Re: Kernel ops with flexfiles

["Mkrtchyan, Tigran" <tigran.mkrtchyan@desy.de>]

Hi Trond,

----- Original Message -----
> From: "Trond Myklebust" <trondmy@primarydata.com>
> To: "Tigran Mkrtchyan" <tigran.mkrtchyan@desy.de>, "linux-nfs" <linux-nfs=
@vger.kernel.org>
> Sent: Wednesday, March 7, 2018 9:09:24 PM
> Subject: Re: Kernel ops with flexfiles

> Hi Tigran,
>=20
> On Mon, 2018-03-05 at 23:26 +0100, Mkrtchyan, Tigran wrote:
>> Hi Trond et al.
>>=20
>> looks like I can always reproduce the following ops with a simple cp:
>>=20
>> [10722.729463] nfs4flexfilelayout_init: NFSv4 Flexfile Layout Driver
>> Registering...
>> [10736.187403] ------------[ cut here ]------------
>> [10736.187405] Kernel BUG at 00000000f71645f7 [verbose debug info
>> unavailable]
>> [10736.187417] ------------[ cut here ]------------
>> [10736.187432] refcount_t hit zero at
>> pnfs_layout_remove_lseg+0x46/0x90 [nfsv4] in kworker/1:0[16409],
>> uid/euid: 0/0
>> [10736.187436] WARNING: CPU: 1 PID: 16409 at kernel/panic.c:657
>> refcount_error_report+0x94/0x9e
>> [10736.187436] Modules linked in: nfs_layout_flexfiles
>> rpcsec_gss_krb5 auth_rpcgss nfsv4 dns_resolver nfs lockd grace
>> fscache xt_nat veth ipt_MASQUERADE nf_nat_masquerade_ipv4 xt_addrtype
>> br_netfilter overlay tun rfcomm fuse ccm nf_conntrack_netbios_ns
>> nf_conntrack_broadcast xt_CT ip6t_rpfilter ip6t_REJECT nf_reject_ipv6
>> xt_conntrack xt_multiport ip_set nfnetlink ebtable_nat ebtable_broute
>> bridge stp llc ip6table_nat nf_conntrack_ipv6 nf_defrag_ipv6
>> nf_nat_ipv6 ip6table_mangle ip6table_raw ip6table_security
>> iptable_nat nf_conntrack_ipv4 nf_defrag_ipv4 nf_nat_ipv4 nf_nat
>> nf_conntrack libcrc32c iptable_mangle iptable_raw iptable_security
>> ebtable_filter ebtables ip6table_filter ip6_tables cmac binfmt_misc
>> bnep sunrpc vfat fat arc4 snd_soc_skl snd_hda_codec_hdmi
>> snd_soc_skl_ipc snd_hda_ext_core snd_soc_sst_dsp
>> [10736.187463]  snd_soc_sst_ipc snd_soc_acpi uvcvideo snd_soc_core
>> videobuf2_vmalloc btusb btrtl intel_rapl btbcm btintel
>> videobuf2_memops x86_pkg_temp_thermal intel_powerclamp videobuf2_v4l2
>> bluetooth snd_hda_codec_realtek videobuf2_core iwlmvm
>> snd_hda_codec_generic coretemp videodev kvm_intel snd_compress
>> snd_pcm_dmaengine ac97_bus mac80211 media snd_hda_intel kvm
>> snd_hda_codec iTCO_wdt iTCO_vendor_support iwlwifi mei_wdt dell_wmi
>> ecdh_generic wmi_bmof dell_smbios_wmi dell_laptop sparse_keymap
>> snd_hda_core dell_wmi_descriptor ppdev dell_smbios_smm dell_smbios
>> dcdbas snd_hwdep irqbypass crct10dif_pclmul dell_smm_hwmon
>> crc32_pclmul snd_seq cfg80211 snd_seq_device ghash_clmulni_intel
>> snd_pcm intel_cstate intel_uncore i2c_i801 intel_rapl_perf snd_timer
>> joydev rtsx_pci_ms memstick snd mei_me soundcore
>> [10736.187491]  mei processor_thermal_device shpchp
>> intel_soc_dts_iosf intel_pch_thermal wmi parport_pc parport dell_rbtn
>> int3400_thermal acpi_thermal_rel acpi_pad int3403_thermal rfkill
>> int340x_thermal_zone i915 rtsx_pci_sdmmc mmc_core i2c_algo_bit
>> drm_kms_helper drm e1000e crc32c_intel serio_raw rtsx_pci ptp
>> pps_core video
>> [10736.187504] CPU: 1 PID: 16409 Comm: kworker/1:0 Tainted:
>> G        W        4.15.6-300.fc27.x86_64 #1
>> [10736.187505] Hardware name: Dell Inc. Latitude E7470/0T6HHJ, BIOS
>> 1.6.3 06/15/2016
>> [10736.187518] Workqueue: nfsiod rpc_async_release [sunrpc]
>> [10736.187520] RIP: 0010:refcount_error_report+0x94/0x9e
>> [10736.187521] RSP: 0018:ffffa5c6d196fac0 EFLAGS: 00010282
>> [10736.187522] RAX: 0000000000000000 RBX: ffffffff93088273 RCX:
>> 0000000000000006
>> [10736.187523] RDX: 0000000000000007 RSI: 0000000000000082 RDI:
>> ffff9a780dc968f0
>> [10736.187523] RBP: ffffa5c6d196fc08 R08: 0000000000000651 R09:
>> 0000000000000004
>> [10736.187524] R10: ffffffff93206a80 R11: 0000000000000001 R12:
>> ffff9a776fe63f80
>> [10736.187525] R13: 0000000000000000 R14: ffffffff930795b5 R15:
>> 0000000000000004
>> [10736.187526] FS:  0000000000000000(0000) GS:ffff9a780dc80000(0000)
>> knlGS:0000000000000000
>> [10736.187527] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>> [10736.187528] CR2: 000000000000000c CR3: 000000036120a005 CR4:
>> 00000000003606e0
>> [10736.187529] DR0: 0000000000000000 DR1: 0000000000000000 DR2:
>> 0000000000000000
>> [10736.187529] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7:
>> 0000000000000400
>> [10736.187530] Call Trace:
>> [10736.187534]  ex_handler_refcount+0x4e/0x80
>> [10736.187536]  fixup_exception+0x33/0x40
>> [10736.187538]  do_trap+0x83/0x140
>> [10736.187540]  do_error_trap+0x9d/0x120
>> [10736.187550]  ? nfs42_proc_clone+0x2c7/0x309 [nfsv4]
>> [10736.187552]  ? __update_load_avg_se.isra.30+0x1b6/0x1c0
>> [10736.187553]  ? __update_load_avg_se.isra.30+0x1b6/0x1c0
>> [10736.187555]  ? update_load_avg+0x558/0x6b0
>> [10736.187557]  invalid_op+0x51/0x70
>> [10736.187566] RIP: 0010:pnfs_layout_remove_lseg+0x46/0x90 [nfsv4]
>> [10736.187566] RSP: 0018:ffffa5c6d196fcb8 EFLAGS: 00010246
>> [10736.187567] RAX: ffff9a765cc07f28 RBX: ffff9a757fe5f000 RCX:
>> ffff9a765cc07f00
>> [10736.187568] RDX: ffff9a765cc07f28 RSI: ffff9a757fe5f000 RDI:
>> ffff9a757fe5f000
>> [10736.187569] RBP: ffff9a765cc07f00 R08: 0000000000000004 R09:
>> 0fb19d5a01000000
>> [10736.187570] R10: ffffa5c6d196fda8 R11: ffffffffc10e3470 R12:
>> ffff9a765cc07f01
>> [10736.187570] R13: ffff9a756cc2d8c0 R14: 0000000000000000 R15:
>> ffff9a765cc07f00
>> [10736.187579]  ? pnfs_layout_remove_lseg+0x1d/0x90 [nfsv4]
>> [10736.187585]  pnfs_lseg_dec_and_remove_zero+0x37/0x70 [nfsv4]
>> [10736.187593]  mark_lseg_invalid+0x29/0x50 [nfsv4]
>> [10736.187600]  pnfs_roc+0x1ad/0x310 [nfsv4]
>> [10736.187606]  ? nfs4_do_close+0x1c9/0x2e0 [nfsv4]
>> [10736.187611]  nfs4_do_close+0x1c9/0x2e0 [nfsv4]
>> [10736.187620]  __put_nfs_open_context+0x7c/0x100 [nfs]
>> [10736.187627]  nfs_commitdata_release+0x15/0x30 [nfs]
>> [10736.187634]  rpc_free_task+0x2d/0x70 [sunrpc]
>> [10736.187637]  process_one_work+0x175/0x390
>> [10736.187640]  worker_thread+0x2e/0x380
>> [10736.187641]  ? process_one_work+0x390/0x390
>> [10736.187644]  kthread+0x113/0x130
>> [10736.187645]  ? kthread_create_worker_on_cpu+0x70/0x70
>> [10736.187647]  ? kthread_create_worker_on_cpu+0x70/0x70
>> [10736.187648]  ret_from_fork+0x35/0x40
>> [10736.187649] Code: 48 8b 95 80 00 00 00 41 55 49 8d 8c 24 48 07 00
>> 00 45 8b 84 24 68 05 00 00 41 89 c1 48 89 de 48 c7 c7 c0 c7 08 93 e8
>> 4c fa ff ff <0f> 0b 58 5b 5d 41 5c 41 5d c3 0f 1f 44 00 00 55 48 89
>> e5 41 56
>> [10736.187673] ---[ end trace ae865330f8bfd4f0 ]---
>> [10736.187699] ------------[ cut here ]------------
>> [10736.187700] Kernel BUG at 00000000c11ed4b1 [verbose debug info
>> unavailable]
>>=20
>>=20
>> If I downgrade my kernel to any of 4.14 - ops is gone. The 4.16-rc4
>> panics as well.
>> Let me know if you need more info or testing.
>=20
> Does the following patch fix it?


Looks good.

Tested-by: ...

Tigran.


> 8<--------------------------------------------------------
> From a5c81f5040a9e986eafb728719dfab2d588fe2d0 Mon Sep 17 00:00:00 2001
> From: Trond Myklebust <trond.myklebust@primarydata.com>
> Date: Wed, 7 Mar 2018 14:49:06 -0500
> Subject: [PATCH] pNFS: Prevent the layout header refcount going to zero i=
n
> pnfs_roc()
>=20
> Ensure that we hold a reference to the layout header when processing
> the pNFS return-on-close so that the refcount value does not inadvertentl=
y
> go to zero.
>=20
> Reported-by: Tigran Mkrtchyan <tigran.mkrtchyan@desy.de>
> Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
> Cc: stable@vger.kernel.org # v4.10+
> ---
> fs/nfs/pnfs.c | 13 ++++++++++---
> 1 file changed, 10 insertions(+), 3 deletions(-)
>=20
> diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
> index c13e826614b5..ee723aa153a3 100644
> --- a/fs/nfs/pnfs.c
> +++ b/fs/nfs/pnfs.c
> @@ -292,8 +292,11 @@ pnfs_detach_layout_hdr(struct pnfs_layout_hdr *lo)
> void
> pnfs_put_layout_hdr(struct pnfs_layout_hdr *lo)
> {
> -=09struct inode *inode =3D lo->plh_inode;
> +=09struct inode *inode;
>=20
> +=09if (!lo)
> +=09=09return;
> +=09inode =3D lo->plh_inode;
> =09pnfs_layoutreturn_before_put_layout_hdr(lo);
>=20
> =09if (refcount_dec_and_lock(&lo->plh_refcount, &inode->i_lock)) {
> @@ -1241,10 +1244,12 @@ bool pnfs_roc(struct inode *ino,
> =09spin_lock(&ino->i_lock);
> =09lo =3D nfsi->layout;
> =09if (!lo || !pnfs_layout_is_valid(lo) ||
> -=09    test_bit(NFS_LAYOUT_BULK_RECALL, &lo->plh_flags))
> +=09    test_bit(NFS_LAYOUT_BULK_RECALL, &lo->plh_flags)) {
> +=09=09lo =3D NULL;
> =09=09goto out_noroc;
> +=09}
> +=09pnfs_get_layout_hdr(lo);
> =09if (test_bit(NFS_LAYOUT_RETURN_LOCK, &lo->plh_flags)) {
> -=09=09pnfs_get_layout_hdr(lo);
> =09=09spin_unlock(&ino->i_lock);
> =09=09wait_on_bit(&lo->plh_flags, NFS_LAYOUT_RETURN,
> =09=09=09=09TASK_UNINTERRUPTIBLE);
> @@ -1312,10 +1317,12 @@ bool pnfs_roc(struct inode *ino,
> =09=09struct pnfs_layoutdriver_type *ld =3D NFS_SERVER(ino)->pnfs_curr_ld=
;
> =09=09if (ld->prepare_layoutreturn)
> =09=09=09ld->prepare_layoutreturn(args);
> +=09=09pnfs_put_layout_hdr(lo);
> =09=09return true;
> =09}
> =09if (layoutreturn)
> =09=09pnfs_send_layoutreturn(lo, &stateid, iomode, true);
> +=09pnfs_put_layout_hdr(lo);
> =09return false;
> }
>=20
> --
> 2.14.3
>=20
> --
> Trond Myklebust
> Linux NFS client maintainer, PrimaryData
> trond.myklebust@primarydata.com
> N=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BDr=EF=BF=BD=EF=BF=BDy=EF=BF=
=BD=EF=BF=BD=EF=BF=BDb=EF=BF=BDX=EF=BF=BD=EF=BF=BD=C7=A7v=EF=BF=BD^=EF=BF=
=BD)=DE=BA{.n=EF=BF=BD+=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD{=EF=BF=BD=EF=BF=
=BD=EF=BF=BD"=EF=BF=BD=EF=BF=BD^n=EF=BF=BDr=EF=BF=BD=EF=BF=BD=EF=BF=BDz=EF=
=BF=BD=EF=BF=BD=EF=BF=BDh=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=BD&=EF=BF=BD=EF=
=BF=BD=EF=BF=BDG=EF=BF=BD=EF=BF=BD=EF=BF=BDh=EF=BF=BD(=EF=BF=BD=E9=9A=8E=EF=
=BF=BD=DD=A2j"=EF=BF=BD=EF=BF=BD=EF=BF=BDm=EF=BF=BD=EF=BF=BD=EF=BF=BD=EF=BF=
=BD=EF=BF=BDz=EF=BF=BD=DE=96=EF=BF=BD=EF=BF=BD=EF=BF=BDf=EF=BF=BD=EF=BF=BD=
=EF=BF=BDh=EF=BF=BD=EF=BF=BD=EF=BF=BD~=EF=BF=BDm=EF=BF=BD