[PATCH] fuse: BUG_ON's correction in fuse_dev_splice_write()

[PATCH] fuse: BUG_ON's correction in fuse_dev_splice_write()

[Vasily Averin]

commit 963545357202 ("fuse: reduce allocation size for splice_write")
changed size of bufs array, so first BUG_ON should be corrected too.
Second BUG_ON become useless, first one also includes the second check:
any unsigned nbuf value cannot be less than 0.

Fixes: 963545357202 ("fuse: reduce allocation size for splice_write")
Signed-off-by: Vasily Averin <vvs@virtuozzo.com>
---
 fs/fuse/dev.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/fs/fuse/dev.c b/fs/fuse/dev.c
index ea8237513dfa..311c7911271c 100644
--- a/fs/fuse/dev.c
+++ b/fs/fuse/dev.c
@@ -2064,8 +2064,7 @@ static ssize_t fuse_dev_splice_write(struct pipe_inode_info *pipe,
 		struct pipe_buffer *ibuf;
 		struct pipe_buffer *obuf;
 
-		BUG_ON(nbuf >= pipe->buffers);
-		BUG_ON(!pipe->nrbufs);
+		BUG_ON(nbuf >= pipe->nrbufs);
 		ibuf = &pipe->bufs[pipe->curbuf];
 		obuf = &bufs[nbuf];
 
-- 
2.17.1

Re: [PATCH] fuse: BUG_ON's correction in fuse_dev_splice_write()

[Miklos Szeredi]

On Tue, Jul 23, 2019 at 8:33 AM Vasily Averin <vvs@virtuozzo.com> wrote:
>
> commit 963545357202 ("fuse: reduce allocation size for splice_write")
> changed size of bufs array, so first BUG_ON should be corrected too.
> Second BUG_ON become useless, first one also includes the second check:
> any unsigned nbuf value cannot be less than 0.

This patch seems broken: it assumes that pipe->nrbufs doesn't change.
Have you actually tested it?

Thanks,
Miklos

Re: [PATCH] fuse: BUG_ON's correction in fuse_dev_splice_write()

[Vasily Averin]

On 8/1/19 2:01 PM, Miklos Szeredi wrote:
> On Tue, Jul 23, 2019 at 8:33 AM Vasily Averin <vvs@virtuozzo.com> wrote:
>>
>> commit 963545357202 ("fuse: reduce allocation size for splice_write")
>> changed size of bufs array, so first BUG_ON should be corrected too.
>> Second BUG_ON become useless, first one also includes the second check:
>> any unsigned nbuf value cannot be less than 0.
> 
> This patch seems broken: it assumes that pipe->nrbufs doesn't change.
> Have you actually tested it?

You're right, I've missed it.
I've prepared second patch version which fixes first BUG_ON only.
checkpatch.pl also advises to replace BUG_ONs to WARN_ONs and 'unsigned' to 'unsigned int'
however I'm don't understand what it's better here:
- keep all as is, 
- or merge all changes together,
- or do it in separate patches,
- or do something else?

I believe it makes sense to remove BUG_ONs in separate patch, or may be merge it with current one,
but I do not like an idea to fight against bare 'unsigned' in fuse.

Could you please comment it?

Thank you,
	Vasily Averin

[PATCH] fuse: BUG_ON correction in fuse_dev_splice_write()

[Vasily Averin]

commit 963545357202 ("fuse: reduce allocation size for splice_write")
changed size of bufs array, so BUG_ON which checks the index of the array
shold also be fixed.

Fixes: 963545357202 ("fuse: reduce allocation size for splice_write")
Signed-off-by: Vasily Averin <vvs@virtuozzo.com>
---
 fs/fuse/dev.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/fs/fuse/dev.c b/fs/fuse/dev.c
index ea8237513dfa..f4ef6e01642c 100644
--- a/fs/fuse/dev.c
+++ b/fs/fuse/dev.c
@@ -2029,7 +2029,7 @@ static ssize_t fuse_dev_splice_write(struct pipe_inode_info *pipe,
 				     struct file *out, loff_t *ppos,
 				     size_t len, unsigned int flags)
 {
-	unsigned nbuf;
+	unsigned nbuf, bsize;
 	unsigned idx;
 	struct pipe_buffer *bufs;
 	struct fuse_copy_state cs;
@@ -2043,7 +2043,8 @@ static ssize_t fuse_dev_splice_write(struct pipe_inode_info *pipe,
 
 	pipe_lock(pipe);
 
-	bufs = kvmalloc_array(pipe->nrbufs, sizeof(struct pipe_buffer),
+	bsize = pipe->nrbufs;
+	bufs = kvmalloc_array(bsize, sizeof(struct pipe_buffer),
 			      GFP_KERNEL);
 	if (!bufs) {
 		pipe_unlock(pipe);
@@ -2064,7 +2065,7 @@ static ssize_t fuse_dev_splice_write(struct pipe_inode_info *pipe,
 		struct pipe_buffer *ibuf;
 		struct pipe_buffer *obuf;
 
-		BUG_ON(nbuf >= pipe->buffers);
+		BUG_ON(nbuf >= bsize);
 		BUG_ON(!pipe->nrbufs);
 		ibuf = &pipe->bufs[pipe->curbuf];
 		obuf = &bufs[nbuf];
-- 
2.17.1

Re: [PATCH] fuse: BUG_ON correction in fuse_dev_splice_write()

[Miklos Szeredi]

On Mon, Aug 19, 2019 at 8:53 AM Vasily Averin <vvs@virtuozzo.com> wrote:
>
> commit 963545357202 ("fuse: reduce allocation size for splice_write")
> changed size of bufs array, so BUG_ON which checks the index of the array
> shold also be fixed.
>
> Fixes: 963545357202 ("fuse: reduce allocation size for splice_write")
> Signed-off-by: Vasily Averin <vvs@virtuozzo.com>
> ---
>  fs/fuse/dev.c | 7 ++++---
>  1 file changed, 4 insertions(+), 3 deletions(-)
>
> diff --git a/fs/fuse/dev.c b/fs/fuse/dev.c
> index ea8237513dfa..f4ef6e01642c 100644
> --- a/fs/fuse/dev.c
> +++ b/fs/fuse/dev.c
> @@ -2029,7 +2029,7 @@ static ssize_t fuse_dev_splice_write(struct pipe_inode_info *pipe,
>                                      struct file *out, loff_t *ppos,
>                                      size_t len, unsigned int flags)
>  {
> -       unsigned nbuf;
> +       unsigned nbuf, bsize;
>         unsigned idx;
>         struct pipe_buffer *bufs;
>         struct fuse_copy_state cs;
> @@ -2043,7 +2043,8 @@ static ssize_t fuse_dev_splice_write(struct pipe_inode_info *pipe,
>
>         pipe_lock(pipe);
>
> -       bufs = kvmalloc_array(pipe->nrbufs, sizeof(struct pipe_buffer),
> +       bsize = pipe->nrbufs;
> +       bufs = kvmalloc_array(bsize, sizeof(struct pipe_buffer),
>                               GFP_KERNEL);
>         if (!bufs) {
>                 pipe_unlock(pipe);
> @@ -2064,7 +2065,7 @@ static ssize_t fuse_dev_splice_write(struct pipe_inode_info *pipe,
>                 struct pipe_buffer *ibuf;
>                 struct pipe_buffer *obuf;
>
> -               BUG_ON(nbuf >= pipe->buffers);
> +               BUG_ON(nbuf >= bsize);
>                 BUG_ON(!pipe->nrbufs);

Better turn these into WARN_ON's..  Fixed and applied.

Thanks,
Miklos