Re: [Xen-devel] [PATCH] net: xen-netbank: hash.c: Use built-in RCU list checking

[Madhuparna Bhowmik <madhuparnabhowmik04@xxxxxxxxx>]

On Wed, Jan 15, 2020 at 7:26 PM Wei Liu <wei.liu@xxxxxxxxxx> wrote:

Thanks for the patch.

There is a typo in the subject line. It should say xen-netback, not
xen-netbank.

Hi,

I am sorry about this, I will send this patch again.

 

On Wed, Jan 15, 2020 at 06:11:28PM +0530, madhuparnabhowmik04@xxxxxxxxx wrote:
> From: Madhuparna Bhowmik <madhuparnabhowmik04@xxxxxxxxx>
>
> list_for_each_entry_rcu has built-in RCU and lock checking.
> Pass cond argument to list_for_each_entry_rcu.
>
> Signed-off-by: Madhuparna Bhowmik <madhuparnabhowmik04@xxxxxxxxx>
> ---
>  drivers/net/xen-netback/hash.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/net/xen-netback/hash.c b/drivers/net/xen-netback/hash.c
> index 10d580c3dea3..30709bc9d170 100644
> --- a/drivers/net/xen-netback/hash.c
> +++ b/drivers/net/xen-netback/hash.c
> @@ -51,7 +51,8 @@ static void xenvif_add_hash(struct xenvif *vif, const u8 *tag,
> 
>       found = false;
>       oldest = NULL;
> -     list_for_each_entry_rcu(entry, &vif->hash.cache.list, link) {
> +     list_for_each_entry_rcu(entry, &vif->hash.cache.list, link,
> +                                                     lockdep_is_held(&vif->hash.cache.lock)) {

There are probably too many tabs here. Indentation looks wrong.

I will correct this when I resend this patch.

 

The surrounding code makes it pretty clear that the lock is already held
by the time list_for_each_entry_rcu is called, yet the checking involved
in lockdep_is_held is not trivial, so I'm afraid I don't consider this a
strict improvement over the existing code.

Actually,  we want to make CONFIG_PROVE_LIST_RCU enabled by default.

And if the cond argument is not passed when the usage of list_for_each_entry_rcu()

is outside of rcu_read_lock(), it will lead to a false positive.

Therefore, I think this patch is required.

Let me know if you have any objections.

Thank you,

Madhuparna

 

If there is something I misunderstood, let me know.

Wei.

>               /* Make sure we don't add duplicate entries */
>               if (entry->len == len &&
>                   memcmp(entry->tag, tag, len) == 0)
> @@ -102,7 +103,8 @@ static void xenvif_flush_hash(struct xenvif *vif)
> 
>       spin_lock_irqsave(&vif->hash.cache.lock, flags);
> 
> -     list_for_each_entry_rcu(entry, &vif->hash.cache.list, link) {
> +     list_for_each_entry_rcu(entry, &vif->hash.cache.list, link,
> +                                                     lockdep_is_held(&vif->hash.cache.lock)) {
>               list_del_rcu(&entry->link);
>               vif->hash.cache.count--;
>               kfree_rcu(entry, rcu);
> --
> 2.17.1
>

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel