When an application invokes posix_fadvise(..., POSIX_FADV_DONTNEED) or other fadvise advice on a FUSE passthrough file descriptor, the VFS currently only executes generic_fadvise() on the upper FUSE inode mapping. Because FUSE passthrough does not forward fadvise operations to the underlying backing file, pages residing in the lower filesystem's pagecache remain resident in memory. During media-heavy workloads (such as image thumbnail generation or streaming), single-access file data accumulates in the inactive pagecache of the lower filesystem. Under memory pressure, the kernel retains these stale media pages and instead evicts active executable pages from running processes, leading to severe direct reclaim latency spikes. Architecturally, Linux VFS supports filesystem-specific fadvise delegation via the .fadvise hook in struct file_operations. Implement fuse_file_fadvise() in fs/fuse/file.c and register it in fuse_file_operations. When passthrough is active on a fuse_file, fuse_file_fadvise() forwards the vfs_fadvise() call to the lower backing_file under the opened credentials, allowing POSIX_FADV_DONTNEED to cleanly invalidate the physical backing filesystem pagecache. Signed-off-by: Shai Barack Cc: Miklos Szeredi Cc: Amir Goldstein Cc: Bernd Schubert Cc: Sandeep Dhavale Cc: Akilesh Kailash Cc: Suren Baghdasaryan Cc: fuse-devel@lists.linux.dev Cc: linux-fsdevel@vger.kernel.org Cc: linux-kernel@vger.kernel.org --- v2: - Remove redundant !backing_file check in fuse_passthrough_fadvise() (Amir Goldstein) - Remove unused fuse_passthrough_fadvise() stub under #ifndef CONFIG_FUSE_PASSTHROUGH (Amir Goldstein) - Reorder fuse_passthrough_fadvise() above fuse_passthrough_open() alongside other passthrough ops (Amir Goldstein) - Add fuse-devel@lists.linux.dev to CC list (Amir Goldstein, TJ Mercier) fs/fuse/file.c | 11 +++++++++++ fs/fuse/fuse_i.h | 1 + fs/fuse/passthrough.c | 13 +++++++++++++ 3 files changed, 25 insertions(+) diff --git a/fs/fuse/file.c b/fs/fuse/file.c index a4d736945..81709ab82 100644 --- a/fs/fuse/file.c +++ b/fs/fuse/file.c @@ -3537,6 +3537,16 @@ static ssize_t fuse_copy_file_range(struct file *src_file, loff_t src_off, return ret; } +static int fuse_file_fadvise(struct file *file, loff_t offset, loff_t len, int advice) +{ + struct fuse_file *ff = file->private_data; + + if (fuse_file_passthrough(ff)) + return fuse_passthrough_fadvise(ff, offset, len, advice); + + return generic_fadvise(file, offset, len, advice); +} + static const struct file_operations fuse_file_operations = { .llseek = fuse_file_llseek, .read_iter = fuse_file_read_iter, @@ -3557,6 +3567,7 @@ static const struct file_operations fuse_file_operations = { .fallocate = fuse_file_fallocate, .copy_file_range = fuse_copy_file_range, .fop_flags = FOP_DONTCACHE, + .fadvise = fuse_file_fadvise, }; static const struct address_space_operations fuse_file_aops = { diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h index 939553b34..6c60716e7 100644 --- a/fs/fuse/fuse_i.h +++ b/fs/fuse/fuse_i.h @@ -1581,6 +1581,7 @@ ssize_t fuse_passthrough_splice_write(struct pipe_inode_info *pipe, struct file *out, loff_t *ppos, size_t len, unsigned int flags); ssize_t fuse_passthrough_mmap(struct file *file, struct vm_area_struct *vma); +int fuse_passthrough_fadvise(struct fuse_file *ff, loff_t offset, loff_t len, int advice); /* backing.c */ diff --git a/fs/fuse/passthrough.c b/fs/fuse/passthrough.c index 059e57969..8a8fa6b8c 100644 --- a/fs/fuse/passthrough.c +++ b/fs/fuse/passthrough.c @@ -160,6 +160,19 @@ ssize_t fuse_passthrough_mmap(struct file *file, struct vm_area_struct *vma) return backing_file_mmap(backing_file, vma, &ctx); } +int fuse_passthrough_fadvise(struct fuse_file *ff, loff_t offset, loff_t len, int advice) +{ + struct file *backing_file = fuse_file_passthrough(ff); + const struct cred *old_cred; + int ret; + + old_cred = override_creds(ff->cred); + ret = vfs_fadvise(backing_file, offset, len, advice); + revert_creds(old_cred); + + return ret; +} + struct fuse_backing *fuse_backing_get(struct fuse_backing *fb) { if (fb && refcount_inc_not_zero(&fb->count)) -- 2.43.0