Ethernet Bridge development
 help / color / mirror / Atom feed
From: Taehee Yoo <ap420073@gmail.com>
To: Heiner Kallweit <hkallweit1@gmail.com>,
	davem@davemloft.net, kuba@kernel.org, dsahern@kernel.org,
	yoshfuji@linux-ipv6.org, netdev@vger.kernel.org,
	j.vosburgh@gmail.com, vfalico@gmail.com, andy@greyhouse.net,
	roopa@nvidia.com, nikolay@nvidia.com, ast@kernel.org,
	andriin@fb.com, daniel@iogearbox.net, weiwan@google.com,
	cong.wang@bytedance.com, bjorn@kernel.org,
	herbert@gondor.apana.org.au, bridge@lists.linux-foundation.org
Subject: Re: [Bridge] [PATCH net v2 1/2] net: core: make bond_get_lowest_level_rcu() generic
Date: Tue, 27 Apr 2021 00:24:31 +0900	[thread overview]
Message-ID: <a43a957a-0fc8-cfff-f04a-cf0bc1ae612b@gmail.com> (raw)
In-Reply-To: <d7cf5368-21a4-a551-169a-00a4cb2b3a0d@gmail.com>

On 4/26/21 3:03 AM, Heiner Kallweit wrote:

Hi Heiner,
Thank you for the review!

 > On 25.04.2021 17:57, Taehee Yoo wrote:
 >> The purpose of bond_get_lowest_level_rcu() is to get nested_level under
 >> RCU. Because dev->nested_level is protected by RTNL, so it shouldn't be
 >> used without RTNL. But bonding module needs this value under RCU without
 >> RTNL.
 >> So, this function was added.
 >>
 >> But, there is another module, which needs this function.
 >> So, make this function generic.
 >> the new name is netdev_get_nest_level_rcu().
 >>
 >> Signed-off-by: Taehee Yoo <ap420073@gmail.com>
 >> ---
 >>
 >> v2:
 >>   - No change
 >>
 >>   drivers/net/bonding/bond_main.c | 45 +--------------------------------
 >>   include/linux/netdevice.h       |  1 +
 >>   net/core/dev.c                  | 44 ++++++++++++++++++++++++++++++++
 >>   3 files changed, 46 insertions(+), 44 deletions(-)
 >>
 >> diff --git a/drivers/net/bonding/bond_main.c 
b/drivers/net/bonding/bond_main.c
 >> index 83ef62db6285..a9feb039ffa6 100644
 >> --- a/drivers/net/bonding/bond_main.c
 >> +++ b/drivers/net/bonding/bond_main.c
 >> @@ -3754,47 +3754,6 @@ static void bond_fold_stats(struct 
rtnl_link_stats64 *_res,
 >>   	}
 >>   }
 >>
 >> -#ifdef CONFIG_LOCKDEP
 >> -static int bond_get_lowest_level_rcu(struct net_device *dev)
 >> -{
 >> -	struct net_device *ldev, *next, *now, *dev_stack[MAX_NEST_DEV + 1];
 >> -	struct list_head *niter, *iter, *iter_stack[MAX_NEST_DEV + 1];
 >> -	int cur = 0, max = 0;
 >> -
 >> -	now = dev;
 >> -	iter = &dev->adj_list.lower;
 >> -
 >> -	while (1) {
 >> -		next = NULL;
 >> -		while (1) {
 >> -			ldev = netdev_next_lower_dev_rcu(now, &iter);
 >> -			if (!ldev)
 >> -				break;
 >> -
 >> -			next = ldev;
 >> -			niter = &ldev->adj_list.lower;
 >> -			dev_stack[cur] = now;
 >> -			iter_stack[cur++] = iter;
 >> -			if (max <= cur)
 >> -				max = cur;
 >> -			break;
 >> -		}
 >> -
 >> -		if (!next) {
 >> -			if (!cur)
 >> -				return max;
 >> -			next = dev_stack[--cur];
 >> -			niter = iter_stack[cur];
 >> -		}
 >> -
 >> -		now = next;
 >> -		iter = niter;
 >> -	}
 >> -
 >> -	return max;
 >> -}
 >> -#endif
 >> -
 >>   static void bond_get_stats(struct net_device *bond_dev,
 >>   			   struct rtnl_link_stats64 *stats)
 >>   {
 >> @@ -3806,9 +3765,7 @@ static void bond_get_stats(struct net_device 
*bond_dev,
 >>
 >>
 >>   	rcu_read_lock();
 >> -#ifdef CONFIG_LOCKDEP
 >> -	nest_level = bond_get_lowest_level_rcu(bond_dev);
 >> -#endif
 >> +	nest_level = netdev_get_nest_level_rcu(bond_dev);
 >>
 >>   	spin_lock_nested(&bond->stats_lock, nest_level);
 >>   	memcpy(stats, &bond->bond_stats, sizeof(*stats));
 >> diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
 >> index 87a5d186faff..507c06bf5dba 100644
 >> --- a/include/linux/netdevice.h
 >> +++ b/include/linux/netdevice.h
 >> @@ -4699,6 +4699,7 @@ int netdev_walk_all_lower_dev(struct 
net_device *dev,
 >>   			      int (*fn)(struct net_device *lower_dev,
 >>   					struct netdev_nested_priv *priv),
 >>   			      struct netdev_nested_priv *priv);
 >> +int netdev_get_nest_level_rcu(struct net_device *dev);
 >>   int netdev_walk_all_lower_dev_rcu(struct net_device *dev,
 >>   				  int (*fn)(struct net_device *lower_dev,
 >>   					    struct netdev_nested_priv *priv),
 >> diff --git a/net/core/dev.c b/net/core/dev.c
 >> index 15fe36332fb8..efc2bf88eafd 100644
 >> --- a/net/core/dev.c
 >> +++ b/net/core/dev.c
 >> @@ -7709,6 +7709,50 @@ static int __netdev_update_lower_level(struct 
net_device *dev,
 >>   	return 0;
 >>   }
 >>
 >> +int netdev_get_nest_level_rcu(struct net_device *dev)
 >> +{
 >> +#ifdef CONFIG_LOCKDEP
 >> +	struct net_device *ldev, *next, *now, *dev_stack[MAX_NEST_DEV + 1];
 >> +	struct list_head *niter, *iter, *iter_stack[MAX_NEST_DEV + 1];
 >> +	int cur = 0, max = 0;
 >> +
 >> +	now = dev;
 >> +	iter = &dev->adj_list.lower;
 >> +
 >> +	while (1) {
 >> +		next = NULL;
 >> +		while (1) {
 >> +			ldev = netdev_next_lower_dev_rcu(now, &iter);
 >> +			if (!ldev)
 >> +				break;
 >> +
 >> +			next = ldev;
 >> +			niter = &ldev->adj_list.lower;
 >> +			dev_stack[cur] = now;
 >> +			iter_stack[cur++] = iter;
 >> +			if (max <= cur)
 >> +				max = cur;
 >> +			break;
 >
 > This looks odd. Why a while loop if it's left in the first iteration
 > anyway? The whole loop looks unnecessarily complex. The following
 > should do the same, just in a simpler way (untested!)
 >
 >          while (1) {
 >                  ldev = netdev_next_lower_dev_rcu(now, &iter);
 >                  if (ldev) {
 >                          dev_stack[cur] = now;
 >                          iter_stack[cur++] = iter;
 >                          if (max <= cur)
 >                                  max = cur;
 >                          now = ldev;
 >                          iter = &ldev->adj_list.lower;
 >                  } else {
 >                          if (!cur)
 >                                  break;
 >                          now = dev_stack[--cur];
 >                          iter = iter_stack[cur];
 >                  }
 >          }
 >
 > I know that you just copied the original function.
 > Simplifying the function should be something for a
 > follow-up patch.
 >
 >> +		}
 >> +
 >> +		if (!next) {
 >> +			if (!cur)
 >> +				return max;
 >> +			next = dev_stack[--cur];
 >> +			niter = iter_stack[cur];
 >> +		}
 >> +
 >> +		now = next;
 >> +		iter = niter;
 >> +	}
 >> +
 >> +	return max;
 >> +#else
 >> +	return 0;
 >> +#endif
 >> +}
 >> +EXPORT_SYMBOL_GPL(netdev_get_nest_level_rcu);
 >> +
 >>   int netdev_walk_all_lower_dev_rcu(struct net_device *dev,
 >>   				  int (*fn)(struct net_device *dev,
 >>   					    struct netdev_nested_priv *priv),
 >>
 >

I think you're right.
Your logic is more simple and stable.
If I send a v3 patch, I will use your logic after some tests.

Thanks a lot!
Taehee

  reply	other threads:[~2021-04-26 15:24 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-04-25 15:57 [Bridge] [PATCH net v2 0/2] net: fix lockdep false positive splat Taehee Yoo
2021-04-25 15:57 ` [Bridge] [PATCH net v2 1/2] net: core: make bond_get_lowest_level_rcu() generic Taehee Yoo
2021-04-25 18:03   ` Heiner Kallweit
2021-04-26 15:24     ` Taehee Yoo [this message]
2021-04-25 15:57 ` [Bridge] [PATCH net 2/2] net: bridge: fix lockdep multicast_lock false positive splat Taehee Yoo
2021-04-25 16:45   ` Nikolay Aleksandrov
2021-04-26 12:48     ` Herbert Xu
2021-04-26 13:15       ` Nikolay Aleksandrov
2021-04-26 15:17         ` Taehee Yoo
2021-04-26 15:39           ` Nikolay Aleksandrov

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=a43a957a-0fc8-cfff-f04a-cf0bc1ae612b@gmail.com \
    --to=ap420073@gmail.com \
    --cc=andriin@fb.com \
    --cc=andy@greyhouse.net \
    --cc=ast@kernel.org \
    --cc=bjorn@kernel.org \
    --cc=bridge@lists.linux-foundation.org \
    --cc=cong.wang@bytedance.com \
    --cc=daniel@iogearbox.net \
    --cc=davem@davemloft.net \
    --cc=dsahern@kernel.org \
    --cc=herbert@gondor.apana.org.au \
    --cc=hkallweit1@gmail.com \
    --cc=j.vosburgh@gmail.com \
    --cc=kuba@kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=nikolay@nvidia.com \
    --cc=roopa@nvidia.com \
    --cc=vfalico@gmail.com \
    --cc=weiwan@google.com \
    --cc=yoshfuji@linux-ipv6.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox