Linux Hardening
 help / color / mirror / Atom feed
* Re: [PATCH net-next 05/12] net/mlx5: fs, add mlx5_fs_pool API
       [not found]     ` <70b3a7b5-abd3-4db4-8415-e0467a565847@nvidia.com>
@ 2024-12-18  5:22       ` Kees Cook
  2024-12-18  8:21         ` Moshe Shemesh
  0 siblings, 1 reply; 2+ messages in thread
From: Kees Cook @ 2024-12-18  5:22 UTC (permalink / raw)
  To: Moshe Shemesh, morbo, qing.zhao
  Cc: Simon Horman, Tariq Toukan, David S. Miller, Jakub Kicinski,
	Paolo Abeni, Eric Dumazet, Andrew Lunn, Leon Romanovsky, netdev,
	Saeed Mahameed, Gal Pressman, linux-rdma, Yevgeny Kliteynik,
	Mark Bloch, linux-hardening

On Sun, Dec 15, 2024 at 03:39:11PM +0200, Moshe Shemesh wrote:
> 
> 
> On 12/12/2024 7:23 PM, Simon Horman wrote:
> > External email: Use caution opening links or attachments
> > 
> > 
> > On Wed, Dec 11, 2024 at 03:42:16PM +0200, Tariq Toukan wrote:
> > > From: Moshe Shemesh <moshe@nvidia.com>
> > > 
> > > Refactor fc_pool API to create generic fs_pool API, as HW steering has
> > > more flow steering elements which can take advantage of the same pool of
> > > bulks API. Change fs_counters code to use the fs_pool API.
> > > 
> > > Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
> > > Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
> > > Reviewed-by: Mark Bloch <mbloch@nvidia.com>
> > > Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
> > 
> > ...
> > 
> > > diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
> > 
> > ...
> > 
> > > @@ -447,11 +437,9 @@ void mlx5_fc_update_sampling_interval(struct mlx5_core_dev *dev,
> > >   /* Flow counter bluks */
> > > 
> > >   struct mlx5_fc_bulk {
> > > -     struct list_head pool_list;
> > > +     struct mlx5_fs_bulk fs_bulk;
> > >        u32 base_id;
> > > -     int bulk_len;
> > > -     unsigned long *bitmask;
> > > -     struct mlx5_fc fcs[] __counted_by(bulk_len);
> > > +     struct mlx5_fc fcs[] __counted_by(fs_bulk.bulk_len);
> > >   };
> > 
> > Unfortunately it seems that clang-19 doesn't know how to handle
> > __counted_by() when used like this:
> > 
> > drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c:442:36: error: 'counted_by' argument must be a simple declaration reference
> >    442 |         struct mlx5_fc fcs[] __counted_by(fs_bulk.bulk_len);
> 
> Thanks Simon, from code perspective, bulk_len should be now in the inner
> struct fs_bulk.
> 
> Keen Cook, is that going to be supported in the future? for now I will just
> remove __counted_by() from this struct.

I am expecting this will be supported in the future, yes, but there isn't
an ETA for it yet. Neither GCC 15 nor Clang are currently supporting
sub-struct members -- the counted_by member needs to be at the same
"level" in the struct as the annotated flexible array.

Is it possible to move "base_id" above "fs_bulk" and move "fcs" into
the of end struct mlx5_fs_bulk?

-Kees

-- 
Kees Cook

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [PATCH net-next 05/12] net/mlx5: fs, add mlx5_fs_pool API
  2024-12-18  5:22       ` [PATCH net-next 05/12] net/mlx5: fs, add mlx5_fs_pool API Kees Cook
@ 2024-12-18  8:21         ` Moshe Shemesh
  0 siblings, 0 replies; 2+ messages in thread
From: Moshe Shemesh @ 2024-12-18  8:21 UTC (permalink / raw)
  To: Kees Cook, morbo, qing.zhao
  Cc: Simon Horman, Tariq Toukan, David S. Miller, Jakub Kicinski,
	Paolo Abeni, Eric Dumazet, Andrew Lunn, Leon Romanovsky, netdev,
	Saeed Mahameed, Gal Pressman, linux-rdma, Yevgeny Kliteynik,
	Mark Bloch, linux-hardening



On 12/18/2024 7:22 AM, Kees Cook wrote:
> 
> On Sun, Dec 15, 2024 at 03:39:11PM +0200, Moshe Shemesh wrote:
>>
>>
>> On 12/12/2024 7:23 PM, Simon Horman wrote:
>>> External email: Use caution opening links or attachments
>>>
>>>
>>> On Wed, Dec 11, 2024 at 03:42:16PM +0200, Tariq Toukan wrote:
>>>> From: Moshe Shemesh <moshe@nvidia.com>
>>>>
>>>> Refactor fc_pool API to create generic fs_pool API, as HW steering has
>>>> more flow steering elements which can take advantage of the same pool of
>>>> bulks API. Change fs_counters code to use the fs_pool API.
>>>>
>>>> Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
>>>> Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
>>>> Reviewed-by: Mark Bloch <mbloch@nvidia.com>
>>>> Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
>>>
>>> ...
>>>
>>>> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c
>>>
>>> ...
>>>
>>>> @@ -447,11 +437,9 @@ void mlx5_fc_update_sampling_interval(struct mlx5_core_dev *dev,
>>>>    /* Flow counter bluks */
>>>>
>>>>    struct mlx5_fc_bulk {
>>>> -     struct list_head pool_list;
>>>> +     struct mlx5_fs_bulk fs_bulk;
>>>>         u32 base_id;
>>>> -     int bulk_len;
>>>> -     unsigned long *bitmask;
>>>> -     struct mlx5_fc fcs[] __counted_by(bulk_len);
>>>> +     struct mlx5_fc fcs[] __counted_by(fs_bulk.bulk_len);
>>>>    };
>>>
>>> Unfortunately it seems that clang-19 doesn't know how to handle
>>> __counted_by() when used like this:
>>>
>>> drivers/net/ethernet/mellanox/mlx5/core/fs_counters.c:442:36: error: 'counted_by' argument must be a simple declaration reference
>>>     442 |         struct mlx5_fc fcs[] __counted_by(fs_bulk.bulk_len);
>>
>> Thanks Simon, from code perspective, bulk_len should be now in the inner
>> struct fs_bulk.
>>
>> Keen Cook, is that going to be supported in the future? for now I will just
>> remove __counted_by() from this struct.
> 
> I am expecting this will be supported in the future, yes, but there isn't
> an ETA for it yet. Neither GCC 15 nor Clang are currently supporting
> sub-struct members -- the counted_by member needs to be at the same
> "level" in the struct as the annotated flexible array.
> 
> Is it possible to move "base_id" above "fs_bulk" and move "fcs" into
> the of end struct mlx5_fs_bulk?
> 

fcs is specific for mlx5_fc_bulk and so can't be part of fs_bulk.
In follow up patches for hws feature I am using mlx5_fs_bulk in other 
type of bulks, fcs is specific for mlx5_fc_bulk.
Unfortunately, I don't see a nice way to have bulk_len and fcs on the 
same struct.

> -Kees
> 
> --
> Kees Cook

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2024-12-18  8:22 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
     [not found] <20241211134223.389616-1-tariqt@nvidia.com>
     [not found] ` <20241211134223.389616-6-tariqt@nvidia.com>
     [not found]   ` <20241212172355.GE73795@kernel.org>
     [not found]     ` <70b3a7b5-abd3-4db4-8415-e0467a565847@nvidia.com>
2024-12-18  5:22       ` [PATCH net-next 05/12] net/mlx5: fs, add mlx5_fs_pool API Kees Cook
2024-12-18  8:21         ` Moshe Shemesh

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox