Skip to contentSkip to conversation

net: dpaa:reduce number of synchronize_net() calls

PATCH Applied Xi Huang 3 participants 4 messages Aug 22, 2024 – Aug 26, 2024

Subsystems (3)
lore
Share & tools
lore

Xi Huang

author
#0 lore raw

PATCH v2 1/1 net: dpaa:reduce number of synchronize_net() calls

In the function dpaa_napi_del(), we execute the netif_napi_del() for each cpu, which is actually a high overhead operation because each call to netif_napi_del() contains a synchronize_net(), i.e. an RCU operation. In fact, it is only necessary to call
__netif_napi_del and use synchronize_net() once outside of the loop.
This change is similar to commit 2543a6000e593a ("gro_cells: reduce number of synchronize_net() calls") and commit 5198d545dba8ad (" net:
remove napi_hash_del() from driver-facing API") 5198d545db.

---
V1 -> V2: Modify the cited commit format and remove useless information

 drivers/net/ethernet/freescale/dpaa/dpaa_eth.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/net/ethernet/freescale/dpaa/dpaa_eth.c +2 -1
index cfe6b57b1..5d99cfb4e 100644
@@ -3156,8 +3156,9 @@ static void dpaa_napi_del(struct net_device *net_dev)
 	for_each_possible_cpu(cpu) {
 		percpu_priv = per_cpu_ptr(priv->percpu_priv, cpu);
 
-		netif_napi_del(&percpu_priv->np.napi);
+		__netif_napi_del(&percpu_priv->np.napi);
 	}
+	synchronize_net();
 }
 
 static inline void dpaa_bp_free_pf(const struct dpaa_bp *bp,
Signature · 1 line
2.34.1
maintainer
#1 lore raw
Quoted 11 lines · On Thu, 22 Aug 2024 15:20:42 +0800 Xi Huang wrote:

In the function dpaa_napi_del(), we execute the netif_napi_del() for each cpu, which is actually a high overhead operation because each call to netif_napi_del() contains a synchronize_net(), i.e. an RCU operation. In fact, it is only necessary to call
__netif_napi_del and use synchronize_net() once outside of the loop.
This change is similar to commit 2543a6000e593a ("gro_cells: reduce number of synchronize_net() calls") and commit 5198d545dba8ad (" net:
remove napi_hash_del() from driver-facing API") 5198d545db.

You missed the part of Eric's response that told you to wait 24 hours:
https://www.kernel.org/doc/html/next/process/maintainer-netdev.html

Signature · 1 line
pv-bot: 24h
author
#2 lore raw

I am very sorry, I misunderstood the meaning of “Please send a V2 in ~24 hours”, I mistakenly thought that it could be sent within 24 hours. Do I need to resend the patchv2 24 hours after the first patchv1?Sorry again and thanks so much for the heads up!

#3 lore raw
In reply to Xi Huang
This patch was applied to netdev/net-next.git (main) by Jakub Kicinski <[email protected]>:

Hello:

This patch was applied to netdev/net-next.git (main) by Jakub Kicinski <[email protected]>:

Quoted 10 lines · On Thu, 22 Aug 2024 15:20:42 +0800 you wrote:

In the function dpaa_napi_del(), we execute the netif_napi_del() for each cpu, which is actually a high overhead operation because each call to netif_napi_del() contains a synchronize_net(), i.e. an RCU operation. In fact, it is only necessary to call
__netif_napi_del and use synchronize_net() once outside of the loop.
This change is similar to commit 2543a6000e593a ("gro_cells: reduce number of synchronize_net() calls") and commit 5198d545dba8ad (" net:
remove napi_hash_del() from driver-facing API") 5198d545db.

[...]

Here is the summary with links:
- [PATCHv2] net: dpaa:reduce number of synchronize_net() calls

    https://git.kernel.org/netdev/net-next/c/2c163922de69

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html