| Message ID | 20200430171814.GA275398@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [v2] fuse, virtiofs: Do not alloc/install fuse device in fuse_fill_super_common() | expand |
On Thu, Apr 30, 2020 at 01:18:14PM -0400, Vivek Goyal wrote: > As of now fuse_fill_super_common() allocates and installs one fuse device. > Filesystems like virtiofs can have more than one filesystem queues and > can have one fuse device per queue. Give, fuse_fill_super_common() only > handles one device, virtiofs allocates and installes fuse devices for > all queues except one. > > This makes logic little twisted and hard to understand. It probably > is better to not do any device allocation/installation in > fuse_fill_super_common() and let caller take care of it instead. > > v2: Removed fuse_dev_alloc_install() call from fuse_fill_super_common(). > > Signed-off-by: Vivek Goyal <vgoyal@redhat.com> > --- > fs/fuse/fuse_i.h | 3 --- > fs/fuse/inode.c | 30 ++++++++++++++---------------- > fs/fuse/virtio_fs.c | 9 +-------- > 3 files changed, 15 insertions(+), 27 deletions(-) Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
On Thu, Apr 30, 2020 at 7:18 PM Vivek Goyal <vgoyal@redhat.com> wrote: > > As of now fuse_fill_super_common() allocates and installs one fuse device. > Filesystems like virtiofs can have more than one filesystem queues and > can have one fuse device per queue. Give, fuse_fill_super_common() only > handles one device, virtiofs allocates and installes fuse devices for > all queues except one. > > This makes logic little twisted and hard to understand. It probably > is better to not do any device allocation/installation in > fuse_fill_super_common() and let caller take care of it instead. Taking a closer look... > > v2: Removed fuse_dev_alloc_install() call from fuse_fill_super_common(). > > Signed-off-by: Vivek Goyal <vgoyal@redhat.com> > --- > fs/fuse/fuse_i.h | 3 --- > fs/fuse/inode.c | 30 ++++++++++++++---------------- > fs/fuse/virtio_fs.c | 9 +-------- > 3 files changed, 15 insertions(+), 27 deletions(-) > > diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h > index ca344bf71404..df0a62f963a8 100644 > --- a/fs/fuse/fuse_i.h > +++ b/fs/fuse/fuse_i.h > @@ -485,9 +485,6 @@ struct fuse_fs_context { > unsigned int max_read; > unsigned int blksize; > const char *subtype; > - > - /* fuse_dev pointer to fill in, should contain NULL on entry */ > - void **fudptr; > }; > > /** > diff --git a/fs/fuse/inode.c b/fs/fuse/inode.c > index 95d712d44ca1..6b38e0391c96 100644 > --- a/fs/fuse/inode.c > +++ b/fs/fuse/inode.c > @@ -1113,7 +1113,6 @@ EXPORT_SYMBOL_GPL(fuse_dev_free); > > int fuse_fill_super_common(struct super_block *sb, struct fuse_fs_context *ctx) > { > - struct fuse_dev *fud; > struct fuse_conn *fc = get_fuse_conn_super(sb); > struct inode *root; > struct dentry *root_dentry; > @@ -1155,15 +1154,11 @@ int fuse_fill_super_common(struct super_block *sb, struct fuse_fs_context *ctx) > if (sb->s_user_ns != &init_user_ns) > sb->s_xattr = fuse_no_acl_xattr_handlers; > > - fud = fuse_dev_alloc_install(fc); > - if (!fud) > - goto err; > - > fc->dev = sb->s_dev; > fc->sb = sb; > err = fuse_bdi_init(fc, sb); > if (err) > - goto err_dev_free; > + goto err; > > /* Handle umasking inside the fuse code */ > if (sb->s_flags & SB_POSIXACL) > @@ -1185,30 +1180,24 @@ int fuse_fill_super_common(struct super_block *sb, struct fuse_fs_context *ctx) > sb->s_d_op = &fuse_root_dentry_operations; > root_dentry = d_make_root(root); > if (!root_dentry) > - goto err_dev_free; > + goto err; > /* Root dentry doesn't have .d_revalidate */ > sb->s_d_op = &fuse_dentry_operations; > > mutex_lock(&fuse_mutex); > err = -EINVAL; > - if (*ctx->fudptr) > - goto err_unlock; > - > err = fuse_ctl_add_conn(fc); > if (err) > goto err_unlock; > > list_add_tail(&fc->entry, &fuse_conn_list); > sb->s_root = root_dentry; > - *ctx->fudptr = fud; > mutex_unlock(&fuse_mutex); > return 0; > > err_unlock: > mutex_unlock(&fuse_mutex); > dput(root_dentry); > - err_dev_free: > - fuse_dev_free(fud); > err: > return err; > } > @@ -1220,6 +1209,7 @@ static int fuse_fill_super(struct super_block *sb, struct fs_context *fsc) > struct file *file; > int err; > struct fuse_conn *fc; > + struct fuse_dev *fud; > > err = -EINVAL; > file = fget(ctx->fd); > @@ -1233,13 +1223,16 @@ static int fuse_fill_super(struct super_block *sb, struct fs_context *fsc) > if ((file->f_op != &fuse_dev_operations) || > (file->f_cred->user_ns != sb->s_user_ns)) > goto err_fput; > - ctx->fudptr = &file->private_data; > > - fc = kmalloc(sizeof(*fc), GFP_KERNEL); > err = -ENOMEM; > - if (!fc) > + fud = fuse_dev_alloc(); > + if (!fud) > goto err_fput; > > + fc = kmalloc(sizeof(*fc), GFP_KERNEL); > + if (!fc) > + goto err_free_dev; > + > fuse_conn_init(fc, sb->s_user_ns, &fuse_dev_fiq_ops, NULL); > fc->release = fuse_free_conn; > sb->s_fs_info = fc; > @@ -1247,6 +1240,9 @@ static int fuse_fill_super(struct super_block *sb, struct fs_context *fsc) > err = fuse_fill_super_common(sb, ctx); > if (err) > goto err_put_conn; > + > + fuse_dev_install(fud, fc); > + file->private_data = fud; We've lost the check for non-null file->private_data; i.e. a fuse fd already bound to a super block. That needs to be restored, together with protection against two such instances racing with each other. Maybe we are better off moving the whole fuse_mutex protected block from the end of fuse_fill_super_common() into the callers. Thanks, Miklos
On Mon, May 04, 2020 at 03:30:39PM +0200, Miklos Szeredi wrote: > On Thu, Apr 30, 2020 at 7:18 PM Vivek Goyal <vgoyal@redhat.com> wrote: > > > > As of now fuse_fill_super_common() allocates and installs one fuse device. > > Filesystems like virtiofs can have more than one filesystem queues and > > can have one fuse device per queue. Give, fuse_fill_super_common() only > > handles one device, virtiofs allocates and installes fuse devices for > > all queues except one. > > > > This makes logic little twisted and hard to understand. It probably > > is better to not do any device allocation/installation in > > fuse_fill_super_common() and let caller take care of it instead. > > Taking a closer look... > > > > > v2: Removed fuse_dev_alloc_install() call from fuse_fill_super_common(). > > > > Signed-off-by: Vivek Goyal <vgoyal@redhat.com> > > --- > > fs/fuse/fuse_i.h | 3 --- > > fs/fuse/inode.c | 30 ++++++++++++++---------------- > > fs/fuse/virtio_fs.c | 9 +-------- > > 3 files changed, 15 insertions(+), 27 deletions(-) > > > > diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h > > index ca344bf71404..df0a62f963a8 100644 > > --- a/fs/fuse/fuse_i.h > > +++ b/fs/fuse/fuse_i.h > > @@ -485,9 +485,6 @@ struct fuse_fs_context { > > unsigned int max_read; > > unsigned int blksize; > > const char *subtype; > > - > > - /* fuse_dev pointer to fill in, should contain NULL on entry */ > > - void **fudptr; > > }; > > > > /** > > diff --git a/fs/fuse/inode.c b/fs/fuse/inode.c > > index 95d712d44ca1..6b38e0391c96 100644 > > --- a/fs/fuse/inode.c > > +++ b/fs/fuse/inode.c > > @@ -1113,7 +1113,6 @@ EXPORT_SYMBOL_GPL(fuse_dev_free); > > > > int fuse_fill_super_common(struct super_block *sb, struct fuse_fs_context *ctx) > > { > > - struct fuse_dev *fud; > > struct fuse_conn *fc = get_fuse_conn_super(sb); > > struct inode *root; > > struct dentry *root_dentry; > > @@ -1155,15 +1154,11 @@ int fuse_fill_super_common(struct super_block *sb, struct fuse_fs_context *ctx) > > if (sb->s_user_ns != &init_user_ns) > > sb->s_xattr = fuse_no_acl_xattr_handlers; > > > > - fud = fuse_dev_alloc_install(fc); > > - if (!fud) > > - goto err; > > - > > fc->dev = sb->s_dev; > > fc->sb = sb; > > err = fuse_bdi_init(fc, sb); > > if (err) > > - goto err_dev_free; > > + goto err; > > > > /* Handle umasking inside the fuse code */ > > if (sb->s_flags & SB_POSIXACL) > > @@ -1185,30 +1180,24 @@ int fuse_fill_super_common(struct super_block *sb, struct fuse_fs_context *ctx) > > sb->s_d_op = &fuse_root_dentry_operations; > > root_dentry = d_make_root(root); > > if (!root_dentry) > > - goto err_dev_free; > > + goto err; > > /* Root dentry doesn't have .d_revalidate */ > > sb->s_d_op = &fuse_dentry_operations; > > > > mutex_lock(&fuse_mutex); > > err = -EINVAL; > > - if (*ctx->fudptr) > > - goto err_unlock; > > - > > err = fuse_ctl_add_conn(fc); > > if (err) > > goto err_unlock; > > > > list_add_tail(&fc->entry, &fuse_conn_list); > > sb->s_root = root_dentry; > > - *ctx->fudptr = fud; > > mutex_unlock(&fuse_mutex); > > return 0; > > > > err_unlock: > > mutex_unlock(&fuse_mutex); > > dput(root_dentry); > > - err_dev_free: > > - fuse_dev_free(fud); > > err: > > return err; > > } > > @@ -1220,6 +1209,7 @@ static int fuse_fill_super(struct super_block *sb, struct fs_context *fsc) > > struct file *file; > > int err; > > struct fuse_conn *fc; > > + struct fuse_dev *fud; > > > > err = -EINVAL; > > file = fget(ctx->fd); > > @@ -1233,13 +1223,16 @@ static int fuse_fill_super(struct super_block *sb, struct fs_context *fsc) > > if ((file->f_op != &fuse_dev_operations) || > > (file->f_cred->user_ns != sb->s_user_ns)) > > goto err_fput; > > - ctx->fudptr = &file->private_data; > > > > - fc = kmalloc(sizeof(*fc), GFP_KERNEL); > > err = -ENOMEM; > > - if (!fc) > > + fud = fuse_dev_alloc(); > > + if (!fud) > > goto err_fput; > > > > + fc = kmalloc(sizeof(*fc), GFP_KERNEL); > > + if (!fc) > > + goto err_free_dev; > > + > > fuse_conn_init(fc, sb->s_user_ns, &fuse_dev_fiq_ops, NULL); > > fc->release = fuse_free_conn; > > sb->s_fs_info = fc; > > @@ -1247,6 +1240,9 @@ static int fuse_fill_super(struct super_block *sb, struct fs_context *fsc) > > err = fuse_fill_super_common(sb, ctx); > > if (err) > > goto err_put_conn; > > + > > + fuse_dev_install(fud, fc); > > + file->private_data = fud; > > We've lost the check for non-null file->private_data; i.e. a fuse fd > already bound to a super block. That needs to be restored, together > with protection against two such instances racing with each other. > > Maybe we are better off moving the whole fuse_mutex protected block > from the end of fuse_fill_super_common() into the callers. I Miklos, I tried moving code under fuse_mutex() outside fuse_fill_super_common() and that meant another 6 symbols had to be used by virtiofs and I had to export these. In the end, final patch looked much bigger. To keep changes small, how about if we retain device installation logic in fuse_fill_super_common() but make it optional and caller has to opt in. Regular fuse will continue to use device installation facility and virtiofs will opt out. IIUC, virtiofs does not have this multiple mount racing issue because sget_fc() takes care of races and only one instance of virtio_fs_fill_super() can be in progress. So I wrote following patch and this looks much simpler. What do you think about it. Thanks Vivek Subject: virtiofs: Do not use fuse_fill_super_common() for fuse device installation fuse_fill_super_common() allocates and installs one fuse_device. Hence virtiofs allocates and install all fuse devices by itself except one. This makes logic little twisted. There does not seem to be any real need that why virtiofs can't allocate and install all fuse devices itself. So opt out of fuse device allocation and installation while calling fuse_fill_super_common(). Regular fuse still wants fuse_fill_super_common() to install fuse_device. It needs to prevent against races where two mounters are trying to mount fuse using same fd. In that case one will succeed while other will get -EINVAL. virtiofs does not have this issue because sget_fc() resolves the race w.r.t multiple mounters and only one instance of virtio_fs_fill_super() should be in progress for same filesystem. Signed-off-by: Vivek Goyal <vgoyal@redhat.com> --- fs/fuse/inode.c | 19 ++++++++++++------- fs/fuse/virtio_fs.c | 9 +++------ 2 files changed, 15 insertions(+), 13 deletions(-) Index: redhat-linux/fs/fuse/inode.c =================================================================== --- redhat-linux.orig/fs/fuse/inode.c 2020-05-04 13:41:50.626368034 -0400 +++ redhat-linux/fs/fuse/inode.c 2020-05-04 14:14:42.343368034 -0400 @@ -1113,7 +1113,7 @@ EXPORT_SYMBOL_GPL(fuse_dev_free); int fuse_fill_super_common(struct super_block *sb, struct fuse_fs_context *ctx) { - struct fuse_dev *fud; + struct fuse_dev *fud = NULL; struct fuse_conn *fc = get_fuse_conn_super(sb); struct inode *root; struct dentry *root_dentry; @@ -1155,9 +1155,12 @@ int fuse_fill_super_common(struct super_ if (sb->s_user_ns != &init_user_ns) sb->s_xattr = fuse_no_acl_xattr_handlers; - fud = fuse_dev_alloc_install(fc); - if (!fud) - goto err; + if (ctx->fudptr) { + err = -ENOMEM; + fud = fuse_dev_alloc_install(fc); + if (!fud) + goto err; + } fc->dev = sb->s_dev; fc->sb = sb; @@ -1191,7 +1194,7 @@ int fuse_fill_super_common(struct super_ mutex_lock(&fuse_mutex); err = -EINVAL; - if (*ctx->fudptr) + if (ctx->fudptr && *ctx->fudptr) goto err_unlock; err = fuse_ctl_add_conn(fc); @@ -1200,7 +1203,8 @@ int fuse_fill_super_common(struct super_ list_add_tail(&fc->entry, &fuse_conn_list); sb->s_root = root_dentry; - *ctx->fudptr = fud; + if (ctx->fudptr) + *ctx->fudptr = fud; mutex_unlock(&fuse_mutex); return 0; @@ -1208,7 +1212,8 @@ int fuse_fill_super_common(struct super_ mutex_unlock(&fuse_mutex); dput(root_dentry); err_dev_free: - fuse_dev_free(fud); + if (fud) + fuse_dev_free(fud); err: return err; } Index: redhat-linux/fs/fuse/virtio_fs.c =================================================================== --- redhat-linux.orig/fs/fuse/virtio_fs.c 2020-05-04 13:41:50.626368034 -0400 +++ redhat-linux/fs/fuse/virtio_fs.c 2020-05-04 13:49:48.567368034 -0400 @@ -1067,7 +1067,7 @@ static int virtio_fs_fill_super(struct s err = -ENOMEM; /* Allocate fuse_dev for hiprio and notification queues */ - for (i = 0; i < VQ_REQUEST; i++) { + for (i = 0; i < fs->nvqs; i++) { struct virtio_fs_vq *fsvq = &fs->vqs[i]; fsvq->fud = fuse_dev_alloc(); @@ -1075,18 +1075,15 @@ static int virtio_fs_fill_super(struct s goto err_free_fuse_devs; } - ctx.fudptr = (void **)&fs->vqs[VQ_REQUEST].fud; + /* virtiofs allocates and installs its own fuse devices */ + ctx.fudptr = NULL; err = fuse_fill_super_common(sb, &ctx); if (err < 0) goto err_free_fuse_devs; - fc = fs->vqs[VQ_REQUEST].fud->fc; - for (i = 0; i < fs->nvqs; i++) { struct virtio_fs_vq *fsvq = &fs->vqs[i]; - if (i == VQ_REQUEST) - continue; /* already initialized */ fuse_dev_install(fsvq->fud, fc); }
On Mon, May 4, 2020 at 8:33 PM Vivek Goyal <vgoyal@redhat.com> wrote: > > Subject: virtiofs: Do not use fuse_fill_super_common() for fuse device installation > > fuse_fill_super_common() allocates and installs one fuse_device. Hence > virtiofs allocates and install all fuse devices by itself except one. > > This makes logic little twisted. There does not seem to be any real need > that why virtiofs can't allocate and install all fuse devices itself. > > So opt out of fuse device allocation and installation while calling > fuse_fill_super_common(). > > Regular fuse still wants fuse_fill_super_common() to install fuse_device. > It needs to prevent against races where two mounters are trying to mount > fuse using same fd. In that case one will succeed while other will get > -EINVAL. > > virtiofs does not have this issue because sget_fc() resolves the race > w.r.t multiple mounters and only one instance of virtio_fs_fill_super() > should be in progress for same filesystem. Thanks, applied. Miklos
diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h index ca344bf71404..df0a62f963a8 100644 --- a/fs/fuse/fuse_i.h +++ b/fs/fuse/fuse_i.h @@ -485,9 +485,6 @@ struct fuse_fs_context { unsigned int max_read; unsigned int blksize; const char *subtype; - - /* fuse_dev pointer to fill in, should contain NULL on entry */ - void **fudptr; }; /** diff --git a/fs/fuse/inode.c b/fs/fuse/inode.c index 95d712d44ca1..6b38e0391c96 100644 --- a/fs/fuse/inode.c +++ b/fs/fuse/inode.c @@ -1113,7 +1113,6 @@ EXPORT_SYMBOL_GPL(fuse_dev_free); int fuse_fill_super_common(struct super_block *sb, struct fuse_fs_context *ctx) { - struct fuse_dev *fud; struct fuse_conn *fc = get_fuse_conn_super(sb); struct inode *root; struct dentry *root_dentry; @@ -1155,15 +1154,11 @@ int fuse_fill_super_common(struct super_block *sb, struct fuse_fs_context *ctx) if (sb->s_user_ns != &init_user_ns) sb->s_xattr = fuse_no_acl_xattr_handlers; - fud = fuse_dev_alloc_install(fc); - if (!fud) - goto err; - fc->dev = sb->s_dev; fc->sb = sb; err = fuse_bdi_init(fc, sb); if (err) - goto err_dev_free; + goto err; /* Handle umasking inside the fuse code */ if (sb->s_flags & SB_POSIXACL) @@ -1185,30 +1180,24 @@ int fuse_fill_super_common(struct super_block *sb, struct fuse_fs_context *ctx) sb->s_d_op = &fuse_root_dentry_operations; root_dentry = d_make_root(root); if (!root_dentry) - goto err_dev_free; + goto err; /* Root dentry doesn't have .d_revalidate */ sb->s_d_op = &fuse_dentry_operations; mutex_lock(&fuse_mutex); err = -EINVAL; - if (*ctx->fudptr) - goto err_unlock; - err = fuse_ctl_add_conn(fc); if (err) goto err_unlock; list_add_tail(&fc->entry, &fuse_conn_list); sb->s_root = root_dentry; - *ctx->fudptr = fud; mutex_unlock(&fuse_mutex); return 0; err_unlock: mutex_unlock(&fuse_mutex); dput(root_dentry); - err_dev_free: - fuse_dev_free(fud); err: return err; } @@ -1220,6 +1209,7 @@ static int fuse_fill_super(struct super_block *sb, struct fs_context *fsc) struct file *file; int err; struct fuse_conn *fc; + struct fuse_dev *fud; err = -EINVAL; file = fget(ctx->fd); @@ -1233,13 +1223,16 @@ static int fuse_fill_super(struct super_block *sb, struct fs_context *fsc) if ((file->f_op != &fuse_dev_operations) || (file->f_cred->user_ns != sb->s_user_ns)) goto err_fput; - ctx->fudptr = &file->private_data; - fc = kmalloc(sizeof(*fc), GFP_KERNEL); err = -ENOMEM; - if (!fc) + fud = fuse_dev_alloc(); + if (!fud) goto err_fput; + fc = kmalloc(sizeof(*fc), GFP_KERNEL); + if (!fc) + goto err_free_dev; + fuse_conn_init(fc, sb->s_user_ns, &fuse_dev_fiq_ops, NULL); fc->release = fuse_free_conn; sb->s_fs_info = fc; @@ -1247,6 +1240,9 @@ static int fuse_fill_super(struct super_block *sb, struct fs_context *fsc) err = fuse_fill_super_common(sb, ctx); if (err) goto err_put_conn; + + fuse_dev_install(fud, fc); + file->private_data = fud; /* * atomic_dec_and_test() in fput() provides the necessary * memory barrier for file->private_data to be visible on all @@ -1259,6 +1255,8 @@ static int fuse_fill_super(struct super_block *sb, struct fs_context *fsc) err_put_conn: fuse_conn_put(fc); sb->s_fs_info = NULL; + err_free_dev: + fuse_dev_free(fud); err_fput: fput(file); err: diff --git a/fs/fuse/virtio_fs.c b/fs/fuse/virtio_fs.c index bade74768903..87a7bc0f193c 100644 --- a/fs/fuse/virtio_fs.c +++ b/fs/fuse/virtio_fs.c @@ -1066,8 +1066,7 @@ static int virtio_fs_fill_super(struct super_block *sb) } err = -ENOMEM; - /* Allocate fuse_dev for hiprio and notification queues */ - for (i = 0; i < VQ_REQUEST; i++) { + for (i = 0; i < fs->nvqs; i++) { struct virtio_fs_vq *fsvq = &fs->vqs[i]; fsvq->fud = fuse_dev_alloc(); @@ -1075,18 +1074,12 @@ static int virtio_fs_fill_super(struct super_block *sb) goto err_free_fuse_devs; } - ctx.fudptr = (void **)&fs->vqs[VQ_REQUEST].fud; err = fuse_fill_super_common(sb, &ctx); if (err < 0) goto err_free_fuse_devs; - fc = fs->vqs[VQ_REQUEST].fud->fc; - for (i = 0; i < fs->nvqs; i++) { struct virtio_fs_vq *fsvq = &fs->vqs[i]; - - if (i == VQ_REQUEST) - continue; /* already initialized */ fuse_dev_install(fsvq->fud, fc); }
As of now fuse_fill_super_common() allocates and installs one fuse device. Filesystems like virtiofs can have more than one filesystem queues and can have one fuse device per queue. Give, fuse_fill_super_common() only handles one device, virtiofs allocates and installes fuse devices for all queues except one. This makes logic little twisted and hard to understand. It probably is better to not do any device allocation/installation in fuse_fill_super_common() and let caller take care of it instead. v2: Removed fuse_dev_alloc_install() call from fuse_fill_super_common(). Signed-off-by: Vivek Goyal <vgoyal@redhat.com> --- fs/fuse/fuse_i.h | 3 --- fs/fuse/inode.c | 30 ++++++++++++++---------------- fs/fuse/virtio_fs.c | 9 +-------- 3 files changed, 15 insertions(+), 27 deletions(-)