Received-SPF: pass (google.com: domain of linux-nfs-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Mon, 13 Jul 2020 09:52:20 -0400
From:   Scott Mayhew <smayhew@redhat.com>
To:     Chuck Lever <chuck.lever@oracle.com>
Cc:     Bruce Fields <bfields@fieldses.org>,
        Linux NFS Mailing List <linux-nfs@vger.kernel.org>
Subject: Re: [PATCH] nfsd: avoid a NULL dereference in __cld_pipe_upcall()
Message-ID: <20200713135220.GB4452@aion.usersys.redhat.com>
References: <20200710203307.2545412-1-smayhew@redhat.com>
 <B6722AC5-A7DD-4EAF-B08A-45C12160D5DF@oracle.com>
 <20200713122955.GY4452@aion.usersys.redhat.com>
 <646E9DB3-0E48-4C2D-A921-69615EAEC940@oracle.com>
MIME-Version: 1.0
In-Reply-To: <646E9DB3-0E48-4C2D-A921-69615EAEC940@oracle.com>
Content-Type: text/plain; charset=WINDOWS-1252
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline
Sender: linux-nfs-owner@vger.kernel.org
Precedence: bulk

On Mon, 13 Jul 2020, Chuck Lever wrote:

>=20
>=20
> > On Jul 13, 2020, at 8:29 AM, Scott Mayhew <smayhew@redhat.com> wrote:
> >=20
> > On Sun, 12 Jul 2020, Chuck Lever wrote:
> >=20
> >> Hi Scott-
> >>=20
> >>> On Jul 10, 2020, at 4:33 PM, Scott Mayhew <smayhew@redhat.com> wrote:
> >>>=20
> >>> If the rpc_pipefs is unmounted, then the rpc_pipe->dentry becomes NUL=
L
> >>> and dereferencing the dentry->d_sb will trigger an oops.  The only
> >>> reason we're doing that is to determine the nfsd_net, which could
> >>> instead be passed in by the caller.  So do that instead.
> >>>=20
> >>> Signed-off-by: Scott Mayhew <smayhew@redhat.com>
> >>=20
> >> Looks straightforward. Applied to the nfsd-5.9 testing tree.
> >>=20
> >> I'm wondering about automatic backports to stable. This fix does not
> >> apply to kernels before v5.6, but IIUC addresses a crash introduced
> >> in f3f8014862d8 ("nfsd: add the infrastructure to handle the cld upcal=
l")
> >> [2012] ?
> >>=20
> >> Is "Cc: <stable@kernel.org> # v5.6+" appropriate?
> >>=20
> > I think it would be 11a60d159259 ("nfsd: add a "GetVersion" upcall for
> > nfsdcld"), so it would only need to go back to v5.4... and to get the
> > patch to apply, in the hunk that modifies nfsd4_cld_grace_done_v0()
> > you would need to add a cast of nn->boot_time to int64_t (which
> > 9cc7680149b2 "nfsd: make 'boot_time' 64-bit wide" removed).
>=20
> Thanks for the detailed background!
>=20
> Then instead of "Cc: stable", can I add:
>=20
> Fixes: 11a60d159259 ("nfsd: add a "GetVersion" upcall for nfsdcld") ?

Yes, please do that.

>=20
>=20
> >> Also, is there a bug report that documents the crash?
> >=20
> > Our QE hit the crash internally while running xfstests on a pNFS SCSI
> > setup.  I don't think either of those is relevant aside from the fact
> > that several nfsd threads where stuck on xfs while calling
> > nfsd4_scsi_proc_layoutcommit().  I think what happened is that the job
> > timed out and the system was in the process of being shut down when the
> > oops occurred, and the rpc_pipefs was unmounted out from under nfsd:
> >=20
> > crash> bt
> > PID: 39572  TASK: ffff8a78f67fddc0  CPU: 1   COMMAND: "kworker/u4:0"
> > #0 [ffffa39a026b3ae0] machine_kexec at ffffffffae65b55e
> > #1 [ffffa39a026b3b38] __crash_kexec at ffffffffae75ea5d
> > #2 [ffffa39a026b3c00] crash_kexec at ffffffffae75f93d
> > #3 [ffffa39a026b3c18] oops_end at ffffffffae622c4d
> > #4 [ffffa39a026b3c38] no_context at ffffffffae66aa2e
> > #5 [ffffa39a026b3c90] do_page_fault at ffffffffae66b552
> > #6 [ffffa39a026b3cc0] async_page_fault at ffffffffaf00123e
> >    [exception RIP: __cld_pipe_upcall+0x3d]
> >    RIP: ffffffffc06b5ded  RSP: ffffa39a026b3d70  RFLAGS: 00010246
> >    RAX: 0000000000000000  RBX: ffff8a78ceaf7000  RCX: 000000000000000b
> >    RDX: ffff8a78ceaf7000  RSI: ffffa39a026b3d70  RDI: ffffa39a026b3d70
> >    RBP: ffffa39a026b3dc0   R8: ffff8a78f37aac50   R9: ffff8a78c7c02800
> >    R10: 8080808080808080  R11: 0000ec193f7575c0  R12: ffff8a78c8ef3038
> >    R13: ffff8a78d3156220  R14: ffffa39a026b3e50  R15: ffff8a78d3156220
> >    ORIG_RAX: ffffffffffffffff  CS: 0010  SS: 0018
> > #7 [ffffa39a026b3dc8] nfsd4_cld_remove at ffffffffc06b7b70 [nfsd]
> > #8 [ffffa39a026b3df8] expire_client at ffffffffc06aa4d6 [nfsd]
> > #9 [ffffa39a026b3e08] laundromat_main at ffffffffc06aa799 [nfsd]
> > #10 [ffffa39a026b3e98] process_one_work at ffffffffae6d1cf7
> > #11 [ffffa39a026b3ed8] worker_thread at ffffffffae6d23c0
> > #12 [ffffa39a026b3f10] kthread at ffffffffae6d7d02
> > #13 [ffffa39a026b3f50] ret_from_fork at ffffffffaf000255
> >=20
> > crash> dis -lr __cld_pipe_upcall+0x3d
> > ...
> > /usr/src/debug/kernel-4.18.0-211.el8/linux-4.18.0-211.el8.x86_64/fs/nfs=
d/nfs4recover.c: 764
> > 0xffffffffc06b5de0 <__cld_pipe_upcall+0x30>:    mov    0x108(%rdi),%rax
> > 0xffffffffc06b5de7 <__cld_pipe_upcall+0x37>:    mov    %rsp,%rsi
> > 0xffffffffc06b5dea <__cld_pipe_upcall+0x3a>:    mov    %rsi,%rdi
> > 0xffffffffc06b5ded <__cld_pipe_upcall+0x3d>:    mov    0x68(%rax),%rax
> >=20
> > That puts us here:
> >=20
> > static int
> > __cld_pipe_upcall(struct rpc_pipe *pipe, void *cmsg)
> > {
> >        int ret;
> >        struct rpc_pipe_msg msg;
> >        struct cld_upcall *cup =3D container_of(cmsg, struct cld_upcall,=
 cu_u);
> >        struct nfsd_net *nn =3D net_generic(pipe->dentry->d_sb->s_fs_inf=
o,
> >                                          nfsd_net_id);
> > ...
> > crash> rpc_pipe.dentry -o
> > struct rpc_pipe {
> >  [0x108] struct dentry *dentry;
> > }
> > crash> dentry.d_sb -o
> > struct dentry {
> >  [0x68] struct super_block *d_sb;
> > }
> >=20
> > If the rpc_pipe->dentry was NULL, that would explain the NULL pointer d=
ereference at 0000000000000068.  Let's confirm.
> >=20
> > crash> dis -lr ffffffffc06b7b70
> > ...
> > /usr/src/debug/kernel-4.18.0-211.el8/linux-4.18.0-211.el8.x86_64/fs/nfs=
d/nfs4recover.c: 794
> > 0xffffffffc06b7b65 <nfsd4_cld_remove+0x85>:     mov    %rbp,%rsi
> > 0xffffffffc06b7b68 <nfsd4_cld_remove+0x88>:     mov    %rbx,%rdi <-----=
-------------------------rpc_pipe
> > 0xffffffffc06b7b6b <nfsd4_cld_remove+0x8b>:     callq  0xffffffffc06b5d=
b0 <__cld_pipe_upcall>
> > /usr/src/debug/kernel-4.18.0-211.el8/linux-4.18.0-211.el8.x86_64/fs/nfs=
d/nfs4recover.c: 795
> > 0xffffffffc06b7b70 <nfsd4_cld_remove+0x90>:     cmp    $0xfffffff5,%eax
> >=20
> > crash> rpc_pipe.dentry ffff8a78ceaf7000
> >  dentry =3D 0x0
> >=20
> > We can also see that the rpc_pipefs isn't mounted:
> >=20
> > crash> mount|grep pipefs
> > (nothing)
> >=20
> > And none of the daemons that require the rpc_pipefs are running:
> >=20
> > crash> ps|grep nfsdcld
> > crash> ps|grep idmapd
> > crash> ps|grep gssd
> > crash> ps|grep blkmapd
> > (nothing)
> >=20
> > I think part of the problem is that the nfsdcld.service unit file has
> > "Requires=3Drpc_pipefs.target", but the nfs-server.service unit file on=
ly
> > has "Wants=3Dnfsdcld.service".  It can't be "Requires" because using th=
e
> > nfsdcld daemon for client tracking is optional.
> >=20
> > -Scott
> >=20
> >>=20
> >>=20
> >>> ---
> >>> fs/nfsd/nfs4recover.c | 24 +++++++++++-------------
> >>> 1 file changed, 11 insertions(+), 13 deletions(-)
> >>>=20
> >>> diff --git a/fs/nfsd/nfs4recover.c b/fs/nfsd/nfs4recover.c
> >>> index 9e40dfecf1b1..186fa2c2c6ba 100644
> >>> --- a/fs/nfsd/nfs4recover.c
> >>> +++ b/fs/nfsd/nfs4recover.c
> >>> @@ -747,13 +747,11 @@ struct cld_upcall {
> >>> };
> >>>=20
> >>> static int
> >>> -__cld_pipe_upcall(struct rpc_pipe *pipe, void *cmsg)
> >>> +__cld_pipe_upcall(struct rpc_pipe *pipe, void *cmsg, struct nfsd_net=
 *nn)
> >>> {
> >>> =09int ret;
> >>> =09struct rpc_pipe_msg msg;
> >>> =09struct cld_upcall *cup =3D container_of(cmsg, struct cld_upcall, c=
u_u);
> >>> -=09struct nfsd_net *nn =3D net_generic(pipe->dentry->d_sb->s_fs_info=
,
> >>> -=09=09=09=09=09  nfsd_net_id);
> >>>=20
> >>> =09memset(&msg, 0, sizeof(msg));
> >>> =09msg.data =3D cmsg;
> >>> @@ -773,7 +771,7 @@ __cld_pipe_upcall(struct rpc_pipe *pipe, void *cm=
sg)
> >>> }
> >>>=20
> >>> static int
> >>> -cld_pipe_upcall(struct rpc_pipe *pipe, void *cmsg)
> >>> +cld_pipe_upcall(struct rpc_pipe *pipe, void *cmsg, struct nfsd_net *=
nn)
> >>> {
> >>> =09int ret;
> >>>=20
> >>> @@ -782,7 +780,7 @@ cld_pipe_upcall(struct rpc_pipe *pipe, void *cmsg=
)
> >>> =09 *  upcalls queued.
> >>> =09 */
> >>> =09do {
> >>> -=09=09ret =3D __cld_pipe_upcall(pipe, cmsg);
> >>> +=09=09ret =3D __cld_pipe_upcall(pipe, cmsg, nn);
> >>> =09} while (ret =3D=3D -EAGAIN);
> >>>=20
> >>> =09return ret;
> >>> @@ -1115,7 +1113,7 @@ nfsd4_cld_create(struct nfs4_client *clp)
> >>> =09memcpy(cup->cu_u.cu_msg.cm_u.cm_name.cn_id, clp->cl_name.data,
> >>> =09=09=09clp->cl_name.len);
> >>>=20
> >>> -=09ret =3D cld_pipe_upcall(cn->cn_pipe, &cup->cu_u.cu_msg);
> >>> +=09ret =3D cld_pipe_upcall(cn->cn_pipe, &cup->cu_u.cu_msg, nn);
> >>> =09if (!ret) {
> >>> =09=09ret =3D cup->cu_u.cu_msg.cm_status;
> >>> =09=09set_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags);
> >>> @@ -1180,7 +1178,7 @@ nfsd4_cld_create_v2(struct nfs4_client *clp)
> >>> =09} else
> >>> =09=09cmsg->cm_u.cm_clntinfo.cc_princhash.cp_len =3D 0;
> >>>=20
> >>> -=09ret =3D cld_pipe_upcall(cn->cn_pipe, cmsg);
> >>> +=09ret =3D cld_pipe_upcall(cn->cn_pipe, cmsg, nn);
> >>> =09if (!ret) {
> >>> =09=09ret =3D cmsg->cm_status;
> >>> =09=09set_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags);
> >>> @@ -1218,7 +1216,7 @@ nfsd4_cld_remove(struct nfs4_client *clp)
> >>> =09memcpy(cup->cu_u.cu_msg.cm_u.cm_name.cn_id, clp->cl_name.data,
> >>> =09=09=09clp->cl_name.len);
> >>>=20
> >>> -=09ret =3D cld_pipe_upcall(cn->cn_pipe, &cup->cu_u.cu_msg);
> >>> +=09ret =3D cld_pipe_upcall(cn->cn_pipe, &cup->cu_u.cu_msg, nn);
> >>> =09if (!ret) {
> >>> =09=09ret =3D cup->cu_u.cu_msg.cm_status;
> >>> =09=09clear_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags);
> >>> @@ -1261,7 +1259,7 @@ nfsd4_cld_check_v0(struct nfs4_client *clp)
> >>> =09memcpy(cup->cu_u.cu_msg.cm_u.cm_name.cn_id, clp->cl_name.data,
> >>> =09=09=09clp->cl_name.len);
> >>>=20
> >>> -=09ret =3D cld_pipe_upcall(cn->cn_pipe, &cup->cu_u.cu_msg);
> >>> +=09ret =3D cld_pipe_upcall(cn->cn_pipe, &cup->cu_u.cu_msg, nn);
> >>> =09if (!ret) {
> >>> =09=09ret =3D cup->cu_u.cu_msg.cm_status;
> >>> =09=09set_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags);
> >>> @@ -1404,7 +1402,7 @@ nfsd4_cld_grace_start(struct nfsd_net *nn)
> >>> =09}
> >>>=20
> >>> =09cup->cu_u.cu_msg.cm_cmd =3D Cld_GraceStart;
> >>> -=09ret =3D cld_pipe_upcall(cn->cn_pipe, &cup->cu_u.cu_msg);
> >>> +=09ret =3D cld_pipe_upcall(cn->cn_pipe, &cup->cu_u.cu_msg, nn);
> >>> =09if (!ret)
> >>> =09=09ret =3D cup->cu_u.cu_msg.cm_status;
> >>>=20
> >>> @@ -1432,7 +1430,7 @@ nfsd4_cld_grace_done_v0(struct nfsd_net *nn)
> >>>=20
> >>> =09cup->cu_u.cu_msg.cm_cmd =3D Cld_GraceDone;
> >>> =09cup->cu_u.cu_msg.cm_u.cm_gracetime =3D nn->boot_time;
> >>> -=09ret =3D cld_pipe_upcall(cn->cn_pipe, &cup->cu_u.cu_msg);
> >>> +=09ret =3D cld_pipe_upcall(cn->cn_pipe, &cup->cu_u.cu_msg, nn);
> >>> =09if (!ret)
> >>> =09=09ret =3D cup->cu_u.cu_msg.cm_status;
> >>>=20
> >>> @@ -1460,7 +1458,7 @@ nfsd4_cld_grace_done(struct nfsd_net *nn)
> >>> =09}
> >>>=20
> >>> =09cup->cu_u.cu_msg.cm_cmd =3D Cld_GraceDone;
> >>> -=09ret =3D cld_pipe_upcall(cn->cn_pipe, &cup->cu_u.cu_msg);
> >>> +=09ret =3D cld_pipe_upcall(cn->cn_pipe, &cup->cu_u.cu_msg, nn);
> >>> =09if (!ret)
> >>> =09=09ret =3D cup->cu_u.cu_msg.cm_status;
> >>>=20
> >>> @@ -1524,7 +1522,7 @@ nfsd4_cld_get_version(struct nfsd_net *nn)
> >>> =09=09goto out_err;
> >>> =09}
> >>> =09cup->cu_u.cu_msg.cm_cmd =3D Cld_GetVersion;
> >>> -=09ret =3D cld_pipe_upcall(cn->cn_pipe, &cup->cu_u.cu_msg);
> >>> +=09ret =3D cld_pipe_upcall(cn->cn_pipe, &cup->cu_u.cu_msg, nn);
> >>> =09if (!ret) {
> >>> =09=09ret =3D cup->cu_u.cu_msg.cm_status;
> >>> =09=09if (ret)
> >>> --=20
> >>> 2.25.4
> >>>=20
> >>=20
> >> --
> >> Chuck Lever
>=20
> --
> Chuck Lever
>=20
>=20
>=20