| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Tue, 1 Mar 2022 17:50:07 -0800
From: Joe Damato <jdamato@...tly.com>
To: Saeed Mahameed <saeed@...nel.org>
Cc: netdev@...r.kernel.org, kuba@...nel.org,
ilias.apalodimas@...aro.org, davem@...emloft.net, hawk@...nel.org,
ttoukan.linux@...il.com, brouer@...hat.com, leon@...nel.org,
linux-rdma@...r.kernel.org, saeedm@...dia.com
Subject: Re: [net-next v8 4/4] mlx5: add support for page_pool_get_stats
On Tue, Mar 1, 2022 at 5:02 PM Saeed Mahameed <saeed@...nel.org> wrote:
>
> On 01 Mar 14:10, Joe Damato wrote:
> >This change adds support for the page_pool_get_stats API to mlx5. If the
> >user has enabled CONFIG_PAGE_POOL_STATS in their kernel, ethtool will
> >output page pool stats.
> >
>
> I was hoping to see something other than ethtool, a driver-less approach,
> page_pool is a first class citizen, it collects own stats and should be
> able to report own stats without the need for driver help.
>
> I understand these stats are per driver ring, but we can always come up with
> a naming convention in the page pool to allow correlating page-pool stats
> with per ring driver stats.
>
> Anyway i can't think of a simple hack, so this patch is a good temporary
> compromise until we come up with the right approach.
>
> >Signed-off-by: Joe Damato <jdamato@...tly.com>
> >---
> > drivers/net/ethernet/mellanox/mlx5/core/en_stats.c | 75 ++++++++++++++++++++++
> > drivers/net/ethernet/mellanox/mlx5/core/en_stats.h | 27 +++++++-
> > 2 files changed, 101 insertions(+), 1 deletion(-)
> >
> >diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c
> >index 3e5d8c7..eb518ec 100644
> >--- a/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c
> >+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c
> >@@ -37,6 +37,10 @@
> > #include "en/ptp.h"
> > #include "en/port.h"
> >
> >+#ifdef CONFIG_PAGE_POOL_STATS
> >+#include <net/page_pool.h>
> >+#endif
> >+
> > static unsigned int stats_grps_num(struct mlx5e_priv *priv)
> > {
> > return !priv->profile->stats_grps_num ? 0 :
> >@@ -183,6 +187,19 @@ static const struct counter_desc sw_stats_desc[] = {
> > { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_congst_umr) },
> > { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_arfs_err) },
> > { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_recover) },
> >+#ifdef CONFIG_PAGE_POOL_STATS
> >+ { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_alloc_fast) },
> >+ { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_alloc_slow) },
> >+ { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_alloc_slow_high_order) },
> >+ { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_alloc_empty) },
> >+ { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_alloc_refill) },
> >+ { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_alloc_waive) },
> >+ { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_recycle_cached) },
> >+ { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_recycle_cache_full) },
> >+ { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_recycle_ring) },
> >+ { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_recycle_ring_full) },
> >+ { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_pp_recycle_released_ref) },
> >+#endif
> > #ifdef CONFIG_MLX5_EN_TLS
> > { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_tls_decrypted_packets) },
> > { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_tls_decrypted_bytes) },
> >@@ -349,6 +366,19 @@ static void mlx5e_stats_grp_sw_update_stats_rq_stats(struct mlx5e_sw_stats *s,
> > s->rx_congst_umr += rq_stats->congst_umr;
> > s->rx_arfs_err += rq_stats->arfs_err;
> > s->rx_recover += rq_stats->recover;
> >+#ifdef CONFIG_PAGE_POOL_STATS
> >+ s->rx_pp_alloc_fast += rq_stats->pp_alloc_fast;
> >+ s->rx_pp_alloc_slow += rq_stats->pp_alloc_slow;
> >+ s->rx_pp_alloc_empty += rq_stats->pp_alloc_empty;
> >+ s->rx_pp_alloc_refill += rq_stats->pp_alloc_refill;
> >+ s->rx_pp_alloc_waive += rq_stats->pp_alloc_waive;
> >+ s->rx_pp_alloc_slow_high_order += rq_stats->pp_alloc_slow_high_order;
> >+ s->rx_pp_recycle_cached += rq_stats->pp_recycle_cached;
> >+ s->rx_pp_recycle_cache_full += rq_stats->pp_recycle_cache_full;
> >+ s->rx_pp_recycle_ring += rq_stats->pp_recycle_ring;
> >+ s->rx_pp_recycle_ring_full += rq_stats->pp_recycle_ring_full;
> >+ s->rx_pp_recycle_released_ref += rq_stats->pp_recycle_released_ref;
> >+#endif
> > #ifdef CONFIG_MLX5_EN_TLS
> > s->rx_tls_decrypted_packets += rq_stats->tls_decrypted_packets;
> > s->rx_tls_decrypted_bytes += rq_stats->tls_decrypted_bytes;
> >@@ -455,6 +485,35 @@ static void mlx5e_stats_grp_sw_update_stats_qos(struct mlx5e_priv *priv,
> > }
> > }
> >
> >+#ifdef CONFIG_PAGE_POOL_STATS
> >+static void mlx5e_stats_update_stats_rq_page_pool(struct mlx5e_channel *c)
> >+{
> >+ struct mlx5e_rq_stats *rq_stats = c->rq.stats;
> >+ struct page_pool *pool = c->rq.page_pool;
> >+ struct page_pool_stats stats = { 0 };
> >+
> you can drop the 0, just {} should be enough.
>
> >+ if (!page_pool_get_stats(pool, &stats))
> >+ return;
> >+
>
> you can contain the whole page_pool_stats objects inside rq_stats object,
> and avoid all the assignments below.
>
> just do:
> page_pool_get_stats(pool, &rq_stats.pp);
> return;
I don't think I can because the maximum stat name size is 32 bytes
(ETH_GSTRING_LEN).
If I do what you are suggesting, I would need to do something like:
{ MLX5E_DECLARE_RX_STAT(struct mlx5e_rq_stats,
pp.recycle_stats.released_refcnt) }
which will generate a string of the form:
"rx%d_pp.recycle_stats.released_refcnt" which is well over 32 bytes,
especially with double-digit queue numbers.
The only options I see are:
- A new define that allows setting a custom field name
(MLX5E_DECLARE_RX_STAT_NAME_OVERRIDE ?), or
- Leaving the code as-is
Can you let me know what you prefer for the v9?
Thanks,
Joe
Powered by blists - more mailing lists