> On Jul 27, 2022, at 12:18 PM, Olga Kornievskaia <aglo@xxxxxxxxx> wrote: > > Hi Chuck, Sorry for the delay, I was traveling. > To make it compile I did: > diff --git a/fs/nfsd/nfs4proc.c b/fs/nfsd/nfs4proc.c > index 7196bcafdd86..f6deffc921d0 100644 > --- a/fs/nfsd/nfs4proc.c > +++ b/fs/nfsd/nfs4proc.c > @@ -1536,7 +1536,7 @@ nfsd4_setup_inter_ssc(struct svc_rqst *rqstp, > if (status) > goto out; > > - status = nfsd4_interssc_connect(©->cp_src, rqstp, mount); > + status = nfsd4_interssc_connect(copy->cp_src, rqstp, mount); > if (status) > goto out; Yes, same bug was reported by the day-0 kbot. v1 was kind of an RFC, as I hadn't fully tested it. Sorry for mislabeling it. I will post a v2 of this series with this fixed and with Dai's fix for nfsd4_decode_copy(). Stand by. > But when I tried to run the nfstest_ssc. The first test (intra01) made > the server oops: > > [ 9569.551100] CPU: 0 PID: 2861 Comm: nfsd Not tainted 5.19.0-rc6+ #73 > [ 9569.552385] Hardware name: VMware, Inc. VMware Virtual > Platform/440BX Desktop Reference Platform, BIOS 6.00 11/12/2020 > [ 9569.555043] RIP: 0010:nfsd4_copy+0x28b/0x4e0 [nfsd] > [ 9569.556662] Code: 24 38 49 89 94 24 10 01 00 00 49 8b 56 08 48 8d > 79 08 49 89 94 24 18 01 00 00 49 8b 56 10 48 83 e7 f8 49 89 94 24 20 > 01 00 00 <48> 8b 06 48 89 01 48 8b 86 04 04 00 00 48 89 81 04 04 00 00 > 48 29 > [ 9569.561792] RSP: 0018:ffffb092c0c97dd0 EFLAGS: 00010282 > [ 9569.563112] RAX: ffff99b5465c2460 RBX: ffff99b5a68828e0 RCX: ffff99b5853b6000 > [ 9569.565196] RDX: 0000000000000001 RSI: 0000000000000000 RDI: ffff99b5853b6008 > [ 9569.567140] RBP: ffffb092c0c97e10 R08: ffffffffc0bf3c24 R09: 0000000000000228 > [ 9569.568929] R10: ffff99b54b0e9268 R11: ffff99b564326998 R12: ffff99b5543dfc00 > [ 9569.570477] R13: ffff99b5a6882950 R14: ffff99b5a68829f0 R15: ffff99b546edc000 > [ 9569.572052] FS: 0000000000000000(0000) GS:ffff99b5bbe00000(0000) > knlGS:0000000000000000 > [ 9569.573926] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 > [ 9569.575281] CR2: 0000000000000000 CR3: 0000000076c36002 CR4: 00000000001706f0 > [ 9569.577586] Call Trace: > [ 9569.578220] <TASK> > [ 9569.578770] ? nfsd4_proc_compound+0x3d2/0x730 [nfsd] > [ 9569.579945] nfsd4_proc_compound+0x3d2/0x730 [nfsd] > [ 9569.581055] nfsd_dispatch+0x146/0x270 [nfsd] > [ 9569.581987] svc_process_common+0x365/0x5c0 [sunrpc] > [ 9569.583122] ? nfsd_svc+0x350/0x350 [nfsd] > [ 9569.583986] ? nfsd_shutdown_threads+0x90/0x90 [nfsd] > [ 9569.585129] svc_process+0xb7/0xf0 [sunrpc] > [ 9569.586169] nfsd+0xd5/0x190 [nfsd] > [ 9569.587170] kthread+0xe8/0x110 > [ 9569.587898] ? kthread_complete_and_exit+0x20/0x20 > [ 9569.588934] ret_from_fork+0x22/0x30 > [ 9569.589759] </TASK> > [ 9569.590224] Modules linked in: rdma_ucm ib_uverbs rpcrdma rdma_cm > iw_cm ib_cm ib_core nfsd nfs_acl lockd grace ext4 mbcache jbd2 fuse > xt_conntrack nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 ipt_REJECT > nf_reject_ipv4 nft_compat nf_tables nfnetlink tun bridge stp llc bnep > vmw_vsock_vmci_transport vsock snd_seq_midi snd_seq_midi_event > intel_rapl_msr intel_rapl_common crct10dif_pclmul crc32_pclmul > vmw_balloon ghash_clmulni_intel joydev pcspkr btusb btrtl btbcm > btintel snd_ens1371 uvcvideo snd_ac97_codec videobuf2_vmalloc ac97_bus > videobuf2_memops videobuf2_v4l2 videobuf2_common snd_seq snd_pcm > videodev bluetooth mc rfkill ecdh_generic ecc snd_timer snd_rawmidi > snd_seq_device snd vmw_vmci soundcore i2c_piix4 auth_rpcgss sunrpc > ip_tables xfs libcrc32c sr_mod cdrom sg ata_generic crc32c_intel > ata_piix nvme ahci libahci nvme_core t10_pi crc64_rocksoft serio_raw > crc64 vmwgfx drm_ttm_helper ttm drm_kms_helper syscopyarea sysfillrect > sysimgblt fb_sys_fops vmxnet3 drm libata > [ 9569.610612] CR2: 0000000000000000 > [ 9569.611375] ---[ end trace 0000000000000000 ]--- > [ 9569.612424] RIP: 0010:nfsd4_copy+0x28b/0x4e0 [nfsd] > [ 9569.613472] Code: 24 38 49 89 94 24 10 01 00 00 49 8b 56 08 48 8d > 79 08 49 89 94 24 18 01 00 00 49 8b 56 10 48 83 e7 f8 49 89 94 24 20 > 01 00 00 <48> 8b 06 48 89 01 48 8b 86 04 04 00 00 48 89 81 04 04 00 00 > 48 29 > [ 9569.617410] RSP: 0018:ffffb092c0c97dd0 EFLAGS: 00010282 > [ 9569.618487] RAX: ffff99b5465c2460 RBX: ffff99b5a68828e0 RCX: ffff99b5853b6000 > [ 9569.620097] RDX: 0000000000000001 RSI: 0000000000000000 RDI: ffff99b5853b6008 > [ 9569.621710] RBP: ffffb092c0c97e10 R08: ffffffffc0bf3c24 R09: 0000000000000228 > [ 9569.623398] R10: ffff99b54b0e9268 R11: ffff99b564326998 R12: ffff99b5543dfc00 > [ 9569.625019] R13: ffff99b5a6882950 R14: ffff99b5a68829f0 R15: ffff99b546edc000 > [ 9569.627456] FS: 0000000000000000(0000) GS:ffff99b5bbe00000(0000) > knlGS:0000000000000000 > [ 9569.629249] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 > [ 9569.630433] CR2: 0000000000000000 CR3: 0000000076c36002 CR4: 00000000001706f0 > [ 9569.632043] Kernel panic - not syncing: Fatal exception > > > > On Tue, Jul 26, 2022 at 3:45 PM Olga Kornievskaia <aglo@xxxxxxxxx> wrote: >> >> Chuck, >> >> Are there pre-reqs for this series? I had tried to apply the patches >> on top of 5-19-rc6 but I get the following compile error: >> >> fs/nfsd/nfs4proc.c: In function ‘nfsd4_setup_inter_ssc’: >> fs/nfsd/nfs4proc.c:1539:34: error: passing argument 1 of >> ‘nfsd4_interssc_connect’ from incompatible pointer type >> [-Werror=incompatible-pointer-types] >> status = nfsd4_interssc_connect(©->cp_src, rqstp, mount); >> ^~~~~~~~~~~~~ >> fs/nfsd/nfs4proc.c:1414:43: note: expected ‘struct nl4_server *’ but >> argument is of type ‘struct nl4_server **’ >> nfsd4_interssc_connect(struct nl4_server *nss, struct svc_rqst *rqstp, >> ~~~~~~~~~~~~~~~~~~~^~~ >> cc1: some warnings being treated as errors >> make[2]: *** [scripts/Makefile.build:249: fs/nfsd/nfs4proc.o] Error 1 >> make[1]: *** [scripts/Makefile.build:466: fs/nfsd] Error 2 >> make: *** [Makefile:1843: fs] Error 2 >> >> On Fri, Jul 22, 2022 at 4:36 PM Chuck Lever <chuck.lever@xxxxxxxxxx> wrote: >>> >>> While testing NFSD for-next, I noticed svc_generic_init_request() >>> was an unexpected hot spot on NFSv4 workloads. Drilling into the >>> perf report, it shows that the hot path in there is: >>> >>> 1208 memset(rqstp->rq_argp, 0, procp->pc_argsize); >>> 1209 memset(rqstp->rq_resp, 0, procp->pc_ressize); >>> >>> For an NFSv4 COMPOUND, >>> >>> procp->pc_argsize = sizeof(nfsd4_compoundargs), >>> >>> struct nfsd4_compoundargs on my system is more than 17KB! This is >>> due to the size of the iops field: >>> >>> struct nfsd4_op iops[8]; >>> >>> Each struct nfsd4_op contains a union of the arguments for each >>> NFSv4 operation. Each argument is typically less than 128 bytes >>> except that struct nfsd4_copy and struct nfsd4_copy_notify are both >>> larger than 2KB each. >>> >>> I'm not yet totally convinced this series never orphans memory, but >>> it does reduce the size of nfsd4_compoundargs to just over 4KB. This >>> is still due to struct nfsd4_copy being almost 500 bytes. I don't >>> see more low-hanging fruit there, though. >>> >>> --- >>> >>> Chuck Lever (11): >>> NFSD: Shrink size of struct nfsd4_copy_notify >>> NFSD: Shrink size of struct nfsd4_copy >>> NFSD: Reorder the fields in struct nfsd4_op >>> NFSD: Make nfs4_put_copy() static >>> NFSD: Make boolean fields in struct nfsd4_copy into atomic bit flags >>> NFSD: Refactor nfsd4_cleanup_inter_ssc() (1/2) >>> NFSD: Refactor nfsd4_cleanup_inter_ssc() (2/2) >>> NFSD: Refactor nfsd4_do_copy() >>> NFSD: Remove kmalloc from nfsd4_do_async_copy() >>> NFSD: Add nfsd4_send_cb_offload() >>> NFSD: Move copy offload callback arguments into a separate structure >>> >>> >>> fs/nfsd/nfs4callback.c | 37 +++++---- >>> fs/nfsd/nfs4proc.c | 165 +++++++++++++++++++++-------------------- >>> fs/nfsd/nfs4xdr.c | 30 +++++--- >>> fs/nfsd/state.h | 1 - >>> fs/nfsd/xdr4.h | 54 ++++++++++---- >>> 5 files changed, 163 insertions(+), 124 deletions(-) >>> >>> -- >>> Chuck Lever >>> -- Chuck Lever
