Re: [PATCH net-next v4 09/17] net: pktgen: align some variable declarations to the most common pattern

[Peter Seiderer <ps.report@gmx.net>]

Hello Simon,

On Thu, 6 Feb 2025 13:25:38 +0000, Simon Horman <horms@kernel.org> wrote:

> On Wed, Feb 05, 2025 at 02:11:45PM +0100, Peter Seiderer wrote:
> > Align some variable declarations (in get_imix_entries and get_labels) to
> > the most common pattern (int instead of ssize_t/long) and adjust function
> > return value accordingly.
> >
> > Signed-off-by: Peter Seiderer <ps.report@gmx.net>
>
> Hi Peter,
>
> These comments are is true in general of this patchset, but particularly so
> in the case of this patch:
>
> * I think a more succinct subject would be nice.
> * I think the patch description should provide some reason
>   _why_ the change is being made.

Yep, will improve...

>
> Also, specifically relating to this patch, I wonder if it's scope ought to
> be extended. For example, the two callers of num_arg(), get_imix_entries() and
> pktgen_if_write() assign the return value of num_arg() to len, which is now
> an int in both functions. But num_args() returns a long.

Aim was to get rid of the int/long mixture in the code (which works flawless
because no one writes to proc with more than a few bytes AND count is limited
to INT_MAX - PAGE_SIZE in vfs_write (see [1], [2])...

I believe the clean way is to use

  size_t i, max;
  ssize_t len;

consequently through out the code and adjust the function signatures
accordingly...., will re-spin...

>
> > ---
> > Changes v3 -> v4
> >   - new patch (factored out of patch 'net: pktgen: fix access outside of user
> >     given buffer in pktgen_if_write()')
> > ---
> >  net/core/pktgen.c | 14 ++++++--------
> >  1 file changed, 6 insertions(+), 8 deletions(-)
> >
> > diff --git a/net/core/pktgen.c b/net/core/pktgen.c
> > index 4f201a2db2dc..279910367ad4 100644
> > --- a/net/core/pktgen.c
> > +++ b/net/core/pktgen.c
> > @@ -850,12 +850,11 @@ static int strn_len(const char __user * user_buffer, unsigned int maxlen)
> >   * where each entry consists of size and weight delimited by commas.
> >   * "size1,weight_1 size2,weight_2 ... size_n,weight_n" for example.
> >   */
> > -static ssize_t get_imix_entries(const char __user *buffer,
> > -				struct pktgen_dev *pkt_dev)
> > +static int get_imix_entries(const char __user *buffer,
> > +			    struct pktgen_dev *pkt_dev)
> >  {
> > -	int i = 0;
> > -	long len;
> >  	char c;
> > +	int i = 0, len;
>
> Given it can be achieved with exactly the same lines changed, just in a
> different order, please arrange the local variable declarations in reverse
> xmas tree order - longest line to shortest.
>
> Likewise for the other hunk of this patch.  And I believe there are also
> other cases in this patchset where this comment applied.
>
> The following tool can be useful:
> https://github.com/ecree-solarflare/xmastree

O.k. will take a look at it...

Thanks for review!

Regards,
Peter


[1] https://elixir.bootlin.com/linux/v6.13.1/source/fs/read_write.c#L673
[2] https://elixir.bootlin.com/linux/v6.13.1/source/include/linux/fs.h#L2704