| Message ID | 20230308143754.1976726-4-dhowells@redhat.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | splice, block: Use page pinning and kill ITER_PIPE | expand |
On Wed, 8 Mar 2023 at 15:38, David Howells <dhowells@redhat.com> wrote: > > Implement splice-read for overlayfs by passing the request down a layer > rather than going through generic_file_splice_read() which is going to be > changed to assume that ->read_folio() is present on buffered files. > > Signed-off-by: David Howells <dhowells@redhat.com> > cc: Christoph Hellwig <hch@lst.de> > cc: Jens Axboe <axboe@kernel.dk> > cc: Al Viro <viro@zeniv.linux.org.uk> > cc: John Hubbard <jhubbard@nvidia.com> > cc: David Hildenbrand <david@redhat.com> > cc: Matthew Wilcox <willy@infradead.org> > cc: Miklos Szeredi <miklos@szeredi.hu> > cc: linux-unionfs@vger.kernel.org > cc: linux-block@vger.kernel.org > cc: linux-fsdevel@vger.kernel.org > cc: linux-mm@kvack.org > --- > > Notes: > ver #15) > - Remove redundant FMODE_CAN_ODIRECT check on real file. > - Do rw_verify_area() on the real file, not the overlay file. > - Fix a file leak. > > fs/overlayfs/file.c | 33 ++++++++++++++++++++++++++++++++- > 1 file changed, 32 insertions(+), 1 deletion(-) > > diff --git a/fs/overlayfs/file.c b/fs/overlayfs/file.c > index 7c04f033aadd..a12919e9ccba 100644 > --- a/fs/overlayfs/file.c > +++ b/fs/overlayfs/file.c > @@ -419,6 +419,37 @@ static ssize_t ovl_write_iter(struct kiocb *iocb, struct iov_iter *iter) > return ret; > } > > +static ssize_t ovl_splice_read(struct file *in, loff_t *ppos, > + struct pipe_inode_info *pipe, size_t len, > + unsigned int flags) > +{ > + const struct cred *old_cred; > + struct fd real; > + ssize_t ret; > + > + ret = ovl_real_fdget(in, &real); > + if (ret) > + return ret; > + > + ret = -EINVAL; > + if (!real.file->f_op->splice_read) > + goto out_fdput; > + > + ret = rw_verify_area(READ, real.file, ppos, len); > + if (unlikely(ret < 0)) > + goto out_fdput; > + > + old_cred = ovl_override_creds(file_inode(in)->i_sb); > + ret = real.file->f_op->splice_read(real.file, ppos, pipe, len, flags); I don't think you replied to my suggestion of using a helper here. E.g. it could be as simple as exporting do_splice_to(), or renaming it to vfs_splice_read() to be more readable. It would remove the boilerplate and be more robust if any changes are done to the splice reading code. Thanks, Miklos
Miklos Szeredi <miklos@szeredi.hu> wrote: > > + ret = -EINVAL; > > + if (!real.file->f_op->splice_read) > > + goto out_fdput; > > + > > + ret = rw_verify_area(READ, real.file, ppos, len); > > + if (unlikely(ret < 0)) > > + goto out_fdput; > > + > > + old_cred = ovl_override_creds(file_inode(in)->i_sb); > > + ret = real.file->f_op->splice_read(real.file, ppos, pipe, len, flags); > > I don't think you replied to my suggestion of using a helper here. > E.g. it could be as simple as exporting do_splice_to(), or renaming it > to vfs_splice_read() to be more readable. It would remove the > boilerplate and be more robust if any changes are done to the splice > reading code. Using do_splice_to() as a helper is probably a good idea, though both Willy and Christoph seem to dislike it. The pipe occupancy check has already been done, so I'm not sure if it should be repeated - though it probably wouldn't hurt. David
On Wed, Mar 08, 2023 at 03:54:36PM +0000, David Howells wrote: > Using do_splice_to() as a helper is probably a good idea, though both Willy > and Christoph seem to dislike it. That's not true. What I'm fundamentlly against is pointless wrappers like the call_* that add no value. do_splice_to adds useful checks, so if properly named and documented, I'm absolutely in favour.
Christoph Hellwig <hch@lst.de> wrote: > On Wed, Mar 08, 2023 at 03:54:36PM +0000, David Howells wrote: > > Using do_splice_to() as a helper is probably a good idea, though both Willy > > and Christoph seem to dislike it. > > That's not true. What I'm fundamentlly against is pointless wrappers > like the call_* that add no value. do_splice_to adds useful checks, > so if properly named and documented, I'm absolutely in favour. Fair enough. Rename to vfs_splice_read() okay with you? David
On Wed, Mar 08, 2023 at 04:04:39PM +0000, David Howells wrote: > Christoph Hellwig <hch@lst.de> wrote: > > > On Wed, Mar 08, 2023 at 03:54:36PM +0000, David Howells wrote: > > > Using do_splice_to() as a helper is probably a good idea, though both Willy > > > and Christoph seem to dislike it. > > > > That's not true. What I'm fundamentlly against is pointless wrappers > > like the call_* that add no value. do_splice_to adds useful checks, > > so if properly named and documented, I'm absolutely in favour. > > Fair enough. Rename to vfs_splice_read() okay with you? Yes.
diff --git a/fs/overlayfs/file.c b/fs/overlayfs/file.c index 7c04f033aadd..a12919e9ccba 100644 --- a/fs/overlayfs/file.c +++ b/fs/overlayfs/file.c @@ -419,6 +419,37 @@ static ssize_t ovl_write_iter(struct kiocb *iocb, struct iov_iter *iter) return ret; } +static ssize_t ovl_splice_read(struct file *in, loff_t *ppos, + struct pipe_inode_info *pipe, size_t len, + unsigned int flags) +{ + const struct cred *old_cred; + struct fd real; + ssize_t ret; + + ret = ovl_real_fdget(in, &real); + if (ret) + return ret; + + ret = -EINVAL; + if (!real.file->f_op->splice_read) + goto out_fdput; + + ret = rw_verify_area(READ, real.file, ppos, len); + if (unlikely(ret < 0)) + goto out_fdput; + + old_cred = ovl_override_creds(file_inode(in)->i_sb); + ret = real.file->f_op->splice_read(real.file, ppos, pipe, len, flags); + + revert_creds(old_cred); + ovl_file_accessed(in); +out_fdput: + fdput(real); + + return ret; +} + /* * Calling iter_file_splice_write() directly from overlay's f_op may deadlock * due to lock order inversion between pipe->mutex in iter_file_splice_write() @@ -695,7 +726,7 @@ const struct file_operations ovl_file_operations = { .fallocate = ovl_fallocate, .fadvise = ovl_fadvise, .flush = ovl_flush, - .splice_read = generic_file_splice_read, + .splice_read = ovl_splice_read, .splice_write = ovl_splice_write, .copy_file_range = ovl_copy_file_range,
Implement splice-read for overlayfs by passing the request down a layer rather than going through generic_file_splice_read() which is going to be changed to assume that ->read_folio() is present on buffered files. Signed-off-by: David Howells <dhowells@redhat.com> cc: Christoph Hellwig <hch@lst.de> cc: Jens Axboe <axboe@kernel.dk> cc: Al Viro <viro@zeniv.linux.org.uk> cc: John Hubbard <jhubbard@nvidia.com> cc: David Hildenbrand <david@redhat.com> cc: Matthew Wilcox <willy@infradead.org> cc: Miklos Szeredi <miklos@szeredi.hu> cc: linux-unionfs@vger.kernel.org cc: linux-block@vger.kernel.org cc: linux-fsdevel@vger.kernel.org cc: linux-mm@kvack.org --- Notes: ver #15) - Remove redundant FMODE_CAN_ODIRECT check on real file. - Do rw_verify_area() on the real file, not the overlay file. - Fix a file leak. fs/overlayfs/file.c | 33 ++++++++++++++++++++++++++++++++- 1 file changed, 32 insertions(+), 1 deletion(-)