From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-1.web.codeaurora.org [10.30.226.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 619BD13CFA5 for ; Mon, 29 Jul 2024 12:26:17 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=10.30.226.201 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1722255977; cv=none; b=Mf8xfejW1rqJsKUYAQ4eMLtXobN9i/AV2KyrpcTCiJSar7m/1thjd+OY2hceVY+obY3Ytcu+QmgB+xPkQzepx8FnH+jwEUpd2WFVvEI6OrVqUhKZbdbGCV1B7igUivHV0EF9JXYublK/yjy2p5AZz5YyGB2y1LgbR/Pvt/4W+S8= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1722255977; c=relaxed/simple; bh=ZT4TU6bRrW63FhRpCGVJyyeP001X8RSO2uD3WJ3D8+Y=; h=Message-ID:Subject:From:To:Cc:Date:In-Reply-To:References: Content-Type:MIME-Version; b=fEbne+bIltk0TjqmV5zZ+V7ieJwOof/54AhWefb20XQ5Jjh8odw2g+gwU76vE+6Nyotedf0WtLiZzqgWl3gFzqupyR76cGcXBnttqIs2FBoV9pSKOx9796EkMlRdDKCNyvSA/dslsQYmd2O3OoHDpePb/9e8cHk1OyYJNBuIiTo= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b=m3T4Qx1x; arc=none smtp.client-ip=10.30.226.201 Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b="m3T4Qx1x" Received: by smtp.kernel.org (Postfix) with ESMTPSA id B14D3C32786; Mon, 29 Jul 2024 12:26:16 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=k20201202; t=1722255977; bh=ZT4TU6bRrW63FhRpCGVJyyeP001X8RSO2uD3WJ3D8+Y=; h=Subject:From:To:Cc:Date:In-Reply-To:References:From; b=m3T4Qx1xIn7viVwFxX7Auaqbugve8th8BIRYEEdCZ82eLu/civljC2i+/4xUo4MIy QLKKwInAj9F4+wEGhR05Bxqwj8ODU5vr4Tl2dM3VFBAwQVRT1HDFNg/ui6PLGiz7Wa Y33cNuFfSk10BCsc1cRSpbw63IqIfon1JOxfOmK7gpiqe5aH3bXPLFKxB0YRCH0L/K aN6NqMJcNV04DrN8crRDV9cuyE6jfMg9954b3ysZtte+hGSdf52XpXdMuqYensFNXo 76Gl8+Vdmlwb424Z8QWEdhXJuoP4kgf7J3cKqClpj5vyp180uVLlbomMdom4QuG+Nk ic70TGm34fbTQ== Message-ID: <78944d2e1ba2831eb6e46bd7009dfd5e176d48bb.camel@kernel.org> Subject: Re: [PATCH v2 2/3] nfsd: Offer write delegations for o_wronly opens From: Jeff Layton To: Sagi Grimberg , Chuck Lever Cc: Dai Ngo , Olga Kornievskaia , linux-nfs@vger.kernel.org Date: Mon, 29 Jul 2024 08:26:15 -0400 In-Reply-To: <20240728204104.519041-3-sagi@grimberg.me> References: <20240728204104.519041-1-sagi@grimberg.me> <20240728204104.519041-3-sagi@grimberg.me> Autocrypt: addr=jlayton@kernel.org; prefer-encrypt=mutual; keydata=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 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable User-Agent: Evolution 3.50.4 (3.50.4-1.fc39) Precedence: bulk X-Mailing-List: linux-nfs@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 On Sun, 2024-07-28 at 23:41 +0300, Sagi Grimberg wrote: > In order to support write delegations with O_WRONLY opens, we need to > allow the clients to read using the write delegation stateid (Per RFC > 8881 section 9.1.2. Use of the Stateid and Locking). >=20 > Hence, we check for NFS4_SHARE_ACCESS_WRITE set in open request, and > in case the share access flag does not set NFS4_SHARE_ACCESS_READ as > well, we'll open the file locally with O_RDWR in order to allow the > client to use the write delegation stateid when issuing a read in case > it may choose to. >=20 > Plus, find_rw_file singular call-site is now removed, remove it altogethe= r. >=20 > Note: reads using special stateids that conflict with pending write > delegations are undetected, and will be covered in a follow on patch. >=20 > Signed-off-by: Sagi Grimberg > --- > =C2=A0fs/nfsd/nfs4proc.c=C2=A0 | 18 +++++++++++++++++- > =C2=A0fs/nfsd/nfs4state.c | 42 ++++++++++++++++++++---------------------- > =C2=A0fs/nfsd/xdr4.h=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 |=C2=A0 2 ++ > =C2=A03 files changed, 39 insertions(+), 23 deletions(-) >=20 > diff --git a/fs/nfsd/nfs4proc.c b/fs/nfsd/nfs4proc.c > index 7b70309ad8fb..041bcc3ab5d7 100644 > --- a/fs/nfsd/nfs4proc.c > +++ b/fs/nfsd/nfs4proc.c > @@ -979,8 +979,22 @@ nfsd4_read(struct svc_rqst *rqstp, struct nfsd4_comp= ound_state *cstate, > =C2=A0 /* check stateid */ > =C2=A0 status =3D nfs4_preprocess_stateid_op(rqstp, cstate, &cstate->curr= ent_fh, > =C2=A0 &read->rd_stateid, RD_STATE, > - &read->rd_nf, NULL); > + &read->rd_nf, &read->rd_wd_stid); > =C2=A0 > + /* > + * rd_wd_stid is needed for nfsd4_encode_read to allow write > + * delegation stateid used for read. Its refcount is decremented > + * by nfsd4_read_release when read is done. > + */ > + if (!status) { > + if (read->rd_wd_stid && > + =C2=A0=C2=A0=C2=A0 (read->rd_wd_stid->sc_type !=3D SC_TYPE_DELEG || > + =C2=A0=C2=A0=C2=A0=C2=A0 delegstateid(read->rd_wd_stid)->dl_type !=3D > + NFS4_OPEN_DELEGATE_WRITE)) { > + nfs4_put_stid(read->rd_wd_stid); > + read->rd_wd_stid =3D NULL; > + } > + } > =C2=A0 read->rd_rqstp =3D rqstp; > =C2=A0 read->rd_fhp =3D &cstate->current_fh; > =C2=A0 return status; > @@ -990,6 +1004,8 @@ nfsd4_read(struct svc_rqst *rqstp, struct nfsd4_comp= ound_state *cstate, > =C2=A0static void > =C2=A0nfsd4_read_release(union nfsd4_op_u *u) > =C2=A0{ > + if (u->read.rd_wd_stid) > + nfs4_put_stid(u->read.rd_wd_stid); > =C2=A0 if (u->read.rd_nf) > =C2=A0 nfsd_file_put(u->read.rd_nf); > =C2=A0 trace_nfsd_read_done(u->read.rd_rqstp, u->read.rd_fhp, > diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c > index 0645fccbf122..538b6e1127a2 100644 > --- a/fs/nfsd/nfs4state.c > +++ b/fs/nfsd/nfs4state.c > @@ -639,18 +639,6 @@ find_readable_file(struct nfs4_file *f) > =C2=A0 return ret; > =C2=A0} > =C2=A0 > -static struct nfsd_file * > -find_rw_file(struct nfs4_file *f) > -{ > - struct nfsd_file *ret; > - > - spin_lock(&f->fi_lock); > - ret =3D nfsd_file_get(f->fi_fds[O_RDWR]); > - spin_unlock(&f->fi_lock); > - > - return ret; > -} > - > =C2=A0struct nfsd_file * > =C2=A0find_any_file(struct nfs4_file *f) > =C2=A0{ > @@ -5784,15 +5772,11 @@ nfs4_set_delegation(struct nfsd4_open *open, stru= ct nfs4_ol_stateid *stp, > =C2=A0 *=C2=A0 "An OPEN_DELEGATE_WRITE delegation allows the client to h= andle, > =C2=A0 *=C2=A0=C2=A0 on its own, all opens." > =C2=A0 * > - * Furthermore the client can use a write delegation for most READ > - * operations as well, so we require a O_RDWR file here. > - * > - * Offer a write delegation in the case of a BOTH open, and ensure > - * we get the O_RDWR descriptor. > + * Offer a write delegation for WRITE or BOTH access > =C2=A0 */ > - if ((open->op_share_access & NFS4_SHARE_ACCESS_BOTH) =3D=3D NFS4_SHARE_= ACCESS_BOTH) { > - nf =3D find_rw_file(fp); > + if (open->op_share_access & NFS4_SHARE_ACCESS_WRITE) { > =C2=A0 dl_type =3D NFS4_OPEN_DELEGATE_WRITE; > + nf =3D find_writeable_file(fp); > =C2=A0 } > =C2=A0 > =C2=A0 /* > @@ -5934,8 +5918,8 @@ static void nfsd4_open_deleg_none_ext(struct nfsd4_= open *open, int status) > =C2=A0 * open or lock state. > =C2=A0 */ > =C2=A0static void > -nfs4_open_delegation(struct nfsd4_open *open, struct nfs4_ol_stateid *st= p, > - =C2=A0=C2=A0=C2=A0=C2=A0 struct svc_fh *currentfh) > +nfs4_open_delegation(struct svc_rqst *rqstp, struct nfsd4_open *open, > + struct nfs4_ol_stateid *stp, struct svc_fh *currentfh) > =C2=A0{ > =C2=A0 struct nfs4_delegation *dp; > =C2=A0 struct nfs4_openowner *oo =3D openowner(stp->st_stateowner); > @@ -5994,6 +5978,20 @@ nfs4_open_delegation(struct nfsd4_open *open, stru= ct nfs4_ol_stateid *stp, > =C2=A0 dp->dl_cb_fattr.ncf_cur_fsize =3D stat.size; > =C2=A0 dp->dl_cb_fattr.ncf_initial_cinfo =3D > =C2=A0 nfsd4_change_attribute(&stat, d_inode(currentfh->fh_dentry)); > + if ((open->op_share_access & NFS4_SHARE_ACCESS_BOTH) !=3D NFS4_SHARE_A= CCESS_BOTH) { More comments on this part: nit: You've already tested for NFS4_SHARE_ACCESS_WRITE here, and this seems easier to read: if (!(open->op_share_access & NFS4_SHARE_ACCESS_READ)) > + struct nfsd_file *nf =3D NULL; > + > + /* make sure the file is opened locally for O_RDWR */ > + status =3D nfsd_file_acquire_opened(rqstp, currentfh, > + nfs4_access_to_access(NFS4_SHARE_ACCESS_BOTH), > + open->op_filp, &nf); > + if (status) { > + nfs4_put_stid(&dp->dl_stid); > + destroy_delegation(dp); > + goto out_no_deleg; > + } > + stp->st_stid.sc_file->fi_fds[O_RDWR] =3D nf; How do you know that this fd isn't already set? Also, this field is protected by the sc_file->fi_lock and that's not held here. > + } > =C2=A0 } else { > =C2=A0 open->op_delegate_type =3D NFS4_OPEN_DELEGATE_READ; > =C2=A0 trace_nfsd_deleg_read(&dp->dl_stid.sc_stateid); > @@ -6123,7 +6121,7 @@ nfsd4_process_open2(struct svc_rqst *rqstp, struct = svc_fh *current_fh, struct nf > =C2=A0 * Attempt to hand out a delegation. No error return, because the > =C2=A0 * OPEN succeeds even if we fail. > =C2=A0 */ > - nfs4_open_delegation(open, stp, &resp->cstate.current_fh); > + nfs4_open_delegation(rqstp, open, stp, &resp->cstate.current_fh); > =C2=A0nodeleg: > =C2=A0 status =3D nfs_ok; > =C2=A0 trace_nfsd_open(&stp->st_stid.sc_stateid); > diff --git a/fs/nfsd/xdr4.h b/fs/nfsd/xdr4.h > index fbdd42cde1fa..434973a6a8b1 100644 > --- a/fs/nfsd/xdr4.h > +++ b/fs/nfsd/xdr4.h > @@ -426,6 +426,8 @@ struct nfsd4_read { > =C2=A0 struct svc_rqst *rd_rqstp;=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2= =A0=C2=A0=C2=A0 /* response */ > =C2=A0 struct svc_fh *rd_fhp;=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0= =C2=A0=C2=A0=C2=A0=C2=A0 /* response */ > =C2=A0 u32 rd_eof;=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2= =A0=C2=A0=C2=A0=C2=A0 /* response */ > + > + struct nfs4_stid *rd_wd_stid; /* internal */ > =C2=A0}; > =C2=A0 > =C2=A0struct nfsd4_readdir { --=20 Jeff Layton