diff options
| author | Miklos Szeredi <mszeredi@redhat.com> | 2026-11-10 11:26:55 +0100 |
|---|---|---|
| committer | Miklos Szeredi <mszeredi@redhat.com> | 2026-11-19 18:25:59 +0100 |
| commit | 6704a78171aa20caab2cd16da9ae2ff8edd4bb1d (patch) | |
| tree | f956689c043c4e686ff39415ffcf8cb3838ec0cf | |
| parent | f7e0ed317736fb33d18e0b903743c2d7f9faa92c (diff) | |
fuse: advertise support for DAX mappings
Reuse the FUSE_HAS_INODE_DAX flag, which previously was only used for the
virtiofs case.
The two uses are unrelated: for virtiofs the flag means that host memory
can be mapped directly into the guest, while here it means that the server
may back an inode with a dax device opened via FUSE_DEV_IOC_BACKING_OPEN.
Gate the advertisement on IS_REACHABLE(CONFIG_DEV_DAX_FSDEV), which is
what fuse_backing_open_file() requires before it accepts a dax device as
backing; CONFIG_FS_DAX alone is not sufficient.
Plain fuse servers should check both FUSE_HAS_INODE_DAX and
FUSE_PASSTHROUGH_V2 before enabling passthrough dax support.
Stop advertising FUSE_PASSTHROUGH(_V2) in the virtiofs case, which doesn't
support passthrough mode.
Note: FUSE_HAS_INODE_DAX flag in the INIT reply is ignored, so it's handled
slightly differently from the virtiofs case. This is deliberate. The
server needs to check both FUSE_PASSTHROUGH_V2 and FUSE_HAS_INODE_DAX
before enabling DAX based extent map if it wants to bail or fall back to
plain I/O mode on kernels not supporting this feature. A broken server
that does not check the features yet sets FUSE_ATTR_DAX will result in an
error on open. This is not backward compatible, since this flag was
silently ignored previously, but unlikely to be an issue in real life.
Fixes: 2fbe4586cff3 ("fuse: add support for opening dax device as backing")
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
| -rw-r--r-- | fs/fuse/fuse_i.h | 13 | ||||
| -rw-r--r-- | fs/fuse/inode.c | 32 | ||||
| -rw-r--r-- | fs/fuse/virtio_fs.c | 1 | ||||
| -rw-r--r-- | include/uapi/linux/fuse.h | 10 |
4 files changed, 47 insertions, 9 deletions
diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h index a033b0bc4576..e0992fcbcc55 100644 --- a/fs/fuse/fuse_i.h +++ b/fs/fuse/fuse_i.h @@ -421,6 +421,7 @@ struct fuse_fs_context { bool no_req_timeout:1; bool legacy_opts_show:1; bool syncfs_capable:1; + bool is_virtiofs:1; enum fuse_vdax_mode vdax_mode; unsigned int max_read; unsigned int blksize; @@ -761,6 +762,9 @@ struct fuse_conn { /** @sync_init: Is synchronous FUSE_INIT allowed? */ unsigned int sync_init:1; + /** @is_virtiofs: Is this a virtio-fs connection? */ + unsigned int is_virtiofs:1; + /** @max_stack_depth: Maximum stack depth for passthrough backing files */ int max_stack_depth; @@ -1273,6 +1277,15 @@ void fuse_free_conn(struct fuse_conn *fc); /* dax.c */ +static inline bool fuse_conn_vdax(struct fuse_conn *fc) +{ +#ifdef CONFIG_FUSE_VDAX + return fc->vdax; +#else + return false; +#endif +} + static inline bool fuse_inode_vdax(struct inode *inode) { #ifdef CONFIG_FUSE_VDAX diff --git a/fs/fuse/inode.c b/fs/fuse/inode.c index c6bfc66317ad..af6d9e0af190 100644 --- a/fs/fuse/inode.c +++ b/fs/fuse/inode.c @@ -385,7 +385,7 @@ static void fuse_change_attributes_i(struct inode *inode, struct fuse_attr *attr invalidate_inode_pages2(inode->i_mapping); } - if (IS_ENABLED(CONFIG_FUSE_VDAX)) + if (fuse_conn_vdax(fc)) fuse_vdax_dontcache(inode, attr->flags); } @@ -419,13 +419,18 @@ static void fuse_init_inode(struct inode *inode, struct fuse_attr *attr, fuse_init_file_inode(inode, attr->flags); /* + * FUSE_ATTR_DAX means a passthrough extent map on a dax + * device, which only exists on a plain fuse connection. On + * virtiofs the same bit means per inode virtiofs DAX, which + * is handled by fuse_vdax_inode_init() instead. + * * Without CONFIG_FS_DAX, S_DAX is zero and IS_DAX() is * constant false, so all the checks that keep DAX inodes away * from the page cache are compiled out. Don't install address * space operations that have no ->read_folio in that case. */ - if (IS_ENABLED(CONFIG_FS_DAX) && (attr->flags & FUSE_ATTR_DAX) && - !fuse_inode_vdax(inode)) { + if (IS_ENABLED(CONFIG_FS_DAX) && !fc->is_virtiofs && + (attr->flags & FUSE_ATTR_DAX)) { inode->i_flags |= S_DAX; inode->i_data.a_ops = &fuse_dax_aops; } @@ -1394,7 +1399,7 @@ static void process_init_reply(struct fuse_args *args, int error) !fuse_vdax_check_alignment(fc, arg->map_alignment)) { ok = false; } - if (flags & FUSE_HAS_INODE_DAX) + if (fuse_conn_vdax(fc) && (flags & FUSE_HAS_INODE_DAX)) fc->inode_vdax = 1; } if (flags & FUSE_HANDLE_KILLPRIV_V2) { @@ -1423,7 +1428,7 @@ static void process_init_reply(struct fuse_args *args, int error) * and FUSE_WRITEBACK_CACHE, current design doesn't handle * them together. */ - if (IS_ENABLED(CONFIG_FUSE_PASSTHROUGH) && + if (IS_ENABLED(CONFIG_FUSE_PASSTHROUGH) && !fm->fc->is_virtiofs && (flags & (FUSE_PASSTHROUGH | FUSE_PASSTHROUGH_V2)) && arg->max_stack_depth > 0 && arg->max_stack_depth <= FILESYSTEM_MAX_STACK_DEPTH && @@ -1516,13 +1521,25 @@ static struct fuse_init_args *fuse_new_init(struct fuse_mount *fm) #ifdef CONFIG_FUSE_VDAX if (fm->fc->vdax) flags |= FUSE_MAP_ALIGNMENT; - if (fuse_is_inode_vdax_mode(fm->fc->vdax_mode)) + if (fm->fc->vdax && fuse_is_inode_vdax_mode(fm->fc->vdax_mode)) flags |= FUSE_HAS_INODE_DAX; #endif if (fm->fc->auto_submounts) flags |= FUSE_SUBMOUNTS; - if (IS_ENABLED(CONFIG_FUSE_PASSTHROUGH)) + /* + * Passthrough requires a backing file registered with + * FUSE_DEV_IOC_BACKING_OPEN on the fuse device, which a virtiofs + * server has no way to do. + */ + if (IS_ENABLED(CONFIG_FUSE_PASSTHROUGH) && !fm->fc->is_virtiofs) { flags |= FUSE_PASSTHROUGH | FUSE_PASSTHROUGH_V2; + /* + * Per inode DAX needs a dax device opened as backing, which + * is unrelated to how virtiofs uses the same flag above. + */ + if (IS_REACHABLE(CONFIG_DEV_DAX_FSDEV)) + flags |= FUSE_HAS_INODE_DAX; + } /* Only offered to sufficiently privileged servers; see * fuse_syncfs_enable(). */ @@ -1830,6 +1847,7 @@ int fuse_fill_super_common(struct super_block *sb, struct fuse_fs_context *ctx) fc->no_control = ctx->no_control; fc->no_force_umount = ctx->no_force_umount; fc->no_req_timeout = ctx->no_req_timeout; + fc->is_virtiofs = ctx->is_virtiofs; err = -ENOMEM; root = fuse_get_root_inode(sb, ctx->rootmode); diff --git a/fs/fuse/virtio_fs.c b/fs/fuse/virtio_fs.c index cdd6f5c50d6d..e17c96fa876e 100644 --- a/fs/fuse/virtio_fs.c +++ b/fs/fuse/virtio_fs.c @@ -1827,6 +1827,7 @@ static int virtio_fs_init_fs_context(struct fs_context *fsc) ctx = kzalloc_obj(struct fuse_fs_context); if (!ctx) return -ENOMEM; + ctx->is_virtiofs = true; fsc->fs_private = ctx; fsc->ops = &virtio_fs_context_ops; return 0; diff --git a/include/uapi/linux/fuse.h b/include/uapi/linux/fuse.h index 616430a2973c..cb9e9554f0ef 100644 --- a/include/uapi/linux/fuse.h +++ b/include/uapi/linux/fuse.h @@ -257,6 +257,7 @@ * - add backing_id_64 to fuse_open_out * - add FUSE_NOTIFY_BACKING_MAP, fuse_notify_backing_map_out, fuse_extent * - add FUSE_BACKING_MAP_CREATE, FUSE_BACKING_MAP_CYCLIC + * - extend the meaning of FUSE_HAS_INODE_DAX and FUSE_ATTR_DAX */ #ifndef _LINUX_FUSE_H @@ -460,7 +461,11 @@ struct fuse_file_lock { * FUSE_INIT_RESERVED: reserved, do not use * FUSE_SECURITY_CTX: add security context to create, mkdir, symlink, and * mknod - * FUSE_HAS_INODE_DAX: use per inode DAX + * FUSE_HAS_INODE_DAX: use per inode DAX. The meaning depends on the + * transport: on virtio-fs it indicates per-inode DAX + * support, while on /dev/fuse (where FUSE_PASSTHROUGH_V2 + * is also offered) it indicates support for passthrough + * extent maps on a DAX device. * FUSE_CREATE_SUPP_GROUP: add supplementary group info to create, mkdir, * symlink and mknod (single group that matches parent) * FUSE_HAS_EXPIRE_ONLY: kernel supports expiry-only entry invalidation @@ -612,7 +617,8 @@ struct fuse_file_lock { * fuse_attr flags * * FUSE_ATTR_SUBMOUNT: Object is a submount root - * FUSE_ATTR_DAX: Enable DAX for this file in per inode DAX mode + * FUSE_ATTR_DAX: Enable DAX for this file in per inode DAX mode and in + * extent mapping passthrough mode */ #define FUSE_ATTR_SUBMOUNT (1 << 0) #define FUSE_ATTR_DAX (1 << 1) |
