[PATCH net-next 0/3] Compile-flag for sock RX queue mapping

[PATCH net-next 0/3] Compile-flag for sock RX queue mapping

[Tariq Toukan]

Hi,

Socket's RX queue mapping logic is useful also for non-XPS use cases.
This series breaks the dependency between the two, introducing a new
kernel config flag SOCK_RX_QUEUE_MAPPING.

Here we select this new kernel flag from TLS_DEVICE, as well as XPS.

Regards,
Tariq

Tariq Toukan (3):
  net/sock: Add kernel config SOCK_RX_QUEUE_MAPPING
  net/tls: Select SOCK_RX_QUEUE_MAPPING from TLS_DEVICE
  net/mlx5: Remove TLS dependencies on XPS

 drivers/net/ethernet/mellanox/mlx5/core/Kconfig |  2 --
 include/net/sock.h                              | 12 ++++++------
 net/Kconfig                                     |  4 ++++
 net/core/filter.c                               |  2 +-
 net/tls/Kconfig                                 |  1 +
 5 files changed, 12 insertions(+), 9 deletions(-)

-- 
2.21.0

[PATCH net-next 1/3] net/sock: Add kernel config SOCK_RX_QUEUE_MAPPING

[Tariq Toukan]

Use a new config SOCK_RX_QUEUE_MAPPING to compile-in the socket
RX queue field and logic, instead of the XPS config.
This breaks dependency in XPS, and allows selecting it from non-XPS
use cases, as we do in the next patch.

In addition, use the new flag to wrap the logic in sk_rx_queue_get()
and protect access to the sk_rx_queue_mapping field, while keeping
the function exposed unconditionally, just like sk_rx_queue_set()
and sk_rx_queue_clear().

Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Maxim Mikityanskiy <maximmi@nvidia.com>
---
 include/net/sock.h | 12 ++++++------
 net/Kconfig        |  4 ++++
 net/core/filter.c  |  2 +-
 3 files changed, 11 insertions(+), 7 deletions(-)

diff --git a/include/net/sock.h b/include/net/sock.h
index 690e496a0e79..855c068c6c86 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -226,7 +226,7 @@ struct sock_common {
 		struct hlist_nulls_node skc_nulls_node;
 	};
 	unsigned short		skc_tx_queue_mapping;
-#ifdef CONFIG_XPS
+#ifdef CONFIG_SOCK_RX_QUEUE_MAPPING
 	unsigned short		skc_rx_queue_mapping;
 #endif
 	union {
@@ -356,7 +356,7 @@ struct sock {
 #define sk_nulls_node		__sk_common.skc_nulls_node
 #define sk_refcnt		__sk_common.skc_refcnt
 #define sk_tx_queue_mapping	__sk_common.skc_tx_queue_mapping
-#ifdef CONFIG_XPS
+#ifdef CONFIG_SOCK_RX_QUEUE_MAPPING
 #define sk_rx_queue_mapping	__sk_common.skc_rx_queue_mapping
 #endif
 
@@ -1838,7 +1838,7 @@ static inline int sk_tx_queue_get(const struct sock *sk)
 
 static inline void sk_rx_queue_set(struct sock *sk, const struct sk_buff *skb)
 {
-#ifdef CONFIG_XPS
+#ifdef CONFIG_SOCK_RX_QUEUE_MAPPING
 	if (skb_rx_queue_recorded(skb)) {
 		u16 rx_queue = skb_get_rx_queue(skb);
 
@@ -1852,20 +1852,20 @@ static inline void sk_rx_queue_set(struct sock *sk, const struct sk_buff *skb)
 
 static inline void sk_rx_queue_clear(struct sock *sk)
 {
-#ifdef CONFIG_XPS
+#ifdef CONFIG_SOCK_RX_QUEUE_MAPPING
 	sk->sk_rx_queue_mapping = NO_QUEUE_MAPPING;
 #endif
 }
 
-#ifdef CONFIG_XPS
 static inline int sk_rx_queue_get(const struct sock *sk)
 {
+#ifdef CONFIG_SOCK_RX_QUEUE_MAPPING
 	if (sk && sk->sk_rx_queue_mapping != NO_QUEUE_MAPPING)
 		return sk->sk_rx_queue_mapping;
+#endif
 
 	return -1;
 }
-#endif
 
 static inline void sk_set_socket(struct sock *sk, struct socket *sock)
 {
diff --git a/net/Kconfig b/net/Kconfig
index f4c32d982af6..8cea808ad9e8 100644
--- a/net/Kconfig
+++ b/net/Kconfig
@@ -256,9 +256,13 @@ config RFS_ACCEL
 	select CPU_RMAP
 	default y
 
+config SOCK_RX_QUEUE_MAPPING
+	bool
+
 config XPS
 	bool
 	depends on SMP
+	select SOCK_RX_QUEUE_MAPPING
 	default y
 
 config HWBM
diff --git a/net/core/filter.c b/net/core/filter.c
index 9ab94e90d660..2ba03f6597bb 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -8816,7 +8816,7 @@ u32 bpf_sock_convert_ctx_access(enum bpf_access_type type,
 				       target_size));
 		break;
 	case offsetof(struct bpf_sock, rx_queue_mapping):
-#ifdef CONFIG_XPS
+#ifdef CONFIG_SOCK_RX_QUEUE_MAPPING
 		*insn++ = BPF_LDX_MEM(
 			BPF_FIELD_SIZEOF(struct sock, sk_rx_queue_mapping),
 			si->dst_reg, si->src_reg,
-- 
2.21.0

[PATCH net-next 2/3] net/tls: Select SOCK_RX_QUEUE_MAPPING from TLS_DEVICE

[Tariq Toukan]

Compile-in the socket RX queue mapping field and logic when TLS_DEVICE
is enabled. This allows device drivers to pick the recorded socket's
RX queue and use it for streams distribution.

Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Maxim Mikityanskiy <maximmi@nvidia.com>
---
 net/tls/Kconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/net/tls/Kconfig b/net/tls/Kconfig
index fa0724fd84b4..0cdc1f7b6b08 100644
--- a/net/tls/Kconfig
+++ b/net/tls/Kconfig
@@ -21,6 +21,7 @@ config TLS_DEVICE
 	bool "Transport Layer Security HW offload"
 	depends on TLS
 	select SOCK_VALIDATE_XMIT
+	select SOCK_RX_QUEUE_MAPPING
 	default n
 	help
 	Enable kernel support for HW offload of the TLS protocol.
-- 
2.21.0

[PATCH net-next 3/3] net/mlx5: Remove TLS dependencies on XPS

[Tariq Toukan]

No real dependency on XPS, but on RX queue mapping, which
is being selected by TLS_DEVICE.

Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Maxim Mikityanskiy <maximmi@nvidia.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/Kconfig | 2 --
 1 file changed, 2 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/Kconfig b/drivers/net/ethernet/mellanox/mlx5/core/Kconfig
index ad45d20f9d44..372d902bca5a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/Kconfig
+++ b/drivers/net/ethernet/mellanox/mlx5/core/Kconfig
@@ -166,7 +166,6 @@ config MLX5_FPGA_TLS
 	depends on TLS=y || MLX5_CORE=m
 	depends on MLX5_CORE_EN
 	depends on MLX5_FPGA
-	depends on XPS
 	select MLX5_EN_TLS
 	default n
 	help
@@ -181,7 +180,6 @@ config MLX5_TLS
 	depends on TLS_DEVICE
 	depends on TLS=y || MLX5_CORE=m
 	depends on MLX5_CORE_EN
-	depends on XPS
 	select MLX5_ACCEL
 	select MLX5_EN_TLS
 	default n
-- 
2.21.0

Re: [PATCH net-next 0/3] Compile-flag for sock RX queue mapping

[Jakub Kicinski]

On Thu, 11 Feb 2021 13:35:50 +0200 Tariq Toukan wrote:
> Socket's RX queue mapping logic is useful also for non-XPS use cases.
> This series breaks the dependency between the two, introducing a new
> kernel config flag SOCK_RX_QUEUE_MAPPING.
> 
> Here we select this new kernel flag from TLS_DEVICE, as well as XPS.

Acked-by: Jakub Kicinski <kuba@kernel.org>

Re: [PATCH net-next 0/3] Compile-flag for sock RX queue mapping

[patchwork-bot+netdevbpf]

Hello:

This series was applied to netdev/net-next.git (refs/heads/master):

On Thu, 11 Feb 2021 13:35:50 +0200 you wrote:
> Hi,
> 
> Socket's RX queue mapping logic is useful also for non-XPS use cases.
> This series breaks the dependency between the two, introducing a new
> kernel config flag SOCK_RX_QUEUE_MAPPING.
> 
> Here we select this new kernel flag from TLS_DEVICE, as well as XPS.
> 
> [...]

Here is the summary with links:
  - [net-next,1/3] net/sock: Add kernel config SOCK_RX_QUEUE_MAPPING
    https://git.kernel.org/netdev/net-next/c/4e1beecc3b58
  - [net-next,2/3] net/tls: Select SOCK_RX_QUEUE_MAPPING from TLS_DEVICE
    https://git.kernel.org/netdev/net-next/c/76f165939ea3
  - [net-next,3/3] net/mlx5: Remove TLS dependencies on XPS
    https://git.kernel.org/netdev/net-next/c/2af3e35c5a04

You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html