[PATCH net] net: drv: netdevsim: don't napi_complete() from netpoll

[PATCH net] net: drv: netdevsim: don't napi_complete() from netpoll

[Jakub Kicinski]

netdevsim supports netpoll. Make sure we don't call napi_complete()
from it, since it may not be scheduled. Breno reports hitting a
warning in napi_complete_done():

WARNING: CPU: 14 PID: 104 at net/core/dev.c:6592 napi_complete_done+0x2cc/0x560
  __napi_poll+0x2d8/0x3a0
  handle_softirqs+0x1fe/0x710

This is presumably after netpoll stole the SCHED bit prematurely.

Reported-by: Breno Leitao <leitao@debian.org>
Fixes: 3762ec05a9fb ("netdevsim: add NAPI support")
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
CC: dw@davidwei.uk
---
 drivers/net/netdevsim/netdev.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/net/netdevsim/netdev.c b/drivers/net/netdevsim/netdev.c
index af545d42961c..fa5fbd97ad69 100644
--- a/drivers/net/netdevsim/netdev.c
+++ b/drivers/net/netdevsim/netdev.c
@@ -371,7 +371,8 @@ static int nsim_poll(struct napi_struct *napi, int budget)
 	int done;
 
 	done = nsim_rcv(rq, budget);
-	napi_complete(napi);
+	if (done < budget)
+		napi_complete_done(napi, done);
 
 	return done;
 }
-- 
2.49.0

Re: [PATCH net] net: drv: netdevsim: don't napi_complete() from netpoll

[Dave Marquardt]

Jakub Kicinski <kuba@kernel.org> writes:

> netdevsim supports netpoll. Make sure we don't call napi_complete()
> from it, since it may not be scheduled. Breno reports hitting a
> warning in napi_complete_done():

I decided to go learn a bit more about netdevsim, and ran across a typo
in Documentation/networking/devlink/netdevsim.rst:

The ``netdevsim`` driver supports rate objects management, which includes:

- registerging/unregistering leaf rate objects per VF devlink port;
  ^^^^^^^^^^^^

-Dave

Re: [PATCH net] net: drv: netdevsim: don't napi_complete() from netpoll

[Breno Leitao]

Hello Dave,

On Wed, Jun 11, 2025 at 05:23:33PM -0500, Dave Marquardt wrote:
> Jakub Kicinski <kuba@kernel.org> writes:
> 
> > netdevsim supports netpoll. Make sure we don't call napi_complete()
> > from it, since it may not be scheduled. Breno reports hitting a
> > warning in napi_complete_done():
> 
> I decided to go learn a bit more about netdevsim, and ran across a typo
> in Documentation/networking/devlink/netdevsim.rst:

I acknowledge the bug exists in the latest netnext/main repo. Would you
like to send a patch and fix it, or, should we?

Thanks for reporting it.
--breno

Re: [PATCH net] net: drv: netdevsim: don't napi_complete() from netpoll

[Dave Marquardt]

Breno Leitao <leitao@debian.org> writes:

> Hello Dave,
>
> On Wed, Jun 11, 2025 at 05:23:33PM -0500, Dave Marquardt wrote:
>> Jakub Kicinski <kuba@kernel.org> writes:
>> 
>> > netdevsim supports netpoll. Make sure we don't call napi_complete()
>> > from it, since it may not be scheduled. Breno reports hitting a
>> > warning in napi_complete_done():
>> 
>> I decided to go learn a bit more about netdevsim, and ran across a typo
>> in Documentation/networking/devlink/netdevsim.rst:
>
> I acknowledge the bug exists in the latest netnext/main repo. Would you
> like to send a patch and fix it, or, should we?

I'll send in a patch. Thanks.

> Thanks for reporting it.

You're welcome!

-Dave

Re: [PATCH net] net: drv: netdevsim: don't napi_complete() from netpoll

[Breno Leitao]

Hello Jakub,

On Wed, Jun 11, 2025 at 10:46:43AM -0700, Jakub Kicinski wrote:
> netdevsim supports netpoll. Make sure we don't call napi_complete()
> from it, since it may not be scheduled. Breno reports hitting a
> warning in napi_complete_done():
> 
> WARNING: CPU: 14 PID: 104 at net/core/dev.c:6592 napi_complete_done+0x2cc/0x560
>   __napi_poll+0x2d8/0x3a0
>   handle_softirqs+0x1fe/0x710
> 
> This is presumably after netpoll stole the SCHED bit prematurely.
> 
> Reported-by: Breno Leitao <leitao@debian.org>
> Fixes: 3762ec05a9fb ("netdevsim: add NAPI support")
> Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Tested-by: Breno Leitao <leitao@debian.org>

Thanks for the quick fix,
--breno

Re: [PATCH net] net: drv: netdevsim: don't napi_complete() from netpoll

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to netdev/net.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Wed, 11 Jun 2025 10:46:43 -0700 you wrote:
> netdevsim supports netpoll. Make sure we don't call napi_complete()
> from it, since it may not be scheduled. Breno reports hitting a
> warning in napi_complete_done():
> 
> WARNING: CPU: 14 PID: 104 at net/core/dev.c:6592 napi_complete_done+0x2cc/0x560
>   __napi_poll+0x2d8/0x3a0
>   handle_softirqs+0x1fe/0x710
> 
> [...]

Here is the summary with links:
  - [net] net: drv: netdevsim: don't napi_complete() from netpoll
    https://git.kernel.org/netdev/net/c/1264971017b4

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