From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:35806) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1W8Xs0-0000Va-JZ for qemu-devel@nongnu.org; Wed, 29 Jan 2014 11:20:15 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1W8Xrw-0002o2-6o for qemu-devel@nongnu.org; Wed, 29 Jan 2014 11:20:08 -0500 Received: from paradis.irqsave.net ([62.212.105.220]:48636) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1W8Xrv-0002eS-Bj for qemu-devel@nongnu.org; Wed, 29 Jan 2014 11:20:04 -0500 Date: Wed, 29 Jan 2014 17:19:59 +0100 From: =?iso-8859-1?Q?Beno=EEt?= Canet Message-ID: <20140129161959.GG3079@irqsave.net> References: <1390985425-13165-1-git-send-email-pl@kamp.de> <1390985425-13165-2-git-send-email-pl@kamp.de> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline In-Reply-To: <1390985425-13165-2-git-send-email-pl@kamp.de> Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [PATCHv7 1/5] block: add native support for NFS List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Peter Lieven Cc: Kevin Wolf , Fam Zheng , Stefan Hajnoczi , Jeff Cody , qemu-devel@nongnu.org, Max Reitz , owasserm@redhat.com, Federico Simoncelli , ronniesahlberg@gmail.com, Wenchao Xia Le Wednesday 29 Jan 2014 =E0 09:50:21 (+0100), Peter Lieven a =E9crit : > This patch adds native support for accessing images on NFS > shares without the requirement to actually mount the entire > NFS share on the host. >=20 > NFS Images can simply be specified by an url of the form: > nfs:////[?param=3Dvalue[¶m2=3Dvalue2[&...]]= ] >=20 > For example: > qemu-img create -f qcow2 nfs://10.0.0.1/qemu-images/test.qcow2 >=20 > You need LibNFS from Ronnie Sahlberg available at: > git://github.com/sahlberg/libnfs.git > for this to work. >=20 > During configure it is automatically probed for libnfs and support > is enabled on-the-fly. You can forbid or enforce libnfs support > with --disable-libnfs or --enable-libnfs respectively. >=20 > Due to NFS restrictions you might need to execute your binaries > as root, allow them to open priviledged ports (<1024) or specify > insecure option on the NFS server. >=20 > For additional information on ROOT vs. non-ROOT operation and URL > format + parameters see: > https://raw.github.com/sahlberg/libnfs/master/README >=20 > Supported by qemu are the uid, gid and tcp-syncnt URL parameters. >=20 > LibNFS currently support NFS version 3 only. >=20 > Signed-off-by: Peter Lieven > --- > MAINTAINERS | 5 + > block/Makefile.objs | 1 + > block/nfs.c | 440 +++++++++++++++++++++++++++++++++++++++++++= ++++++++ > configure | 26 +++ > qapi-schema.json | 1 + > 5 files changed, 473 insertions(+) > create mode 100644 block/nfs.c >=20 > diff --git a/MAINTAINERS b/MAINTAINERS > index fb53242..f8411f9 100644 > --- a/MAINTAINERS > +++ b/MAINTAINERS > @@ -936,6 +936,11 @@ M: Peter Lieven > S: Supported > F: block/iscsi.c > =20 > +NFS > +M: Peter Lieven > +S: Maintained > +F: block/nfs.c > + > SSH > M: Richard W.M. Jones > S: Supported > diff --git a/block/Makefile.objs b/block/Makefile.objs > index 4e8c91e..e254a21 100644 > --- a/block/Makefile.objs > +++ b/block/Makefile.objs > @@ -12,6 +12,7 @@ block-obj-$(CONFIG_LINUX_AIO) +=3D linux-aio.o > ifeq ($(CONFIG_POSIX),y) > block-obj-y +=3D nbd.o nbd-client.o sheepdog.o > block-obj-$(CONFIG_LIBISCSI) +=3D iscsi.o > +block-obj-$(CONFIG_LIBNFS) +=3D nfs.o > block-obj-$(CONFIG_CURL) +=3D curl.o > block-obj-$(CONFIG_RBD) +=3D rbd.o > block-obj-$(CONFIG_GLUSTERFS) +=3D gluster.o > diff --git a/block/nfs.c b/block/nfs.c > new file mode 100644 > index 0000000..2bbf7a2 > --- /dev/null > +++ b/block/nfs.c > @@ -0,0 +1,440 @@ > +/* > + * QEMU Block driver for native access to files on NFS shares > + * > + * Copyright (c) 2014 Peter Lieven > + * > + * Permission is hereby granted, free of charge, to any person obtaini= ng a copy > + * of this software and associated documentation files (the "Software"= ), to deal > + * in the Software without restriction, including without limitation t= he rights > + * to use, copy, modify, merge, publish, distribute, sublicense, and/o= r sell > + * copies of the Software, and to permit persons to whom the Software = is > + * furnished to do so, subject to the following conditions: > + * > + * The above copyright notice and this permission notice shall be incl= uded in > + * all copies or substantial portions of the Software. > + * > + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXP= RESS OR > + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABI= LITY, > + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT S= HALL > + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES O= R OTHER > + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARI= SING FROM, > + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALI= NGS IN > + * THE SOFTWARE. > + */ > + > +#include "config-host.h" > + > +#include > +#include "qemu-common.h" > +#include "qemu/config-file.h" > +#include "qemu/error-report.h" > +#include "block/block_int.h" > +#include "trace.h" > +#include "qemu/iov.h" > +#include "qemu/uri.h" > +#include "sysemu/sysemu.h" > +#include > + > +typedef struct NFSClient { > + struct nfs_context *context; > + struct nfsfh *fh; > + int events; > + bool has_zero_init; > +} NFSClient; > + > +typedef struct NFSRPC { > + int status; > + int complete; > + QEMUIOVector *iov; > + struct stat *st; > + Coroutine *co; > + QEMUBH *bh; > +} NFSRPC; > + > +static void nfs_process_read(void *arg); > +static void nfs_process_write(void *arg); > + > +static void nfs_set_events(NFSClient *client) > +{ > + int ev =3D nfs_which_events(client->context); > + if (ev !=3D client->events) { > + qemu_aio_set_fd_handler(nfs_get_fd(client->context), > + (ev & POLLIN) ? nfs_process_read : NULL, > + (ev & POLLOUT) ? nfs_process_write : NULL, > + client); > + > + } > + client->events =3D ev; > +} > + > +static void nfs_process_read(void *arg) > +{ > + NFSClient *client =3D arg; > + nfs_service(client->context, POLLIN); > + nfs_set_events(client); > +} > + > +static void nfs_process_write(void *arg) > +{ > + NFSClient *client =3D arg; > + nfs_service(client->context, POLLOUT); > + nfs_set_events(client); > +} > + > +static void nfs_co_init_task(NFSClient *client, NFSRPC *task) > +{ > + *task =3D (NFSRPC) { > + .co =3D qemu_coroutine_self(), > + }; > +} > + > +static void nfs_co_generic_bh_cb(void *opaque) > +{ > + NFSRPC *task =3D opaque; > + qemu_bh_delete(task->bh); > + qemu_coroutine_enter(task->co, NULL); > +} > + > +static void > +nfs_co_generic_cb(int status, struct nfs_context *nfs, void *data, > + void *private_data) > +{ > + NFSRPC *task =3D private_data; > + task->complete =3D 1; > + task->status =3D status; > + if (task->status > 0 && task->iov) { > + if (task->status <=3D task->iov->size) { It feel very odd to compare something named status with something named s= ize. > + qemu_iovec_from_buf(task->iov, 0, data, task->status); > + } else { > + task->status =3D -EIO; > + } > + } > + if (task->status =3D=3D 0 && task->st) { > + memcpy(task->st, data, sizeof(struct stat)); > + } > + if (task->co) { > + task->bh =3D qemu_bh_new(nfs_co_generic_bh_cb, task); > + qemu_bh_schedule(task->bh); > + } > +} > + > +static int coroutine_fn nfs_co_readv(BlockDriverState *bs, > + int64_t sector_num, int nb_sector= s, > + QEMUIOVector *iov) > +{ > + NFSClient *client =3D bs->opaque; > + NFSRPC task; > + > + nfs_co_init_task(client, &task); > + task.iov =3D iov; > + > + if (nfs_pread_async(client->context, client->fh, > + sector_num * BDRV_SECTOR_SIZE, > + nb_sectors * BDRV_SECTOR_SIZE, > + nfs_co_generic_cb, &task) !=3D 0) { > + return -ENOMEM; > + } > + > + while (!task.complete) { > + nfs_set_events(client); > + qemu_coroutine_yield(); > + } > + > + if (task.status < 0) { > + return task.status; > + } > + > + /* zero pad short reads */ > + if (task.status < iov->size) { > + qemu_iovec_memset(iov, task.status, 0, iov->size - task.status= ); > + } > + > + return 0; > +} > + > +static int coroutine_fn nfs_co_writev(BlockDriverState *bs, > + int64_t sector_num, int nb_sec= tors, > + QEMUIOVector *iov) > +{ > + NFSClient *client =3D bs->opaque; > + NFSRPC task; > + char *buf =3D NULL; > + > + nfs_co_init_task(client, &task); > + > + buf =3D g_malloc(nb_sectors * BDRV_SECTOR_SIZE); > + qemu_iovec_to_buf(iov, 0, buf, nb_sectors * BDRV_SECTOR_SIZE); > + > + if (nfs_pwrite_async(client->context, client->fh, > + sector_num * BDRV_SECTOR_SIZE, > + nb_sectors * BDRV_SECTOR_SIZE, > + buf, nfs_co_generic_cb, &task) !=3D 0) { > + g_free(buf); > + return -ENOMEM; > + } > + > + while (!task.complete) { > + nfs_set_events(client); > + qemu_coroutine_yield(); > + } > + > + g_free(buf); > + > + if (task.status !=3D nb_sectors * BDRV_SECTOR_SIZE) { > + return task.status < 0 ? task.status : -EIO; > + } > + > + return 0; > +} > + > +static int coroutine_fn nfs_co_flush(BlockDriverState *bs) > +{ > + NFSClient *client =3D bs->opaque; > + NFSRPC task; > + > + nfs_co_init_task(client, &task); > + > + if (nfs_fsync_async(client->context, client->fh, nfs_co_generic_cb= , > + &task) !=3D 0) { > + return -ENOMEM; > + } > + > + while (!task.complete) { > + nfs_set_events(client); > + qemu_coroutine_yield(); > + } > + > + return task.status; > +} > + > +/* TODO Convert to fine grained options */ > +static QemuOptsList runtime_opts =3D { > + .name =3D "nfs", > + .head =3D QTAILQ_HEAD_INITIALIZER(runtime_opts.head), > + .desc =3D { > + { > + .name =3D "filename", > + .type =3D QEMU_OPT_STRING, > + .help =3D "URL to the NFS file", > + }, > + { /* end of list */ } > + }, > +}; > + > +static void nfs_client_close(NFSClient *client) > +{ > + if (client->context) { > + if (client->fh) { > + nfs_close(client->context, client->fh); > + } > + qemu_aio_set_fd_handler(nfs_get_fd(client->context), NULL, NUL= L, NULL); > + nfs_destroy_context(client->context); > + } > + memset(client, 0, sizeof(NFSClient)); > +} > + > +static void nfs_file_close(BlockDriverState *bs) > +{ > + NFSClient *client =3D bs->opaque; > + nfs_client_close(client); > +} > + > +static int64_t nfs_.cilclient_open(NFSClient *client, const char *file= name, > + int flags, Error **errp) > +{ > + int ret =3D -EINVAL, i; > + struct stat st; > + URI *uri; > + QueryParams *qp =3D NULL; > + char *file =3D NULL, *strp =3D NULL; > + > + uri =3D uri_parse(filename); > + if (!uri) { > + error_setg(errp, "Invalid URL specified"); > + goto fail; Should this be goto out since we don't have done nfs_init_context yet ? > + } > + strp =3D strrchr(uri->path, '/'); > + if (strp =3D=3D NULL) { > + error_setg(errp, "Invalid URL specified"); > + goto fail; goto out ? > + } > + file =3D g_strdup(strp); > + *strp =3D 0; > + > + client->context =3D nfs_init_context(); > + if (client->context =3D=3D NULL) { > + error_setg(errp, "Failed to init NFS context"); > + goto fail; > + } > + > + qp =3D query_params_parse(uri->query); > + for (i =3D 0; i < qp->n; i++) { > + if (!qp->p[i].value) { > + error_setg(errp, "Value for NFS parameter expected: %s", > + qp->p[i].name); > + goto fail; > + } > + if (!strncmp(qp->p[i].name, "uid", 3)) { > + nfs_set_uid(client->context, atoi(qp->p[i].value)); > + } else if (!strncmp(qp->p[i].name, "gid", 3)) { > + nfs_set_gid(client->context, atoi(qp->p[i].value)); > + } else if (!strncmp(qp->p[i].name, "tcp-syncnt", 10)) { > + nfs_set_tcp_syncnt(client->context, atoi(qp->p[i].value)); > + } else { > + error_setg(errp, "Unknown NFS parameter name: %s", > + qp->p[i].name); > + goto fail; > + } > + } > + > + ret =3D nfs_mount(client->context, uri->server, uri->path); > + if (ret < 0) { > + error_setg(errp, "Failed to mount nfs share: %s", > + nfs_get_error(client->context)); > + goto fail; > + } > + > + if (flags & O_CREAT) { > + ret =3D nfs_creat(client->context, file, 0600, &client->fh); > + if (ret < 0) { > + error_setg(errp, "Failed to create file: %s", > + nfs_get_error(client->context)); > + goto fail; > + } > + } else { > + ret =3D nfs_open(client->context, file, flags, &client->fh); > + if (ret < 0) { > + error_setg(errp, "Failed to open file : %s", > + nfs_get_error(client->context)); > + goto fail; > + } > + } > + > + ret =3D nfs_fstat(client->context, client->fh, &st); > + if (ret < 0) { > + error_setg(errp, "Failed to fstat file: %s", > + nfs_get_error(client->context)); > + goto fail; > + } > + > + ret =3D DIV_ROUND_UP(st.st_size, BDRV_SECTOR_SIZE); > + client->has_zero_init =3D S_ISREG(st.st_mode); > + goto out; > +fail: > + nfs_client_close(client); > +out: > + if (qp) { > + query_params_free(qp); > + } > + uri_free(uri); > + g_free(file); > + return ret; > +} > + > +static int nfs_file_open(BlockDriverState *bs, QDict *options, int fla= gs, > + Error **errp) { > + NFSClient *client =3D bs->opaque; > + int64_t ret; > + QemuOpts *opts; > + Error *local_err =3D NULL; > + > + opts =3D qemu_opts_create(&runtime_opts, NULL, 0, &error_abort); > + qemu_opts_absorb_qdict(opts, options, &local_err); > + if (error_is_set(&local_err)) { > + qerror_report_err(local_err); I have seen more usage of error_propagate(errp, local_err); in QEMU code. Maybe I am missing the point. > + error_free(local_err); > + return -EINVAL; > + } > + ret =3D nfs_client_open(client, qemu_opt_get(opts, "filename"), > + (flags & BDRV_O_RDWR) ? O_RDWR : O_RDONLY, > + errp); > + if (ret < 0) { > + return ret; > + } > + bs->total_sectors =3D ret; > + return 0; > +} > + > +static int nfs_file_create(const char *url, QEMUOptionParameter *optio= ns, > + Error **errp) > +{ > + int ret =3D 0; > + int64_t total_size =3D 0; > + NFSClient *client =3D g_malloc0(sizeof(NFSClient)); > + > + /* Read out options */ > + while (options && options->name) { > + if (!strcmp(options->name, "size")) { > + total_size =3D options->value.n; > + } > + options++; > + } > + > + ret =3D nfs_client_open(client, url, O_CREAT, errp); > + if (ret < 0) { > + goto out; > + } > + ret =3D nfs_ftruncate(client->context, client->fh, total_size); > +out: Should the out: label be after nfs_client_close(client); since the code w= ill jump to it on nfs_client_open failure ? > + nfs_client_close(client); > + g_free(client); > + return ret; > +} > + > +static int nfs_has_zero_init(BlockDriverState *bs) > +{ > + NFSClient *client =3D bs->opaque; > + return client->has_zero_init; > +} > + > +static int64_t nfs_get_allocated_file_size(BlockDriverState *bs) > +{ > + NFSClient *client =3D bs->opaque; > + NFSRPC task =3D {0}; > + struct stat st; > + > + task.st =3D &st; > + if (nfs_fstat_async(client->context, client->fh, nfs_co_generic_cb= , > + &task) !=3D 0) { > + return -ENOMEM; > + } > + > + while (!task.complete) { > + nfs_set_events(client); > + qemu_aio_wait(); > + } > + > + return (task.status < 0 ? task.status : st.st_blocks * st.st_blksi= ze); > +} > + > +static int nfs_file_truncate(BlockDriverState *bs, int64_t offset) > +{ > + NFSClient *client =3D bs->opaque; > + return nfs_ftruncate(client->context, client->fh, offset); > +} > + > +static BlockDriver bdrv_nfs =3D { > + .format_name =3D "nfs", > + .protocol_name =3D "nfs", > + > + .instance_size =3D sizeof(NFSClient), > + .bdrv_needs_filename =3D true, > + .bdrv_has_zero_init =3D nfs_has_zero_init, > + .bdrv_get_allocated_file_size =3D nfs_get_allocated_file_size, > + .bdrv_truncate =3D nfs_file_truncate, > + > + .bdrv_file_open =3D nfs_file_open, > + .bdrv_close =3D nfs_file_close, > + .bdrv_create =3D nfs_file_create, > + > + .bdrv_co_readv =3D nfs_co_readv, > + .bdrv_co_writev =3D nfs_co_writev, > + .bdrv_co_flush_to_disk =3D nfs_co_flush, > +}; > + > +static void nfs_block_init(void) > +{ > + bdrv_register(&bdrv_nfs); > +} > + > +block_init(nfs_block_init); > diff --git a/configure b/configure > index b472694..9ca0c63 100755 > --- a/configure > +++ b/configure > @@ -251,6 +251,7 @@ vss_win32_sdk=3D"" > win_sdk=3D"no" > want_tools=3D"yes" > libiscsi=3D"" > +libnfs=3D"" > coroutine=3D"" > coroutine_pool=3D"" > seccomp=3D"" > @@ -840,6 +841,10 @@ for opt do > ;; > --enable-libiscsi) libiscsi=3D"yes" > ;; > + --disable-libnfs) libnfs=3D"no" > + ;; > + --enable-libnfs) libnfs=3D"yes" > + ;; > --enable-profiler) profiler=3D"yes" > ;; > --disable-cocoa) cocoa=3D"no" > @@ -1229,6 +1234,8 @@ Advanced options (experts only): > --enable-rbd enable building the rados block device (rbd= ) > --disable-libiscsi disable iscsi support > --enable-libiscsi enable iscsi support > + --disable-libnfs disable nfs support > + --enable-libnfs enable nfs support > --disable-smartcard-nss disable smartcard nss support > --enable-smartcard-nss enable smartcard nss support > --disable-libusb disable libusb (for usb passthrough) > @@ -3600,6 +3607,20 @@ elif test "$debug" =3D "no" ; then > CFLAGS=3D"-O2 -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=3D2 $CFLAGS" > fi > =20 > +########################################## > +# Do we have libnfs > +if test "$libnfs" !=3D "no" ; then > + if $pkg_config --atleast-version=3D1.8.91 libnfs; then > + libnfs=3D"yes" > + libnfs_libs=3D$($pkg_config --libs libnfs) > + LIBS=3D"$LIBS $libnfs_libs" > + else > + if test "$libnfs" =3D "yes" ; then > + feature_not_found "libnfs" > + fi > + libnfs=3D"no" > + fi > +fi > =20 > # Disable zero malloc errors for official releases unless explicitly t= old to > # enable/disable > @@ -3829,6 +3850,7 @@ echo "libiscsi support $libiscsi (1.4.0)" > else > echo "libiscsi support $libiscsi" > fi > +echo "libnfs support $libnfs" > echo "build guest agent $guest_agent" > echo "QGA VSS support $guest_agent_with_vss" > echo "seccomp support $seccomp" > @@ -4165,6 +4187,10 @@ if test "$libiscsi" =3D "yes" ; then > fi > fi > =20 > +if test "$libnfs" =3D "yes" ; then > + echo "CONFIG_LIBNFS=3Dy" >> $config_host_mak > +fi > + > if test "$seccomp" =3D "yes"; then > echo "CONFIG_SECCOMP=3Dy" >> $config_host_mak > fi > diff --git a/qapi-schema.json b/qapi-schema.json > index 05ced9d..7cfb5e5 100644 > --- a/qapi-schema.json > +++ b/qapi-schema.json > @@ -4371,6 +4371,7 @@ > # TODO gluster: Wait for structured options > # TODO iscsi: Wait for structured options > # TODO nbd: Should take InetSocketAddress for 'host'? > +# TODO nfs: Wait for structured options > # TODO rbd: Wait for structured options > # TODO sheepdog: Wait for structured options > # TODO ssh: Should take InetSocketAddress for 'host'? > --=20 > 1.7.9.5 >=20 >=20