[PATCH net v3] batman-adv: Fix double-hold of meshif when getting enabled

[PATCH net v3] batman-adv: Fix double-hold of meshif when getting enabled

[Sven Eckelmann]

It was originally meant to replace the dev_hold with netdev_hold. But this
was missed in this place and thus there was an imbalance when trying to
remove the interfaces.

Fixes: 00b35530811f ("batman-adv: adopt netdev_hold() / netdev_put()")
Signed-off-by: Sven Eckelmann <sven@narfation.org>
---
 net/batman-adv/hard-interface.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/net/batman-adv/hard-interface.c b/net/batman-adv/hard-interface.c
index f145f9662653..7cd4bdcee439 100644
--- a/net/batman-adv/hard-interface.c
+++ b/net/batman-adv/hard-interface.c
@@ -725,7 +725,6 @@ int batadv_hardif_enable_interface(struct batadv_hard_iface *hard_iface,
 
 	kref_get(&hard_iface->refcount);
 
-	dev_hold(mesh_iface);
 	netdev_hold(mesh_iface, &hard_iface->meshif_dev_tracker, GFP_ATOMIC);
 	hard_iface->mesh_iface = mesh_iface;
 	bat_priv = netdev_priv(hard_iface->mesh_iface);
-- 
2.39.5

Re: [PATCH net v3] batman-adv: Fix double-hold of meshif when getting enabled

[Paolo Abeni]

On 4/9/25 9:35 AM, Sven Eckelmann wrote:
> It was originally meant to replace the dev_hold with netdev_hold. But this
> was missed in this place and thus there was an imbalance when trying to
> remove the interfaces.
> 
> Fixes: 00b35530811f ("batman-adv: adopt netdev_hold() / netdev_put()")
> Signed-off-by: Sven Eckelmann <sven@narfation.org>

You must wait at least 24h before reposting, see:

https://elixir.bootlin.com/linux/v6.14-rc6/source/Documentation/process/maintainer-netdev.rst#L15

Also this is somewhat strange: the same patch come from 2 different
persons (sometimes with garbled SoBs), and we usually receive PR for
batman patches.

@Svev, @Simon: could you please double check you are on the same page
and clarify the intent here?

Thanks,

Paolo

Re: [PATCH net v3] batman-adv: Fix double-hold of meshif when getting enabled

[Eric Dumazet]

On Thu, Apr 10, 2025 at 12:13 PM Paolo Abeni <pabeni@redhat.com> wrote:
>
> On 4/9/25 9:35 AM, Sven Eckelmann wrote:
> > It was originally meant to replace the dev_hold with netdev_hold. But this
> > was missed in this place and thus there was an imbalance when trying to
> > remove the interfaces.
> >
> > Fixes: 00b35530811f ("batman-adv: adopt netdev_hold() / netdev_put()")
> > Signed-off-by: Sven Eckelmann <sven@narfation.org>
>
> You must wait at least 24h before reposting, see:
>
> https://elixir.bootlin.com/linux/v6.14-rc6/source/Documentation/process/maintainer-netdev.rst#L15
>
> Also this is somewhat strange: the same patch come from 2 different
> persons (sometimes with garbled SoBs), and we usually receive PR for
> batman patches.
>
> @Svev, @Simon: could you please double check you are on the same page
> and clarify the intent here?

Also I do not see credits given to syzbot  team ?

https://lkml.org/lkml/2025/4/8/1988

Re: [PATCH net v3] batman-adv: Fix double-hold of meshif when getting enabled

[Simon Wunderlich]

[-- Attachment #1: Type: text/plain, Size: 1293 bytes --]

On Thursday, April 10, 2025 1:20:21 PM CEST Eric Dumazet wrote:
> On Thu, Apr 10, 2025 at 12:13 PM Paolo Abeni <pabeni@redhat.com> wrote:
> > On 4/9/25 9:35 AM, Sven Eckelmann wrote:
> > > It was originally meant to replace the dev_hold with netdev_hold. But
> > > this
> > > was missed in this place and thus there was an imbalance when trying to
> > > remove the interfaces.
> > > 
> > > Fixes: 00b35530811f ("batman-adv: adopt netdev_hold() / netdev_put()")
> > > Signed-off-by: Sven Eckelmann <sven@narfation.org>
> > 
> > You must wait at least 24h before reposting, see:
> > 
> > https://elixir.bootlin.com/linux/v6.14-rc6/source/Documentation/process/ma
> > intainer-netdev.rst#L15
> > 
> > Also this is somewhat strange: the same patch come from 2 different
> > persons (sometimes with garbled SoBs), and we usually receive PR for
> > batman patches.
> > 
> > @Svev, @Simon: could you please double check you are on the same page
> > and clarify the intent here?
> 
> Also I do not see credits given to syzbot  team ?
> 
> https://lkml.org/lkml/2025/4/8/1988

Sorry, there were some mistakes on our side. We'll take a deep breath and 
repost this on another day, most likely as part of a usual pull request. Sorry 
for the noise.

Cheers,
       Simon

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [PATCH net v3] batman-adv: Fix double-hold of meshif when getting enabled

[Sven Eckelmann]

[-- Attachment #1: Type: text/plain, Size: 3028 bytes --]

On Thursday, 10 April 2025 12:13:25 CEST Paolo Abeni wrote:
> Also this is somewhat strange: the same patch come from 2 different
> persons (sometimes with garbled SoBs), 

Ok, all of that was my fault. Even the original duplicated *dev_hold. Just as summary:

* v1: accidentally submitted the patch with correct content but from a test
  git repo which didn't had the actual commit message
* v2: send correct patch but it was done from the folder which is used to 
  prepare the branches which will be submitted later by Simon -> git-send-
  email picked up the Simon's name (see below) and submitted it using my 
  mailserver
* v3: submitted it "correctly" and marked the day as "shouldn't have waken up 
  in the first place"

Regarding the v2 situation: This is definitely odd but it had to be done this 
way because there were complains in the past from netdev when Simon submitted 
the PR and not all patches in the PR were Signed-off-by him. As result, when I 
add something in the queue, I directly apply the patches as him (including my 
own Signed-off-by). And Simon will go through the patches again before 
actually sending the PR, create a signed tag and submits the PR. I would love 
not to do this preparation/fakery anymore. But then you will not have the 
requested full signed-off-by - something which you usually don't have for PRs 
but which was for some reason required for netdev.

> and we usually receive PR for
> batman patches.

This was just an attempt to get syzbot happy again (sooner). Besides my direct 
patch submission, we have different options:

* wait for Simon's PR
* let Eric Dumazet integrate his (earlier posted) patch from
  https://lore.kernel.org/r/CANn89iJTHf-sJCqcyrFJiLMLBOBgtN0+KrfPSuW0mhOzLS08Rw@mail.gmail.com
  This change was also published earlier

> Also I do not see credits given to syzbot  team ?

Correct. Is there a proper way when the reports received were actually about 
different problems (just the bisecting went the wrong way due to the 
batman-adv bug)?

For example, I saw the problem with bisecting in:

* https://syzkaller.appspot.com/bug?extid=ff3aa851d46ab82953a3
  Reported-by: syzbot+ff3aa851d46ab82953a3@syzkaller.appspotmail.com
* https://syzkaller.appspot.com/bug?extid=4036165fc595a74b09b2
  Reported-by: syzbot+4036165fc595a74b09b2@syzkaller.appspotmail.com
* https://syzkaller.appspot.com/bug?extid=c35d73ce910d86c0026e
  Reported-by: syzbot+c35d73ce910d86c0026e@syzkaller.appspotmail.com
* https://syzkaller.appspot.com/bug?extid=48c14f61594bdfadb086
  Reported-by: syzbot+48c14f61594bdfadb086@syzkaller.appspotmail.com

So, a lot of different problems which unfortunately all ended up in bisecting 
to the batman-adv problem.

On Thursday, 10 April 2025 13:20:21 CEST Eric Dumazet wrote:
> https://lkml.org/lkml/2025/4/8/1988

You also posted your patch here. Feel free to directly add it. And sorry for 
adding the problem in the first placed - just tried to make Antonio happy (and 
then created a big mess for everyone else)

Kind regards,
	Sven

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

Re: [PATCH net v3] batman-adv: Fix double-hold of meshif when getting enabled

[Eric Dumazet]

On Thu, Apr 10, 2025 at 3:08 PM Sven Eckelmann <sven@narfation.org> wrote:
>
> On Thursday, 10 April 2025 12:13:25 CEST Paolo Abeni wrote:
> > Also this is somewhat strange: the same patch come from 2 different
> > persons (sometimes with garbled SoBs),
>
> Ok, all of that was my fault. Even the original duplicated *dev_hold. Just as summary:
>
> * v1: accidentally submitted the patch with correct content but from a test
>   git repo which didn't had the actual commit message
> * v2: send correct patch but it was done from the folder which is used to
>   prepare the branches which will be submitted later by Simon -> git-send-
>   email picked up the Simon's name (see below) and submitted it using my
>   mailserver
> * v3: submitted it "correctly" and marked the day as "shouldn't have waken up
>   in the first place"
>
> Regarding the v2 situation: This is definitely odd but it had to be done this
> way because there were complains in the past from netdev when Simon submitted
> the PR and not all patches in the PR were Signed-off-by him. As result, when I
> add something in the queue, I directly apply the patches as him (including my
> own Signed-off-by). And Simon will go through the patches again before
> actually sending the PR, create a signed tag and submits the PR. I would love
> not to do this preparation/fakery anymore. But then you will not have the
> requested full signed-off-by - something which you usually don't have for PRs
> but which was for some reason required for netdev.
>
> > and we usually receive PR for
> > batman patches.
>
> This was just an attempt to get syzbot happy again (sooner). Besides my direct
> patch submission, we have different options:
>
> * wait for Simon's PR
> * let Eric Dumazet integrate his (earlier posted) patch from
>   https://lore.kernel.org/r/CANn89iJTHf-sJCqcyrFJiLMLBOBgtN0+KrfPSuW0mhOzLS08Rw@mail.gmail.com
>   This change was also published earlier
>
> > Also I do not see credits given to syzbot  team ?
>
> Correct. Is there a proper way when the reports received were actually about
> different problems (just the bisecting went the wrong way due to the
> batman-adv bug)?
>
> For example, I saw the problem with bisecting in:
>
> * https://syzkaller.appspot.com/bug?extid=ff3aa851d46ab82953a3
>   Reported-by: syzbot+ff3aa851d46ab82953a3@syzkaller.appspotmail.com
> * https://syzkaller.appspot.com/bug?extid=4036165fc595a74b09b2
>   Reported-by: syzbot+4036165fc595a74b09b2@syzkaller.appspotmail.com
> * https://syzkaller.appspot.com/bug?extid=c35d73ce910d86c0026e
>   Reported-by: syzbot+c35d73ce910d86c0026e@syzkaller.appspotmail.com
> * https://syzkaller.appspot.com/bug?extid=48c14f61594bdfadb086
>   Reported-by: syzbot+48c14f61594bdfadb086@syzkaller.appspotmail.com
>
> So, a lot of different problems which unfortunately all ended up in bisecting
> to the batman-adv problem.
>
> On Thursday, 10 April 2025 13:20:21 CEST Eric Dumazet wrote:
> > https://lkml.org/lkml/2025/4/8/1988
>
> You also posted your patch here. Feel free to directly add it. And sorry for
> adding the problem in the first placed - just tried to make Antonio happy (and
> then created a big mess for everyone else)

No worries, could you post the fix today with some of the
Reported-by: you are aware of ?

You can add a 'Suggested-by: Eric Dumazet <edumazet@google.com>'

Thanks !

Re: [PATCH net v3] batman-adv: Fix double-hold of meshif when getting enabled

[Sven Eckelmann]

[-- Attachment #1: Type: text/plain, Size: 356 bytes --]

On Thursday, 10 April 2025 15:31:47 CEST Eric Dumazet wrote:
> No worries, could you post the fix today with some of the
> Reported-by: you are aware of ?
> 
> You can add a 'Suggested-by: Eric Dumazet <edumazet@google.com>'

It was queued in the batman-adv repo earlier today. But I am guessing that 
you're request is more important.

Kind regards,
	Sven

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 228 bytes --]