linux-next: manual merge of the fuse tree with Linus' tree

linux-next: manual merge of the fuse tree with Linus' tree

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 737 bytes --]

Hi all,

Today's linux-next merge of the fuse tree got a conflict in:

  Documentation/filesystems/index.rst

between commit:

  77ce1a47ebca ("docs: filesystems: add overlayfs to index.rst")

from Linus' tree and commit:

  8ab13bca428b ("Documentation: filesystems: convert fuse to RST")

from the fuse tree.

I fixed it up (see below) and can carry the fix as necessary. This
is now fixed as far as linux-next is concerned, but any non trivial
conflicts should be mentioned to your upstream maintainer when your tree
is submitted for merging.  You may also want to consider cooperating
with the maintainer of the conflicting tree to minimise any particularly
complex conflicts.

-- 
Cheers,
Stephen Rothwell

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

linux-next: manual merge of the fuse tree with Linus' tree

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1668 bytes --]

Hi all,

Today's linux-next merge of the fuse tree got a conflict in:

  fs/fuse/file.c

between commit:

  933a3752babc ("fuse: fix the ->direct_IO() treatment of iov_iter")

from Linus' tree and commit:

  fcee216beb9c ("fuse: split fuse_mount off of fuse_conn")

from the fuse tree.

I fixed it up (see below) and can carry the fix as necessary. This
is now fixed as far as linux-next is concerned, but any non trivial
conflicts should be mentioned to your upstream maintainer when your tree
is submitted for merging.  You may also want to consider cooperating
with the maintainer of the conflicting tree to minimise any particularly
complex conflicts.

-- 
Cheers,
Stephen Rothwell

diff --cc fs/fuse/file.c
index 43c165e796da,53d4dd1ab992..000000000000
--- a/fs/fuse/file.c
+++ b/fs/fuse/file.c
@@@ -3120,17 -3156,10 +3146,18 @@@ fuse_direct_IO(struct kiocb *iocb, stru
  	 * By default, we want to optimize all I/Os with async request
  	 * submission to the client filesystem if supported.
  	 */
- 	io->async = ff->fc->async_dio;
 -	io->async = async_dio;
++	io->async = ff->fm->fc->async_dio;
  	io->iocb = iocb;
  	io->blocking = is_sync_kiocb(iocb);
  
 +	/* optimization for short read */
 +	if (io->async && !io->write && offset + count > i_size) {
- 		iov_iter_truncate(iter, fuse_round_up(ff->fc, i_size - offset));
++		iov_iter_truncate(iter, fuse_round_up(ff->fm->fc,
++				  i_size - offset));
 +		shortened = count - iov_iter_count(iter);
 +		count -= shortened;
 +	}
 +
  	/*
  	 * We cannot asynchronously extend the size of a file.
  	 * In such case the aio will behave exactly like sync io.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: linux-next: manual merge of the fuse tree with Linus' tree

[Max Reitz]

[-- Attachment #1.1: Type: text/plain, Size: 2090 bytes --]

On 13.10.20 01:47, Stephen Rothwell wrote:
> Hi all,
> 
> Today's linux-next merge of the fuse tree got a conflict in:
> 
>   fs/fuse/file.c
> 
> between commit:
> 
>   933a3752babc ("fuse: fix the ->direct_IO() treatment of iov_iter")
> 
> from Linus' tree and commit:
> 
>   fcee216beb9c ("fuse: split fuse_mount off of fuse_conn")
> 
> from the fuse tree.
> 
> I fixed it up (see below) and can carry the fix as necessary. This
> is now fixed as far as linux-next is concerned, but any non trivial
> conflicts should be mentioned to your upstream maintainer when your tree
> is submitted for merging.  You may also want to consider cooperating
> with the maintainer of the conflicting tree to minimise any particularly
> complex conflicts.

Thanks!  Semantically, the change looks good to me, I just noticed
something about the indentation:

> diff --cc fs/fuse/file.c
> index 43c165e796da,53d4dd1ab992..000000000000
> --- a/fs/fuse/file.c
> +++ b/fs/fuse/file.c
> @@@ -3120,17 -3156,10 +3146,18 @@@ fuse_direct_IO(struct kiocb *iocb, stru
>   	 * By default, we want to optimize all I/Os with async request
>   	 * submission to the client filesystem if supported.
>   	 */
> - 	io->async = ff->fc->async_dio;
>  -	io->async = async_dio;
> ++	io->async = ff->fm->fc->async_dio;
>   	io->iocb = iocb;
>   	io->blocking = is_sync_kiocb(iocb);
>   
>  +	/* optimization for short read */
>  +	if (io->async && !io->write && offset + count > i_size) {
> - 		iov_iter_truncate(iter, fuse_round_up(ff->fc, i_size - offset));
> ++		iov_iter_truncate(iter, fuse_round_up(ff->fm->fc,
> ++				  i_size - offset));

Personally, I’d align this second line to the opening parenthesis of
fuse_round_up() (as it is done in fcee216beb9c, in a hunk not shown
here, probably because it’s been dropped in the merge.).

Max

>  +		shortened = count - iov_iter_count(iter);
>  +		count -= shortened;
>  +	}
>  +
>   	/*
>   	 * We cannot asynchronously extend the size of a file.
>   	 * In such case the aio will behave exactly like sync io.



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

linux-next: manual merge of the fuse tree with Linus' tree

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1088 bytes --]

Hi all,

Today's linux-next merge of the fuse tree got a conflict in:

  include/uapi/linux/fuse.h

between commit:

  7d37539037c2 ("fuse: implement ->tmpfile()")

from Linus' tree and commit:

  22fb2e02a194 ("fuse: add "expire only" mode to FUSE_NOTIFY_INVAL_ENTRY")

from the fuse tree.

I fixed it up (see below) and can carry the fix as necessary. This
is now fixed as far as linux-next is concerned, but any non trivial
conflicts should be mentioned to your upstream maintainer when your tree
is submitted for merging.  You may also want to consider cooperating
with the maintainer of the conflicting tree to minimise any particularly
complex conflicts.

-- 
Cheers,
Stephen Rothwell

diff --cc include/uapi/linux/fuse.h
index 76ee8f9e024a,1688fdea550f..000000000000
--- a/include/uapi/linux/fuse.h
+++ b/include/uapi/linux/fuse.h
@@@ -196,7 -196,7 +196,8 @@@
   *  - add FUSE_HAS_INODE_DAX, FUSE_ATTR_DAX
   *
   *  7.37
 + *  - add FUSE_TMPFILE
+  *  - add FUSE_EXPIRE_ONLY flag to fuse_notify_inval_entry
   */
  
  #ifndef _LINUX_FUSE_H

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

linux-next: manual merge of the fuse tree with Linus' tree

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1145 bytes --]

Hi all,

Today's linux-next merge of the fuse tree got a conflict in:

  fs/fuse/inode.c

between commit:

  053fc4f755ad ("fuse: fix UAF in rcu pathwalks")

from Linus' tree and commit:

  9fc953f59383 ("fuse: implement ioctls to manage backing files")

from the fuse tree.

I fixed it up (see below) and can carry the fix as necessary. This
is now fixed as far as linux-next is concerned, but any non trivial
conflicts should be mentioned to your upstream maintainer when your tree
is submitted for merging.  You may also want to consider cooperating
with the maintainer of the conflicting tree to minimise any particularly
complex conflicts.

-- 
Cheers,
Stephen Rothwell

diff --cc fs/fuse/inode.c
index 516ea2979a90,c26a84439934..000000000000
--- a/fs/fuse/inode.c
+++ b/fs/fuse/inode.c
@@@ -1373,7 -1395,9 +1402,9 @@@ EXPORT_SYMBOL_GPL(fuse_send_init)
  void fuse_free_conn(struct fuse_conn *fc)
  {
  	WARN_ON(!list_empty(&fc->devices));
+ 	if (IS_ENABLED(CONFIG_FUSE_PASSTHROUGH))
+ 		fuse_backing_files_free(fc);
 -	kfree_rcu(fc, rcu);
 +	kfree(fc);
  }
  EXPORT_SYMBOL_GPL(fuse_free_conn);
  

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

linux-next: manual merge of the fuse tree with Linus' tree

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1311 bytes --]

Hi all,

Today's linux-next merge of the fuse tree got a conflict in:

  fs/fuse/inode.c

between commit:

  053fc4f755ad ("fuse: fix UAF in rcu pathwalks")

from Linus' tree and commit:

  44350256ab94 ("fuse: implement ioctls to manage backing files")

from the fuse tree.

I fixed it up (I think? see below) and can carry the fix as
necessary. This is now fixed as far as linux-next is concerned, but any
non trivial conflicts should be mentioned to your upstream maintainer
when your tree is submitted for merging.  You may also want to consider
cooperating with the maintainer of the conflicting tree to minimise any
particularly complex conflicts.

-- 
Cheers,
Stephen Rothwell

diff --cc fs/fuse/inode.c
index 516ea2979a90,02869edf72f3..000000000000
--- a/fs/fuse/inode.c
+++ b/fs/fuse/inode.c
@@@ -930,14 -942,6 +942,16 @@@ void fuse_conn_init(struct fuse_conn *f
  }
  EXPORT_SYMBOL_GPL(fuse_conn_init);
  
 +static void delayed_release(struct rcu_head *p)
 +{
 +	struct fuse_conn *fc = container_of(p, struct fuse_conn, rcu);
 +
 +	put_user_ns(fc->user_ns);
++	if (IS_ENABLED(CONFIG_FUSE_PASSTHROUGH))
++		fuse_backing_files_free(fc);
 +	fc->release(fc);
 +}
 +
  void fuse_conn_put(struct fuse_conn *fc)
  {
  	if (refcount_dec_and_test(&fc->count)) {

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: linux-next: manual merge of the fuse tree with Linus' tree

[Miklos Szeredi]

On Wed, 6 Mar 2024 at 00:34, Stephen Rothwell <sfr@canb.auug.org.au> wrote:

> diff --cc fs/fuse/inode.c
> index 516ea2979a90,02869edf72f3..000000000000
> --- a/fs/fuse/inode.c
> +++ b/fs/fuse/inode.c
> @@@ -930,14 -942,6 +942,16 @@@ void fuse_conn_init(struct fuse_conn *f
>   }
>   EXPORT_SYMBOL_GPL(fuse_conn_init);
>
>  +static void delayed_release(struct rcu_head *p)
>  +{
>  +      struct fuse_conn *fc = container_of(p, struct fuse_conn, rcu);
>  +
>  +      put_user_ns(fc->user_ns);
> ++      if (IS_ENABLED(CONFIG_FUSE_PASSTHROUGH))
> ++              fuse_backing_files_free(fc);
>  +      fc->release(fc);
>  +}
>  +

fuse_backing_files_free() doesn't need to be called after an RCU
delay, so it should be resolved like this:

diff --cc fs/fuse/inode.c
index 516ea2979a90,02869edf72f3..000000000000
--- a/fs/fuse/inode.c
+++ b/fs/fuse/inode.c
@@@ -954,7 -959,9 +966,9 @@@ void fuse_conn_put(struct fuse_conn *fc
                        WARN_ON(atomic_read(&bucket->count) != 1);
                        kfree(bucket);
                }
+               if (IS_ENABLED(CONFIG_FUSE_PASSTHROUGH))
+                       fuse_backing_files_free(fc);
 -              fc->release(fc);
 +              call_rcu(&fc->rcu, delayed_release);
        }
  }
  EXPORT_SYMBOL_GPL(fuse_conn_put);

Thanks,
Miklos

Re: linux-next: manual merge of the fuse tree with Linus' tree

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1579 bytes --]

Hi Miklos,

On Thu, 7 Mar 2024 15:16:00 +0100 Miklos Szeredi <miklos@szeredi.hu> wrote:
>
> On Wed, 6 Mar 2024 at 00:34, Stephen Rothwell <sfr@canb.auug.org.au> wrote:
> 
> > diff --cc fs/fuse/inode.c
> > index 516ea2979a90,02869edf72f3..000000000000
> > --- a/fs/fuse/inode.c
> > +++ b/fs/fuse/inode.c
> > @@@ -930,14 -942,6 +942,16 @@@ void fuse_conn_init(struct fuse_conn *f
> >   }
> >   EXPORT_SYMBOL_GPL(fuse_conn_init);
> >
> >  +static void delayed_release(struct rcu_head *p)
> >  +{
> >  +      struct fuse_conn *fc = container_of(p, struct fuse_conn, rcu);
> >  +
> >  +      put_user_ns(fc->user_ns);
> > ++      if (IS_ENABLED(CONFIG_FUSE_PASSTHROUGH))
> > ++              fuse_backing_files_free(fc);
> >  +      fc->release(fc);
> >  +}
> >  +  
> 
> fuse_backing_files_free() doesn't need to be called after an RCU
> delay, so it should be resolved like this:
> 
> diff --cc fs/fuse/inode.c
> index 516ea2979a90,02869edf72f3..000000000000
> --- a/fs/fuse/inode.c
> +++ b/fs/fuse/inode.c
> @@@ -954,7 -959,9 +966,9 @@@ void fuse_conn_put(struct fuse_conn *fc
>                         WARN_ON(atomic_read(&bucket->count) != 1);
>                         kfree(bucket);
>                 }
> +               if (IS_ENABLED(CONFIG_FUSE_PASSTHROUGH))
> +                       fuse_backing_files_free(fc);
>  -              fc->release(fc);
>  +              call_rcu(&fc->rcu, delayed_release);
>         }
>   }
>   EXPORT_SYMBOL_GPL(fuse_conn_put);

Thanks, I will do that from today.

-- 
Cheers,
Stephen Rothwell

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]