| Message ID | 20240129151507.14885-2-tony.solomonik@gmail.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | io_uring: add support for ftruncate | expand |
On Mon, Jan 29, 2024 at 05:15:06PM +0200, Tony Solomonik wrote: > do_sys_ftruncate receives a file descriptor, fgets the struct file, and > finally actually truncates the file. > > do_ftruncate allows for passing in a file directly, with the caller > already holding a reference to it. > > Signed-off-by: Tony Solomonik <tony.solomonik@gmail.com> Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
On Mon, Jan 29, 2024 at 05:15:06PM +0200, Tony Solomonik wrote: > do_sys_ftruncate receives a file descriptor, fgets the struct file, and > finally actually truncates the file. > > do_ftruncate allows for passing in a file directly, with the caller > already holding a reference to it. > > Signed-off-by: Tony Solomonik <tony.solomonik@gmail.com> > --- > fs/internal.h | 1 + > fs/open.c | 52 ++++++++++++++++++++++++++++----------------------- > 2 files changed, 30 insertions(+), 23 deletions(-) > > diff --git a/fs/internal.h b/fs/internal.h > index 58e43341aebf..d35b1c05cf6d 100644 > --- a/fs/internal.h > +++ b/fs/internal.h > @@ -182,6 +182,7 @@ extern struct open_how build_open_how(int flags, umode_t mode); > extern int build_open_flags(const struct open_how *how, struct open_flags *op); > extern struct file *__close_fd_get_file(unsigned int fd); > > +long do_ftruncate(struct file *file, loff_t length); > long do_sys_ftruncate(unsigned int fd, loff_t length, int small); > int chmod_common(const struct path *path, umode_t mode); > int do_fchownat(int dfd, const char __user *filename, uid_t user, gid_t group, > diff --git a/fs/open.c b/fs/open.c > index 02dc608d40d8..6d608ff4a3f7 100644 > --- a/fs/open.c > +++ b/fs/open.c > @@ -154,49 +154,55 @@ COMPAT_SYSCALL_DEFINE2(truncate, const char __user *, path, compat_off_t, length > } > #endif > > -long do_sys_ftruncate(unsigned int fd, loff_t length, int small) > +long do_ftruncate(struct file *file, loff_t length) > { > struct inode *inode; > struct dentry *dentry; > + int error; > + > + dentry = file->f_path.dentry; > + inode = dentry->d_inode; > + if (!S_ISREG(inode->i_mode) || !(file->f_mode & FMODE_WRITE)) > + return -EINVAL; > + > + /* Check IS_APPEND on real upper inode */ > + if (IS_APPEND(file_inode(file))) > + return -EPERM; > + sb_start_write(inode->i_sb); > + error = security_file_truncate(file); > + if (!error) > + error = do_truncate(file_mnt_idmap(file), dentry, length, > + ATTR_MTIME | ATTR_CTIME, file); > + sb_end_write(inode->i_sb); > + > + return error; > +} > + > +long do_sys_ftruncate(unsigned int fd, loff_t length, int small) > +{ > struct fd f; > int error; > > - error = -EINVAL; > if (length < 0) > - goto out; > - error = -EBADF; > + return -EINVAL; > f = fdget(fd); > if (!f.file) > - goto out; > + return -EBADF; > > /* explicitly opened as large or we are on 64-bit box */ > if (f.file->f_flags & O_LARGEFILE) > small = 0; Why is the O_LARGEFILE handling not needed when used from io_uring?
On 1/31/24 2:14 AM, Tony Solomonik wrote: > Actually, I'm not quite sure anymore, @Jens Axboe > <mailto:axboe@kernel.dk> is there any guarantee in io_uring that the > file is always opened as LARGE / 64 bit? From looking at the code, it > simply accepts a user made fd, so the user might have not opened it as > LARGE on a 32bit system, which might be bad news. Yeah, we probably want to retain that. Though it'd be a very odd case where an application using io_uring isn't opening "large" files by default, but we'd still have to ensure that it is.
On Thu, Feb 01, 2024 at 04:20:06PM -0700, Jens Axboe wrote: > On 1/31/24 2:14 AM, Tony Solomonik wrote: > > Actually, I'm not quite sure anymore, @Jens Axboe > > <mailto:axboe@kernel.dk> is there any guarantee in io_uring that the > > file is always opened as LARGE / 64 bit? From looking at the code, it > > simply accepts a user made fd, so the user might have not opened it as > > LARGE on a 32bit system, which might be bad news. > > Yeah, we probably want to retain that. Though it'd be a very odd case > where an application using io_uring isn't opening "large" files by > default, but we'd still have to ensure that it is. Oh; my mistake. I thought we required O_LARGEFILE when using io_uring.
diff --git a/fs/internal.h b/fs/internal.h index 58e43341aebf..d35b1c05cf6d 100644 --- a/fs/internal.h +++ b/fs/internal.h @@ -182,6 +182,7 @@ extern struct open_how build_open_how(int flags, umode_t mode); extern int build_open_flags(const struct open_how *how, struct open_flags *op); extern struct file *__close_fd_get_file(unsigned int fd); +long do_ftruncate(struct file *file, loff_t length); long do_sys_ftruncate(unsigned int fd, loff_t length, int small); int chmod_common(const struct path *path, umode_t mode); int do_fchownat(int dfd, const char __user *filename, uid_t user, gid_t group, diff --git a/fs/open.c b/fs/open.c index 02dc608d40d8..6d608ff4a3f7 100644 --- a/fs/open.c +++ b/fs/open.c @@ -154,49 +154,55 @@ COMPAT_SYSCALL_DEFINE2(truncate, const char __user *, path, compat_off_t, length } #endif -long do_sys_ftruncate(unsigned int fd, loff_t length, int small) +long do_ftruncate(struct file *file, loff_t length) { struct inode *inode; struct dentry *dentry; + int error; + + dentry = file->f_path.dentry; + inode = dentry->d_inode; + if (!S_ISREG(inode->i_mode) || !(file->f_mode & FMODE_WRITE)) + return -EINVAL; + + /* Check IS_APPEND on real upper inode */ + if (IS_APPEND(file_inode(file))) + return -EPERM; + sb_start_write(inode->i_sb); + error = security_file_truncate(file); + if (!error) + error = do_truncate(file_mnt_idmap(file), dentry, length, + ATTR_MTIME | ATTR_CTIME, file); + sb_end_write(inode->i_sb); + + return error; +} + +long do_sys_ftruncate(unsigned int fd, loff_t length, int small) +{ struct fd f; int error; - error = -EINVAL; if (length < 0) - goto out; - error = -EBADF; + return -EINVAL; f = fdget(fd); if (!f.file) - goto out; + return -EBADF; /* explicitly opened as large or we are on 64-bit box */ if (f.file->f_flags & O_LARGEFILE) small = 0; - dentry = f.file->f_path.dentry; - inode = dentry->d_inode; error = -EINVAL; - if (!S_ISREG(inode->i_mode) || !(f.file->f_mode & FMODE_WRITE)) - goto out_putf; - error = -EINVAL; /* Cannot ftruncate over 2^31 bytes without large file support */ if (small && length > MAX_NON_LFS) - goto out_putf; + goto out; + + error = do_ftruncate(f.file, length); - error = -EPERM; - /* Check IS_APPEND on real upper inode */ - if (IS_APPEND(file_inode(f.file))) - goto out_putf; - sb_start_write(inode->i_sb); - error = security_file_truncate(f.file); - if (!error) - error = do_truncate(file_mnt_idmap(f.file), dentry, length, - ATTR_MTIME | ATTR_CTIME, f.file); - sb_end_write(inode->i_sb); -out_putf: - fdput(f); out: + fdput(f); return error; }
do_sys_ftruncate receives a file descriptor, fgets the struct file, and finally actually truncates the file. do_ftruncate allows for passing in a file directly, with the caller already holding a reference to it. Signed-off-by: Tony Solomonik <tony.solomonik@gmail.com> --- fs/internal.h | 1 + fs/open.c | 52 ++++++++++++++++++++++++++++----------------------- 2 files changed, 30 insertions(+), 23 deletions(-)