Re: [PATCH 1/2] fs: fork do_splice_copy_file_range() from do_splice_direct()

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Thu, Nov 30, 2023 at 3:18 PM Christian Brauner <brauner@xxxxxxxxxx> wrote:
>
> On Wed, Nov 29, 2023 at 10:07:08PM +0200, Amir Goldstein wrote:
> > The new helper is meant to be called from context of ->copy_file_range()
> > methods instead of do_splice_direct().
> >
> > Currently, the only difference is that do_splice_copy_file_range() does
> > not take a splice flags argument and it asserts that file_start_write()
> > was called.
> >
> > Soon, do_splice_direct() will be called without file_start_write() held.
> >
> > Use the new helper from __ceph_copy_file_range(), that was incorrectly
> > passing the copy_file_range() flags argument as splice flags argument
> > to do_splice_direct(). the value of flags was 0, so no actual bug fix.
> >
> > Move the definition of both helpers to linux/splice.h.
> >
> > Signed-off-by: Amir Goldstein <amir73il@xxxxxxxxx>
> > ---
> >  fs/ceph/file.c         |  9 ++---
> >  fs/read_write.c        |  6 ++--
> >  fs/splice.c            | 82 ++++++++++++++++++++++++++++++------------
> >  include/linux/fs.h     |  2 --
> >  include/linux/splice.h | 13 ++++---
> >  5 files changed, 75 insertions(+), 37 deletions(-)
> >
> > diff --git a/fs/ceph/file.c b/fs/ceph/file.c
> > index 3b5aae29e944..7c2db78e2c6e 100644
> > --- a/fs/ceph/file.c
> > +++ b/fs/ceph/file.c
> > @@ -12,6 +12,7 @@
> >  #include <linux/falloc.h>
> >  #include <linux/iversion.h>
> >  #include <linux/ktime.h>
> > +#include <linux/splice.h>
> >
> >  #include "super.h"
> >  #include "mds_client.h"
> > @@ -3010,8 +3011,8 @@ static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
> >                * {read,write}_iter, which will get caps again.
> >                */
> >               put_rd_wr_caps(src_ci, src_got, dst_ci, dst_got);
> > -             ret = do_splice_direct(src_file, &src_off, dst_file,
> > -                                    &dst_off, src_objlen, flags);
> > +             ret = do_splice_copy_file_range(src_file, &src_off, dst_file,
> > +                                             &dst_off, src_objlen);
> >               /* Abort on short copies or on error */
> >               if (ret < (long)src_objlen) {
> >                       doutc(cl, "Failed partial copy (%zd)\n", ret);
> > @@ -3065,8 +3066,8 @@ static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
> >        */
> >       if (len && (len < src_ci->i_layout.object_size)) {
> >               doutc(cl, "Final partial copy of %zu bytes\n", len);
> > -             bytes = do_splice_direct(src_file, &src_off, dst_file,
> > -                                      &dst_off, len, flags);
> > +             bytes = do_splice_copy_file_range(src_file, &src_off, dst_file,
> > +                                               &dst_off, len);
> >               if (bytes > 0)
> >                       ret += bytes;
> >               else
> > diff --git a/fs/read_write.c b/fs/read_write.c
> > index f791555fa246..555514cdad53 100644
> > --- a/fs/read_write.c
> > +++ b/fs/read_write.c
> > @@ -1423,10 +1423,8 @@ ssize_t generic_copy_file_range(struct file *file_in, loff_t pos_in,
> >                               struct file *file_out, loff_t pos_out,
> >                               size_t len, unsigned int flags)
> >  {
>
> Hm, the low-level helper takes a @flags argument but it's completely
> ignored. I think that helper should remove it or it should check:
>
> if (flags)
>         return -EINVAL;
>

It's a good point.
The upstream code and in this v1, generic_copy_file_range() can actually
be called with flag COPY_FILE_SPLICE, but it is a mistake that
I fixed it in my branch for v2, so in v2 I can add this check.

> in case it's ever called from codepaths where @flags hasn't been
> sanitized imho.
>
> > -     lockdep_assert(file_write_started(file_out));
> > -
> > -     return do_splice_direct(file_in, &pos_in, file_out, &pos_out,
> > -                             len > MAX_RW_COUNT ? MAX_RW_COUNT : len, 0);
> > +     return do_splice_copy_file_range(file_in, &pos_in, file_out, &pos_out,
> > +                             len > MAX_RW_COUNT ? MAX_RW_COUNT : len);
>
> clamp(len, 0, MAX_RW_COUNT)
>

It is a low level helper, so I don't want to worry about negative len value.
Already changed to min_t(size_t, len, MAX_RW_COUNT) in my branch.

Thanks!
Amir.





[Index of Archives]     [Linux Ext4 Filesystem]     [Union Filesystem]     [Filesystem Testing]     [Ceph Users]     [Ecryptfs]     [NTFS 3]     [AutoFS]     [Kernel Newbies]     [Share Photos]     [Security]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux Cachefs]     [Reiser Filesystem]     [Linux RAID]     [NTFS 3]     [Samba]     [Device Mapper]     [CEPH Development]

  Powered by Linux