplify queue mapping & schedule with each >>>> possisble CPU) >>>> >>>> And this can be triggered even without putting down any CPUs. >>>> >>>> The blk-mq CPU hotplug handler is removed in 4b855ad37194, and we >>>> can't >>>> remap queue any more when CPU topo is changed, so the static & >>>> fixed mapping >>>> has to be setup from the beginning. >>>> >>>> Then if there are less enough online CPUs compared with number of >>>> hw queues, >>>> some of hctxes can be mapped with all offline CPUs. For example, if >>>> one device >>>> has 4 hw queues, but there are only 2 online CPUs and 6 offline >>>> CPUs, at most >>>> 2 hw queues are assigned to online CPUs, and the other two are all >>>> with offline >>>> CPUs. >>> >>> That is fine, but the problem that I gave in the example below which >>> has >>> nr_hw_queues == num_online_cpus but because of the mapping, we still >>> have unmapped hctxs. >> >> For FC's case, there may be some hctxs not 'mapped', which is caused by >> blk_mq_map_queues(), but that should one bug. >> >> So the patch(blk-mq: don't keep offline CPUs mapped to hctx 0)[1] is >> fixing the issue: >> >> [1] https://marc.info/?l=linux-block&m=152318093725257&w=2 >> >> Once this patch is in, any hctx should be mapped by at least one CPU. > > I think this will solve the problem Yi is stepping on. > >> Then later, the patch(blk-mq: reimplement blk_mq_hw_queue_mapped)[2] >> extends the mapping concept, maybe it should have been renamed as >> blk_mq_hw_queue_active(), will do it in V2. >> >> [2] https://marc.info/?l=linux-block&m=152318099625268&w=2 > > This is also a good patch. > > ... > >>> But when we reset the controller, we call blk_mq_update_nr_hw_queues() >>> with the current number of nr_hw_queues which never exceeds >>> num_online_cpus. This in turn, remaps the mq_map which results >>> in unmapped queues because of the mapping function, not because we >>> have more hctx than online cpus... >> >> As I mentioned, num_online_cpus() isn't one stable variable, and it >> can change any time. > > Correct, but I'm afraid num_possible_cpus might not work either. > >> After patch(blk-mq: don't keep offline CPUs mapped to hctx 0) is in, >> there won't be unmapped queue any more. > > Yes. > >>> An easy fix, is to allocate num_present_cpus queues, and only connect >>> the oneline ones, but as you said, we have unused resources this way. >> >> Yeah, it should be num_possible_cpus queues because physical CPU hotplug >> is needed to be supported for KVM or S390, or even some X86_64 system. > > num_present_cpus is a waste of resources (as I said, both on the host > and on the target), but num_possible_cpus is even worse as this is > all cpus that _can_ be populated. > >>> We also have an issue with blk_mq_rdma_map_queues with the only >>> device that supports it because it doesn't use managed affinity (code >>> was reverted) and can have irq affinity redirected in case of cpu >>> offlining... >> >> That can be one corner case, looks I have to re-consider the patch >> (blk-mq: remove code for dealing with remapping queue), which may cause >> regression for this RDMA case, but I guess CPU hotplug may break this >> case easily. >> >> [3] https://marc.info/?l=linux-block&m=152318100625284&w=2 >> >> Also this case will make blk-mq's queue mapping much complicated, >> could you provide one link about the reason for reverting managed >> affinity >> on this device? > > The problem was that users reported a regression because now > /proc/irp/$IRQ/smp_affinity is immutable. Looks like netdev users do > this on a regular basis (and also rely on irq_banacer at times) while > nvme users (and other HBAs) do not care about it. > > Thread starts here: > https://www.spinics.net/lists/netdev/msg464301.html > >> Recently we fix quite a few issues on managed affinity, maybe the >> original issue for RDMA affinity has been addressed already. > > That is not specific to RDMA affinity, its because RDMA devices are > also network devices and people want to apply their irq affinity > scripts on it like their used to with other devices. > >>> The goal here I think, should be to allocate just enough queues (not >>> more than the number online cpus) and spread it 1x1 with online cpus, >>> and also make sure to allocate completion vectors that align to online >>> cpus. I just need to figure out how to do that... >> >> I think we have to support CPU hotplug, so your goal may be hard to >> reach if you don't want to waste memory resource. > > Well, not so much if I make blk_mq_rdma_map_queues do the right thing? > > As I said, for the first go, I'd like to fix the mapping for the simple > case where we map the queues with some cpus offlined. Having queues > being added dynamically is a different story and I agree would require > more work. > > _______________________________________________ > Linux-nvme mailing list > Linux-nvme@lists.infradead.org > http://lists.infradead.org/mailman/listinfo/linux-nvme
On 04/09/2018 04:54 PM, Yi Zhang wrote: > > > On 04/09/2018 04:31 PM, Sagi Grimberg wrote: >> >>>> My device exposes nr_hw_queues which is not higher than >>>> num_online_cpus >>>> so I want to connect all hctxs with hope that they will be used. >>> >>> The issue is that CPU online & offline can happen any time, and after >>> blk-mq removes CPU hotplug handler, there is no way to remap queue >>> when CPU topo is changed. >>> >>> For example: >>> >>> 1) after nr_hw_queues is set as num_online_cpus() and hw queues >>> are initialized, then some of CPUs become offline, and the issue >>> reported by Zhang Yi is triggered, but in this case, we should fail >>> the allocation since 1:1 mapping doesn't need to use this inactive >>> hw queue. >> >> Normal cpu offlining is fine, as the hctxs are already connected. When >> we reset the controller and re-establish the queues, the issue triggers >> because we call blk_mq_alloc_request_hctx. >> >> The question is, for this particular issue, given that the request >> execution is guaranteed to run from an online cpu, will the below work? > Hi Sagi > Sorry for the late response, bellow patch works, here is the full log: And this issue cannot be reproduced on 4.15. So I did bisect testing today, found it was introduced from bellow commit: bf9ae8c blk-mq: fix bad clear of RQF_MQ_INFLIGHT in blk_mq_ct_ctx_init() > > [ 117.370832] nvme nvme0: new ctrl: NQN > "nqn.2014-08.org.nvmexpress.discovery", addr 172.31.0.90:4420 > [ 117.427385] nvme nvme0: creating 40 I/O queues. > [ 117.736806] nvme nvme0: new ctrl: NQN "testnqn", addr 172.31.0.90:4420 > [ 122.531891] smpboot: CPU 1 is now offline > [ 122.573007] IRQ 37: no longer affine to CPU2 > [ 122.577775] IRQ 54: no longer affine to CPU2 > [ 122.582532] IRQ 70: no longer affine to CPU2 > [ 122.587300] IRQ 98: no longer affine to CPU2 > [ 122.592069] IRQ 140: no longer affine to CPU2 > [ 122.596930] IRQ 141: no longer affine to CPU2 > [ 122.603166] smpboot: CPU 2 is now offline > [ 122.840577] smpboot: CPU 3 is now offline > [ 125.204901] print_req_error: operation not supported error, dev > nvme0n1, sector 143212504 > [ 125.204907] print_req_error: operation not supported error, dev > nvme0n1, sector 481004984 > [ 125.204922] print_req_error: operation not supported error, dev > nvme0n1, sector 436594584 > [ 125.204924] print_req_error: operation not supported error, dev > nvme0n1, sector 461363784 > [ 125.204945] print_req_error: operation not supported error, dev > nvme0n1, sector 308124792 > [ 125.204957] print_req_error: operation not supported error, dev > nvme0n1, sector 513395784 > [ 125.204959] print_req_error: operation not supported error, dev > nvme0n1, sector 432260176 > [ 125.204961] print_req_error: operation not supported error, dev > nvme0n1, sector 251704096 > [ 125.204963] print_req_error: operation not supported error, dev > nvme0n1, sector 234819336 > [ 125.204966] print_req_error: operation not supported error, dev > nvme0n1, sector 181874128 > [ 125.938858] nvme nvme0: Reconnecting in 10 seconds... > [ 125.938862] Buffer I/O error on dev nvme0n1, logical block 367355, > lost async page write > [ 125.942587] Buffer I/O error on dev nvme0n1, logical block 586, > lost async page write > [ 125.942589] Buffer I/O error on dev nvme0n1, logical block 375453, > lost async page write > [ 125.942591] Buffer I/O error on dev nvme0n1, logical block 587, > lost async page write > [ 125.942592] Buffer I/O error on dev nvme0n1, logical block 588, > lost async page write > [ 125.942593] Buffer I/O error on dev nvme0n1, logical block 375454, > lost async page write > [ 125.942594] Buffer I/O error on dev nvme0n1, logical block 589, > lost async page write > [ 125.942595] Buffer I/O error on dev nvme0n1, logical block 590, > lost async page write > [ 125.942596] Buffer I/O error on dev nvme0n1, logical block 591, > lost async page write > [ 125.942597] Buffer I/O error on dev nvme0n1, logical block 592, > lost async page write > [ 130.205584] print_req_error: 537000 callbacks suppressed > [ 130.205586] print_req_error: I/O error, dev nvme0n1, sector 471135288 > [ 130.218763] print_req_error: I/O error, dev nvme0n1, sector 471137240 > [ 130.225985] print_req_error: I/O error, dev nvme0n1, sector 471138328 > [ 130.233206] print_req_error: I/O error, dev nvme0n1, sector 471140096 > [ 130.240433] print_req_error: I/O error, dev nvme0n1, sector 471140184 > [ 130.247659] print_req_error: I/O error, dev nvme0n1, sector 471140960 > [ 130.254874] print_req_error: I/O error, dev nvme0n1, sector 471141864 > [ 130.262095] print_req_error: I/O error, dev nvme0n1, sector 471143296 > [ 130.269317] print_req_error: I/O error, dev nvme0n1, sector 471143776 > [ 130.276537] print_req_error: I/O error, dev nvme0n1, sector 471144224 > [ 132.954315] nvme nvme0: Identify namespace failed > [ 132.959698] buffer_io_error: 3801549 callbacks suppressed > [ 132.959699] Buffer I/O error on dev nvme0n1, logical block 0, async > page read > [ 132.974669] Buffer I/O error on dev nvme0n1, logical block 0, async > page read > [ 132.983078] Buffer I/O error on dev nvme0n1, logical block 0, async > page read > [ 132.991476] Buffer I/O error on dev nvme0n1, logical block 0, async > page read > [ 132.999859] Buffer I/O error on dev nvme0n1, logical block 0, async > page read > [ 133.008217] Buffer I/O error on dev nvme0n1, logical block 0, async > page read > [ 133.016575] Dev nvme0n1: unable to read RDB block 0 > [ 133.022423] Buffer I/O error on dev nvme0n1, logical block 0, async > page read > [ 133.030800] Buffer I/O error on dev nvme0n1, logical block 0, async > page read > [ 133.039151] nvme0n1: unable to read partition table > [ 133.050221] Buffer I/O error on dev nvme0n1, logical block > 65535984, async page read > [ 133.060154] Buffer I/O error on dev nvme0n1, logical block > 65535984, async page read > [ 136.334516] nvme nvme0: creating 37 I/O queues. > [ 136.636012] no online cpu for hctx 1 > [ 136.640448] no online cpu for hctx 2 > [ 136.644832] no online cpu for hctx 3 > [ 136.650432] nvme nvme0: Successfully reconnected (1 attempts) > [ 184.894584] x86: Booting SMP configuration: > [ 184.899694] smpboot: Booting Node 1 Processor 1 APIC 0x20 > [ 184.913923] smpboot: Booting Node 0 Processor 2 APIC 0x2 > [ 184.929556] smpboot: Booting Node 1 Processor 3 APIC 0x22 > > And here is the debug ouput: > [root@rdma-virt-01 linux (test)]$ gdb vmlinux > GNU gdb (GDB) Red Hat Enterprise Linux 7.6.1-110.el7 > Copyright (C) 2013 Free Software Foundation, Inc. > License GPLv3+: GNU GPL version 3 or later > <http://gnu.org/licenses/gpl.html> > This is free software: you are free to change and redistribute it. > There is NO WARRANTY, to the extent permitted by law. Type "show > copying" > and "show warranty" for details. > This GDB was configured as "x86_64-redhat-linux-gnu". > For bug reporting instructions, please see: > <http://www.gnu.org/software/gdb/bugs/>... > Reading symbols from /home/test/linux/vmlinux...done. > (gdb) l *(blk_mq_get_request+0x23e) > 0xffffffff8136bf9e is in blk_mq_get_request (block/blk-mq.c:327). > 322 rq->rl = NULL; > 323 set_start_time_ns(rq); > 324 rq->io_start_time_ns = 0; > 325 #endif > 326 > 327 data->ctx->rq_dispatched[op_is_sync(op)]++; > 328 return rq; > 329 } > 330 > 331 static struct request *blk_mq_get_request(struct request_queue *q, > > > >> -- >> diff --git a/block/blk-mq.c b/block/blk-mq.c >> index 75336848f7a7..81ced3096433 100644 >> --- a/block/blk-mq.c >> +++ b/block/blk-mq.c >> @@ -444,6 +444,10 @@ struct request *blk_mq_alloc_request_hctx(struct >> request_queue *q, >> return ERR_PTR(-EXDEV); >> } >> cpu = cpumask_first_and(alloc_data.hctx->cpumask, >> cpu_online_mask); >> + if (cpu >= nr_cpu_ids) { >> + pr_warn("no online cpu for hctx %d\n", hctx_idx); >> + cpu = cpumask_first(alloc_data.hctx->cpumask); >> + } >> alloc_data.ctx = __blk_mq_get_ctx(q, cpu); >> >> rq = blk_mq_get_request(q, NULL, op, &alloc_data); >> -- >> >>> 2) when nr_hw_queues is set as num_online_cpus(), there may be >>> much less online CPUs, so the hw queue number can be initialized as >>> much smaller, then performance is degraded much even if some CPUs >>> become online later. >> >> That is correct, when the controller will be reset though, more queues >> will be added to the system. I agree it would be good if we can change >> stuff dynamically. >> >>> So the current policy is to map all possible CPUs for handing CPU >>> hotplug, and if you want to get 1:1 mapping between hw queue and >>> online CPU, the nr_hw_queues can be set as num_possible_cpus. >> >> Having nr_hw_queues == num_possible_cpus cannot work as it requires >> establishing an RDMA queue-pair with a set of HW resources both on >> the host side _and_ on the controller side, which are idle. >> >>> Please see commit 16ccfff28976130 (nvme: pci: pass max vectors as >>> num_possible_cpus() to pci_alloc_irq_vectors). >> >> Yes, I am aware of this patch, however I not sure it'll be a good idea >> for nvmf as it takes resources from both the host and the target for >> for cpus that may never come online... >> >>> It will waste some memory resource just like percpu variable, but it >>> simplifies the queue mapping logic a lot, and can support both hard >>> and soft CPU online/offline without CPU hotplug handler, which may >>> cause very complicated queue dependency issue. >> >> Yes, but these some memory resources are becoming an issue when it >> takes HW (RDMA) resources on the local device and on the target device. >> >>>> I agree we don't want to connect hctx which doesn't have an online >>>> cpu, that's redundant, but this is not the case here. >>> >>> OK, I will explain below, and it can be fixed by the following patch >>> too: >>> >>> https://marc.info/?l=linux-block&m=152318093725257&w=2 >>> >> >> I agree this patch is good! >> >>>>>>> Or I may understand you wrong, :-) >>>>>> >>>>>> In the report we connected 40 hctxs (which was exactly the number of >>>>>> online cpus), after Yi removed 3 cpus, we tried to connect 37 hctxs. >>>>>> I'm not sure why some hctxs are left without any online cpus. >>>>> >>>>> That is possible after the following two commits: >>>>> >>>>> 4b855ad37194 ("blk-mq: Create hctx for each present CPU) >>>>> 20e4d8139319 (blk-mq: simplify queue mapping & schedule with each >>>>> possisble CPU) >>>>> >>>>> And this can be triggered even without putting down any CPUs. >>>>> >>>>> The blk-mq CPU hotplug handler is removed in 4b855ad37194, and we >>>>> can't >>>>> remap queue any more when CPU topo is changed, so the static & >>>>> fixed mapping >>>>> has to be setup from the beginning. >>>>> >>>>> Then if there are less enough online CPUs compared with number of >>>>> hw queues, >>>>> some of hctxes can be mapped with all offline CPUs. For example, >>>>> if one device >>>>> has 4 hw queues, but there are only 2 online CPUs and 6 offline >>>>> CPUs, at most >>>>> 2 hw queues are assigned to online CPUs, and the other two are all >>>>> with offline >>>>> CPUs. >>>> >>>> That is fine, but the problem that I gave in the example below >>>> which has >>>> nr_hw_queues == num_online_cpus but because of the mapping, we still >>>> have unmapped hctxs. >>> >>> For FC's case, there may be some hctxs not 'mapped', which is caused by >>> blk_mq_map_queues(), but that should one bug. >>> >>> So the patch(blk-mq: don't keep offline CPUs mapped to hctx 0)[1] is >>> fixing the issue: >>> >>> [1] https://marc.info/?l=linux-block&m=152318093725257&w=2 >>> >>> Once this patch is in, any hctx should be mapped by at least one CPU. >> >> I think this will solve the problem Yi is stepping on. >> >>> Then later, the patch(blk-mq: reimplement blk_mq_hw_queue_mapped)[2] >>> extends the mapping concept, maybe it should have been renamed as >>> blk_mq_hw_queue_active(), will do it in V2. >>> >>> [2] https://marc.info/?l=linux-block&m=152318099625268&w=2 >> >> This is also a good patch. >> >> ... >> >>>> But when we reset the controller, we call blk_mq_update_nr_hw_queues() >>>> with the current number of nr_hw_queues which never exceeds >>>> num_online_cpus. This in turn, remaps the mq_map which results >>>> in unmapped queues because of the mapping function, not because we >>>> have more hctx than online cpus... >>> >>> As I mentioned, num_online_cpus() isn't one stable variable, and it >>> can change any time. >> >> Correct, but I'm afraid num_possible_cpus might not work either. >> >>> After patch(blk-mq: don't keep offline CPUs mapped to hctx 0) is in, >>> there won't be unmapped queue any more. >> >> Yes. >> >>>> An easy fix, is to allocate num_present_cpus queues, and only connect >>>> the oneline ones, but as you said, we have unused resources this way. >>> >>> Yeah, it should be num_possible_cpus queues because physical CPU >>> hotplug >>> is needed to be supported for KVM or S390, or even some X86_64 system. >> >> num_present_cpus is a waste of resources (as I said, both on the host >> and on the target), but num_possible_cpus is even worse as this is >> all cpus that _can_ be populated. >> >>>> We also have an issue with blk_mq_rdma_map_queues with the only >>>> device that supports it because it doesn't use managed affinity (code >>>> was reverted) and can have irq affinity redirected in case of cpu >>>> offlining... >>> >>> That can be one corner case, looks I have to re-consider the patch >>> (blk-mq: remove code for dealing with remapping queue), which may cause >>> regression for this RDMA case, but I guess CPU hotplug may break this >>> case easily. >>> >>> [3] https://marc.info/?l=linux-block&m=152318100625284&w=2 >>> >>> Also this case will make blk-mq's queue mapping much complicated, >>> could you provide one link about the reason for reverting managed >>> affinity >>> on this device? >> >> The problem was that users reported a regression because now >> /proc/irp/$IRQ/smp_affinity is immutable. Looks like netdev users do >> this on a regular basis (and also rely on irq_banacer at times) while >> nvme users (and other HBAs) do not care about it. >> >> Thread starts here: >> https://www.spinics.net/lists/netdev/msg464301.html >> >>> Recently we fix quite a few issues on managed affinity, maybe the >>> original issue for RDMA affinity has been addressed already. >> >> That is not specific to RDMA affinity, its because RDMA devices are >> also network devices and people want to apply their irq affinity >> scripts on it like their used to with other devices. >> >>>> The goal here I think, should be to allocate just enough queues (not >>>> more than the number online cpus) and spread it 1x1 with online cpus, >>>> and also make sure to allocate completion vectors that align to online >>>> cpus. I just need to figure out how to do that... >>> >>> I think we have to support CPU hotplug, so your goal may be hard to >>> reach if you don't want to waste memory resource. >> >> Well, not so much if I make blk_mq_rdma_map_queues do the right thing? >> >> As I said, for the first go, I'd like to fix the mapping for the simple >> case where we map the queues with some cpus offlined. Having queues >> being added dynamically is a different story and I agree would require >> more work. >> >> _______________________________________________ >> Linux-nvme mailing list >> Linux-nvme@lists.infradead.org >> http://lists.infradead.org/mailman/listinfo/linux-nvme > > > _______________________________________________ > Linux-nvme mailing list > Linux-nvme@lists.infradead.org > http://lists.infradead.org/mailman/listinfo/linux-nvme
On Mon, Apr 09, 2018 at 11:31:37AM +0300, Sagi Grimberg wrote: > > > > My device exposes nr_hw_queues which is not higher than num_online_cpus > > > so I want to connect all hctxs with hope that they will be used. > > > > The issue is that CPU online & offline can happen any time, and after > > blk-mq removes CPU hotplug handler, there is no way to remap queue > > when CPU topo is changed. > > > > For example: > > > > 1) after nr_hw_queues is set as num_online_cpus() and hw queues > > are initialized, then some of CPUs become offline, and the issue > > reported by Zhang Yi is triggered, but in this case, we should fail > > the allocation since 1:1 mapping doesn't need to use this inactive > > hw queue. > > Normal cpu offlining is fine, as the hctxs are already connected. When > we reset the controller and re-establish the queues, the issue triggers > because we call blk_mq_alloc_request_hctx. That is right, blk_mq_alloc_request_hctx() is one insane interface. Also could you share a bit why the request has to be allocated in this way? I may have to read the NVMe connect protocol and related code for understanding this mechanism. > > The question is, for this particular issue, given that the request > execution is guaranteed to run from an online cpu, will the below work? > -- > diff --git a/block/blk-mq.c b/block/blk-mq.c > index 75336848f7a7..81ced3096433 100644 > --- a/block/blk-mq.c > +++ b/block/blk-mq.c > @@ -444,6 +444,10 @@ struct request *blk_mq_alloc_request_hctx(struct > request_queue *q, > return ERR_PTR(-EXDEV); > } > cpu = cpumask_first_and(alloc_data.hctx->cpumask, cpu_online_mask); > + if (cpu >= nr_cpu_ids) { > + pr_warn("no online cpu for hctx %d\n", hctx_idx); > + cpu = cpumask_first(alloc_data.hctx->cpumask); > + } We may do this way for the special case, but it is ugly, IMO. > alloc_data.ctx = __blk_mq_get_ctx(q, cpu); > > rq = blk_mq_get_request(q, NULL, op, &alloc_data); > -- > > > 2) when nr_hw_queues is set as num_online_cpus(), there may be > > much less online CPUs, so the hw queue number can be initialized as > > much smaller, then performance is degraded much even if some CPUs > > become online later. > > That is correct, when the controller will be reset though, more queues > will be added to the system. I agree it would be good if we can change > stuff dynamically. > > > So the current policy is to map all possible CPUs for handing CPU > > hotplug, and if you want to get 1:1 mapping between hw queue and > > online CPU, the nr_hw_queues can be set as num_possible_cpus. > > Having nr_hw_queues == num_possible_cpus cannot work as it requires > establishing an RDMA queue-pair with a set of HW resources both on > the host side _and_ on the controller side, which are idle. OK, can I understand it just because there isn't so many hw resources? > > > Please see commit 16ccfff28976130 (nvme: pci: pass max vectors as > > num_possible_cpus() to pci_alloc_irq_vectors). > > Yes, I am aware of this patch, however I not sure it'll be a good idea > for nvmf as it takes resources from both the host and the target for > for cpus that may never come online... > > > It will waste some memory resource just like percpu variable, but it > > simplifies the queue mapping logic a lot, and can support both hard > > and soft CPU online/offline without CPU hotplug handler, which may > > cause very complicated queue dependency issue. > > Yes, but these some memory resources are becoming an issue when it > takes HW (RDMA) resources on the local device and on the target device. Maybe both host & target resource can be allocated until there is any CPU coming for this hctx in host side. But CPU hotplug handler has to be re-introduced, maybe callback of .hctx_activate or .hctx_deactivate can be added for allocating/releasing these resources in CPU hotplug path. Since queue mapping won't be changed, and queue freezing may be avoided, it should be fine. > > > > I agree we don't want to connect hctx which doesn't have an online > > > cpu, that's redundant, but this is not the case here. > > > > OK, I will explain below, and it can be fixed by the following patch too: > > > > https://marc.info/?l=linux-block&m=152318093725257&w=2 > > > > I agree this patch is good! > > > > > > > Or I may understand you wrong, :-) > > > > > > > > > > In the report we connected 40 hctxs (which was exactly the number of > > > > > online cpus), after Yi removed 3 cpus, we tried to connect 37 hctxs. > > > > > I'm not sure why some hctxs are left without any online cpus. > > > > > > > > That is possible after the following two commits: > > > > > > > > 4b855ad37194 ("blk-mq: Create hctx for each present CPU) > > > > 20e4d8139319 (blk-mq: simplify queue mapping & schedule with each possisble CPU) > > > > > > > > And this can be triggered even without putting down any CPUs. > > > > > > > > The blk-mq CPU hotplug handler is removed in 4b855ad37194, and we can't > > > > remap queue any more when CPU topo is changed, so the static & fixed mapping > > > > has to be setup from the beginning. > > > > > > > > Then if there are less enough online CPUs compared with number of hw queues, > > > > some of hctxes can be mapped with all offline CPUs. For example, if one device > > > > has 4 hw queues, but there are only 2 online CPUs and 6 offline CPUs, at most > > > > 2 hw queues are assigned to online CPUs, and the other two are all with offline > > > > CPUs. > > > > > > That is fine, but the problem that I gave in the example below which has > > > nr_hw_queues == num_online_cpus but because of the mapping, we still > > > have unmapped hctxs. > > > > For FC's case, there may be some hctxs not 'mapped', which is caused by > > blk_mq_map_queues(), but that should one bug. > > > > So the patch(blk-mq: don't keep offline CPUs mapped to hctx 0)[1] is > > fixing the issue: > > > > [1] https://marc.info/?l=linux-block&m=152318093725257&w=2 > > > > Once this patch is in, any hctx should be mapped by at least one CPU. > > I think this will solve the problem Yi is stepping on. > > > Then later, the patch(blk-mq: reimplement blk_mq_hw_queue_mapped)[2] > > extends the mapping concept, maybe it should have been renamed as > > blk_mq_hw_queue_active(), will do it in V2. > > > > [2] https://marc.info/?l=linux-block&m=152318099625268&w=2 > > This is also a good patch. > > ... > > > > But when we reset the controller, we call blk_mq_update_nr_hw_queues() > > > with the current number of nr_hw_queues which never exceeds > > > num_online_cpus. This in turn, remaps the mq_map which results > > > in unmapped queues because of the mapping function, not because we > > > have more hctx than online cpus... > > > > As I mentioned, num_online_cpus() isn't one stable variable, and it > > can change any time. > > Correct, but I'm afraid num_possible_cpus might not work either Why? > > > After patch(blk-mq: don't keep offline CPUs mapped to hctx 0) is in, > > there won't be unmapped queue any more. > > Yes. > > > > An easy fix, is to allocate num_present_cpus queues, and only connect > > > the oneline ones, but as you said, we have unused resources this way. > > > > Yeah, it should be num_possible_cpus queues because physical CPU hotplug > > is needed to be supported for KVM or S390, or even some X86_64 system. > > num_present_cpus is a waste of resources (as I said, both on the host > and on the target), but num_possible_cpus is even worse as this is > all cpus that _can_ be populated. Yes, that can be one direction for improving queue mapping. > > > > We also have an issue with blk_mq_rdma_map_queues with the only > > > device that supports it because it doesn't use managed affinity (code > > > was reverted) and can have irq affinity redirected in case of cpu > > > offlining... > > > > That can be one corner case, looks I have to re-consider the patch > > (blk-mq: remove code for dealing with remapping queue), which may cause > > regression for this RDMA case, but I guess CPU hotplug may break this > > case easily. > > > > [3] https://marc.info/?l=linux-block&m=152318100625284&w=2 > > > > Also this case will make blk-mq's queue mapping much complicated, > > could you provide one link about the reason for reverting managed affinity > > on this device? > > The problem was that users reported a regression because now > /proc/irp/$IRQ/smp_affinity is immutable. Looks like netdev users do > this on a regular basis (and also rely on irq_banacer at times) while > nvme users (and other HBAs) do not care about it. > > Thread starts here: > https://www.spinics.net/lists/netdev/msg464301.html > > > Recently we fix quite a few issues on managed affinity, maybe the > > original issue for RDMA affinity has been addressed already. > > That is not specific to RDMA affinity, its because RDMA devices are > also network devices and people want to apply their irq affinity > scripts on it like their used to with other devices. OK, got it, then seems RDMA can't use managed IRQ affinity any more, and it has to be treated a bit special now. > > > > The goal here I think, should be to allocate just enough queues (not > > > more than the number online cpus) and spread it 1x1 with online cpus, > > > and also make sure to allocate completion vectors that align to online > > > cpus. I just need to figure out how to do that... > > > > I think we have to support CPU hotplug, so your goal may be hard to > > reach if you don't want to waste memory resource. > > Well, not so much if I make blk_mq_rdma_map_queues do the right thing? > > As I said, for the first go, I'd like to fix the mapping for the simple > case where we map the queues with some cpus offlined. Having queues > being added dynamically is a different story and I agree would require > more work. It may not be a simple case, since Zhang Yi is running CPU hotplug stress test with NVMe disconnection & connection meantime. Thanks, Ming
>> diff --git a/block/blk-mq.c b/block/blk-mq.c >> index 75336848f7a7..81ced3096433 100644 >> --- a/block/blk-mq.c >> +++ b/block/blk-mq.c >> @@ -444,6 +444,10 @@ struct request *blk_mq_alloc_request_hctx(struct >> request_queue *q, >> return ERR_PTR(-EXDEV); >> } >> cpu = cpumask_first_and(alloc_data.hctx->cpumask, cpu_online_mask); >> + if (cpu >= nr_cpu_ids) { >> + pr_warn("no online cpu for hctx %d\n", hctx_idx); >> + cpu = cpumask_first(alloc_data.hctx->cpumask); >> + } > > We may do this way for the special case, but it is ugly, IMO. Christoph?
diff --git a/block/blk-mq.c b/block/blk-mq.c index 75336848f7a7..81ced3096433 100644 --- a/block/blk-mq.c +++ b/block/blk-mq.c @@ -444,6 +444,10 @@ struct request *blk_mq_alloc_request_hctx(struct request_queue *q, return ERR_PTR(-EXDEV); } cpu = cpumask_first_and(alloc_data.hctx->cpumask, cpu_online_mask); + if (cpu >= nr_cpu_ids) { + pr_warn("no online cpu for hctx %d\n", hctx_idx); + cpu = cpumask_first(alloc_data.hctx->cpumask); + } alloc_data.ctx = __blk_mq_get_ctx(q, cpu);