* Re: [PATCH 2/2] net: mvneta: improve suspend/resume
From: Jisheng Zhang @ 2018-03-30 9:15 UTC (permalink / raw)
To: Thomas Petazzoni; +Cc: David Miller, netdev, linux-arm-kernel, linux-kernel
In-Reply-To: <20180329135432.7da1299b@windsurf>
On Thu, 29 Mar 2018 13:54:32 +0200 Thomas Petazzoni wrote:
> Hello Jisheng,
Hi Thomas,
>
> On Thu, 29 Mar 2018 18:15:36 +0800, Jisheng Zhang wrote:
> > Current suspend/resume implementation reuses the mvneta_open() and
> > mvneta_close(), but it could be optimized to take only necessary
> > actions during suspend/resume.
> >
> > One obvious problem of current implementation is: after hundreds of
> > system suspend/resume cycles, the resume of mvneta could fail due to
> > fragmented dma coherent memory. After this patch, the non-necessary
> > memory alloc/free is optimized out.
>
> Indeed, this needs to be fixed, you're totally right.
>
> > Signed-off-by: Jisheng Zhang <Jisheng.Zhang@synaptics.com>
> > ---
> > drivers/net/ethernet/marvell/mvneta.c | 76 ++++++++++++++++++++++++++++++-----
> > 1 file changed, 66 insertions(+), 10 deletions(-)
> >
> > diff --git a/drivers/net/ethernet/marvell/mvneta.c b/drivers/net/ethernet/marvell/mvneta.c
> > index 4ec69bbd1eb4..1870f1dd7093 100644
> > --- a/drivers/net/ethernet/marvell/mvneta.c
> > +++ b/drivers/net/ethernet/marvell/mvneta.c
> > @@ -4575,14 +4575,46 @@ static int mvneta_remove(struct platform_device *pdev)
> > #ifdef CONFIG_PM_SLEEP
> > static int mvneta_suspend(struct device *device)
> > {
> > + int queue;
> > struct net_device *dev = dev_get_drvdata(device);
> > struct mvneta_port *pp = netdev_priv(dev);
> >
> > - rtnl_lock();
> > - if (netif_running(dev))
> > - mvneta_stop(dev);
> > - rtnl_unlock();
> > + if (!netif_running(dev))
> > + return 0;
>
> This is changing the behavior I believe. The current code is:
>
> rtnl_lock();
> if (netif_running(dev))
> mvneta_stop(dev);
> rtnl_unlock();
> netif_device_detach(dev);
> clk_disable_unprepare(pp->clk_bus);
> clk_disable_unprepare(pp->clk);
> return 0;
>
> So, when netif_running(dev) is false, we're indeed not calling
> mvneta_stop(), but we're still doing netif_device_detach(), and
> disabling the clocks. With your change, we're no longer doing these
> steps.
Indeed, will try to keep the behavior in v2
>
> > +
> > netif_device_detach(dev);
> > +
> > + mvneta_stop_dev(pp);
> > +
> > + if (!pp->neta_armada3700) {
> > + spin_lock(&pp->lock);
> > + pp->is_stopped = true;
> > + spin_unlock(&pp->lock);
>
> Real question: is it OK to set pp->is_stopped *after* calling
> mvneta_stop_dev(), while it was set before calling mvneta_stop_dev() in
> the current code ?
oops, you are right. Fixed in v2
>
> > +
> > + cpuhp_state_remove_instance_nocalls(online_hpstate,
> > + &pp->node_online);
> > + cpuhp_state_remove_instance_nocalls(CPUHP_NET_MVNETA_DEAD,
> > + &pp->node_dead);
>
> Do we need to remove/add those CPU notifiers when suspending/resuming ?
Take mvneta_cpu_online() as an example, if we don't remove it during
suspend, when system is resume back, it will touch mac when secondary
cpu is ON, but at this point the mvneta isn't resumed, this is not safe.
>
> > + }
> > +
> > + for (queue = 0; queue < rxq_number; queue++) {
> > + struct mvneta_rx_queue *rxq = &pp->rxqs[queue];
> > +
> > + mvneta_rxq_drop_pkts(pp, rxq);
> > + }
>
> Wouldn't it make sense to have
> mvneta_rxq_sw_deinit/mvneta_rxq_hw_deinit(), like you did for the
> initialization ?
For rxq deinit, we'd like to drop rx pkts, this is both HW and SW operation.
So we reuse mvneta_rxq_drop_pkts() here.
>
> > +
> > + for (queue = 0; queue < txq_number; queue++) {
> > + struct mvneta_tx_queue *txq = &pp->txqs[queue];
> > +
> > + /* Set minimum bandwidth for disabled TXQs */
> > + mvreg_write(pp, MVETH_TXQ_TOKEN_CFG_REG(txq->id), 0);
> > + mvreg_write(pp, MVETH_TXQ_TOKEN_COUNT_REG(txq->id), 0);
> > +
> > + /* Set Tx descriptors queue starting address and size */
> > + mvreg_write(pp, MVNETA_TXQ_BASE_ADDR_REG(txq->id), 0);
> > + mvreg_write(pp, MVNETA_TXQ_SIZE_REG(txq->id), 0);
> > + }
>
> Same comment here: a mvneta_txq_sw_deinit()/mvneta_txq_hw_deinit()
> would be good, and would avoid duplicating this logic.
yep, will do in v2.
Thanks a lot for the kind review.
^ permalink raw reply
* [PATCH net-next] net: Revert net_rwsem
From: Kirill Tkhai @ 2018-03-30 9:23 UTC (permalink / raw)
To: davem, ktkhai, netdev
This reverts:
152f253152cc net: Remove rtnl_lock() in nf_ct_iterate_destroy()
ec9c780925c5 ovs: Remove rtnl_lock() from ovs_exit_net()
350311aab4c0 security: Remove rtnl_lock() in selinux_xfrm_notify_policyload()
10256debb918 net: Don't take rtnl_lock() in wireless_nlevent_flush()
f0b07bb151b0 net: Introduce net_rwsem to protect net_namespace_list
There are missed, that down_read() can't be taken recursive.
This is because of rw_semaphore design, which prevents it
to be occupied by only readers forever.
So, we can't take it in register_netdevice_notifier(), as it's also
taken in wext_netdev_notifier_call()->wireless_nlevent_flush().
The solution is to protect net_namespace_list modifications
in register_netdevice_notifier() via pernet_ops_rwsem, as it's
made in:
https://patchwork.ozlabs.org/project/netdev/list/?series=36495
Then, we won't have to take net_rwsem in call_netevent_notifiers().
But since the patchset is not in kernel, let's just revert net_rwsem
for now, and I'll resubmit it later (after the above patchset).
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
---
drivers/infiniband/core/roce_gid_mgmt.c | 2 --
include/linux/rtnetlink.h | 1 -
include/net/net_namespace.h | 1 -
net/core/dev.c | 5 -----
net/core/fib_notifier.c | 2 --
net/core/net_namespace.c | 18 +++++-------------
net/core/rtnetlink.c | 5 -----
net/netfilter/nf_conntrack_core.c | 4 ++--
net/openvswitch/datapath.c | 4 ++--
net/wireless/wext-core.c | 6 ++++--
security/selinux/include/xfrm.h | 4 ++--
11 files changed, 15 insertions(+), 37 deletions(-)
diff --git a/drivers/infiniband/core/roce_gid_mgmt.c b/drivers/infiniband/core/roce_gid_mgmt.c
index cc2966380c0c..5a52ec77940a 100644
--- a/drivers/infiniband/core/roce_gid_mgmt.c
+++ b/drivers/infiniband/core/roce_gid_mgmt.c
@@ -403,12 +403,10 @@ static void enum_all_gids_of_dev_cb(struct ib_device *ib_dev,
* our feet
*/
rtnl_lock();
- down_read(&net_rwsem);
for_each_net(net)
for_each_netdev(net, ndev)
if (is_eth_port_of_netdev(ib_dev, port, rdma_ndev, ndev))
add_netdev_ips(ib_dev, port, rdma_ndev, ndev);
- up_read(&net_rwsem);
rtnl_unlock();
}
diff --git a/include/linux/rtnetlink.h b/include/linux/rtnetlink.h
index 5225832bd6ff..c7d1e4689325 100644
--- a/include/linux/rtnetlink.h
+++ b/include/linux/rtnetlink.h
@@ -37,7 +37,6 @@ extern int rtnl_lock_killable(void);
extern wait_queue_head_t netdev_unregistering_wq;
extern struct rw_semaphore pernet_ops_rwsem;
-extern struct rw_semaphore net_rwsem;
#ifdef CONFIG_PROVE_LOCKING
extern bool lockdep_rtnl_is_held(void);
diff --git a/include/net/net_namespace.h b/include/net/net_namespace.h
index 47e35cce3b64..1ab4f920f109 100644
--- a/include/net/net_namespace.h
+++ b/include/net/net_namespace.h
@@ -291,7 +291,6 @@ static inline struct net *read_pnet(const possible_net_t *pnet)
#endif
}
-/* Protected by net_rwsem */
#define for_each_net(VAR) \
list_for_each_entry(VAR, &net_namespace_list, list)
diff --git a/net/core/dev.c b/net/core/dev.c
index eca5458b2753..e13807b5c84d 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -1629,7 +1629,6 @@ int register_netdevice_notifier(struct notifier_block *nb)
goto unlock;
if (dev_boot_phase)
goto unlock;
- down_read(&net_rwsem);
for_each_net(net) {
for_each_netdev(net, dev) {
err = call_netdevice_notifier(nb, NETDEV_REGISTER, dev);
@@ -1643,7 +1642,6 @@ int register_netdevice_notifier(struct notifier_block *nb)
call_netdevice_notifier(nb, NETDEV_UP, dev);
}
}
- up_read(&net_rwsem);
unlock:
rtnl_unlock();
@@ -1666,7 +1664,6 @@ int register_netdevice_notifier(struct notifier_block *nb)
}
outroll:
- up_read(&net_rwsem);
raw_notifier_chain_unregister(&netdev_chain, nb);
goto unlock;
}
@@ -1697,7 +1694,6 @@ int unregister_netdevice_notifier(struct notifier_block *nb)
if (err)
goto unlock;
- down_read(&net_rwsem);
for_each_net(net) {
for_each_netdev(net, dev) {
if (dev->flags & IFF_UP) {
@@ -1708,7 +1704,6 @@ int unregister_netdevice_notifier(struct notifier_block *nb)
call_netdevice_notifier(nb, NETDEV_UNREGISTER, dev);
}
}
- up_read(&net_rwsem);
unlock:
rtnl_unlock();
return err;
diff --git a/net/core/fib_notifier.c b/net/core/fib_notifier.c
index 13a40b831d6d..b793b523aba3 100644
--- a/net/core/fib_notifier.c
+++ b/net/core/fib_notifier.c
@@ -39,7 +39,6 @@ static unsigned int fib_seq_sum(void)
struct net *net;
rtnl_lock();
- down_read(&net_rwsem);
for_each_net(net) {
rcu_read_lock();
list_for_each_entry_rcu(ops, &net->fib_notifier_ops, list) {
@@ -50,7 +49,6 @@ static unsigned int fib_seq_sum(void)
}
rcu_read_unlock();
}
- up_read(&net_rwsem);
rtnl_unlock();
return fib_seq;
diff --git a/net/core/net_namespace.c b/net/core/net_namespace.c
index 7fdf321d4997..b5796d17a302 100644
--- a/net/core/net_namespace.c
+++ b/net/core/net_namespace.c
@@ -33,10 +33,6 @@ static struct list_head *first_device = &pernet_list;
LIST_HEAD(net_namespace_list);
EXPORT_SYMBOL_GPL(net_namespace_list);
-/* Protects net_namespace_list. Nests iside rtnl_lock() */
-DECLARE_RWSEM(net_rwsem);
-EXPORT_SYMBOL_GPL(net_rwsem);
-
struct net init_net = {
.count = REFCOUNT_INIT(1),
.dev_base_head = LIST_HEAD_INIT(init_net.dev_base_head),
@@ -313,9 +309,9 @@ static __net_init int setup_net(struct net *net, struct user_namespace *user_ns)
if (error < 0)
goto out_undo;
}
- down_write(&net_rwsem);
+ rtnl_lock();
list_add_tail_rcu(&net->list, &net_namespace_list);
- up_write(&net_rwsem);
+ rtnl_unlock();
out:
return error;
@@ -454,7 +450,7 @@ static void unhash_nsid(struct net *net, struct net *last)
* and this work is the only process, that may delete
* a net from net_namespace_list. So, when the below
* is executing, the list may only grow. Thus, we do not
- * use for_each_net_rcu() or net_rwsem.
+ * use for_each_net_rcu() or rtnl_lock().
*/
for_each_net(tmp) {
int id;
@@ -489,7 +485,7 @@ static void cleanup_net(struct work_struct *work)
down_read(&pernet_ops_rwsem);
/* Don't let anyone else find us. */
- down_write(&net_rwsem);
+ rtnl_lock();
llist_for_each_entry(net, net_kill_list, cleanup_list)
list_del_rcu(&net->list);
/* Cache last net. After we unlock rtnl, no one new net
@@ -503,7 +499,7 @@ static void cleanup_net(struct work_struct *work)
* useless anyway, as netns_ids are destroyed there.
*/
last = list_last_entry(&net_namespace_list, struct net, list);
- up_write(&net_rwsem);
+ rtnl_unlock();
llist_for_each_entry(net, net_kill_list, cleanup_list) {
unhash_nsid(net, last);
@@ -904,9 +900,6 @@ static int __register_pernet_operations(struct list_head *list,
list_add_tail(&ops->list, list);
if (ops->init || (ops->id && ops->size)) {
- /* We held write locked pernet_ops_rwsem, and parallel
- * setup_net() and cleanup_net() are not possible.
- */
for_each_net(net) {
error = ops_init(ops, net);
if (error)
@@ -930,7 +923,6 @@ static void __unregister_pernet_operations(struct pernet_operations *ops)
LIST_HEAD(net_exit_list);
list_del(&ops->list);
- /* See comment in __register_pernet_operations() */
for_each_net(net)
list_add_tail(&net->exit_list, &net_exit_list);
ops_exit_list(ops, &net_exit_list);
diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index e86b28482ca7..2d3949789cef 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -418,11 +418,9 @@ void __rtnl_link_unregister(struct rtnl_link_ops *ops)
{
struct net *net;
- down_read(&net_rwsem);
for_each_net(net) {
__rtnl_kill_links(net, ops);
}
- up_read(&net_rwsem);
list_del(&ops->list);
}
EXPORT_SYMBOL_GPL(__rtnl_link_unregister);
@@ -440,9 +438,6 @@ static void rtnl_lock_unregistering_all(void)
for (;;) {
unregistering = false;
rtnl_lock();
- /* We held write locked pernet_ops_rwsem, and parallel
- * setup_net() and cleanup_net() are not possible.
- */
for_each_net(net) {
if (net->dev_unreg_count > 0) {
unregistering = true;
diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c
index 41ff04ee2554..705198de671d 100644
--- a/net/netfilter/nf_conntrack_core.c
+++ b/net/netfilter/nf_conntrack_core.c
@@ -1763,14 +1763,14 @@ nf_ct_iterate_destroy(int (*iter)(struct nf_conn *i, void *data), void *data)
{
struct net *net;
- down_read(&net_rwsem);
+ rtnl_lock();
for_each_net(net) {
if (atomic_read(&net->ct.count) == 0)
continue;
__nf_ct_unconfirmed_destroy(net);
nf_queue_nf_hook_drop(net);
}
- up_read(&net_rwsem);
+ rtnl_unlock();
/* Need to wait for netns cleanup worker to finish, if its
* running -- it might have deleted a net namespace from
diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index 015e24e08909..ef38e5aecd28 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -2363,10 +2363,10 @@ static void __net_exit ovs_exit_net(struct net *dnet)
list_for_each_entry_safe(dp, dp_next, &ovs_net->dps, list_node)
__dp_destroy(dp);
- down_read(&net_rwsem);
+ rtnl_lock();
for_each_net(net)
list_vports_from_net(net, dnet, &head);
- up_read(&net_rwsem);
+ rtnl_unlock();
/* Detach all vports from given namespace. */
list_for_each_entry_safe(vport, vport_next, &head, detach_list) {
diff --git a/net/wireless/wext-core.c b/net/wireless/wext-core.c
index 5e677dac2a0c..9efbfc753347 100644
--- a/net/wireless/wext-core.c
+++ b/net/wireless/wext-core.c
@@ -347,13 +347,13 @@ void wireless_nlevent_flush(void)
struct sk_buff *skb;
struct net *net;
- down_read(&net_rwsem);
+ ASSERT_RTNL();
+
for_each_net(net) {
while ((skb = skb_dequeue(&net->wext_nlevents)))
rtnl_notify(skb, net, 0, RTNLGRP_LINK, NULL,
GFP_KERNEL);
}
- up_read(&net_rwsem);
}
EXPORT_SYMBOL_GPL(wireless_nlevent_flush);
@@ -410,7 +410,9 @@ subsys_initcall(wireless_nlevent_init);
/* Process events generated by the wireless layer or the driver. */
static void wireless_nlevent_process(struct work_struct *work)
{
+ rtnl_lock();
wireless_nlevent_flush();
+ rtnl_unlock();
}
static DECLARE_WORK(wireless_nlevent_work, wireless_nlevent_process);
diff --git a/security/selinux/include/xfrm.h b/security/selinux/include/xfrm.h
index a0b465316292..1f173a7a4daa 100644
--- a/security/selinux/include/xfrm.h
+++ b/security/selinux/include/xfrm.h
@@ -47,10 +47,10 @@ static inline void selinux_xfrm_notify_policyload(void)
{
struct net *net;
- down_read(&net_rwsem);
+ rtnl_lock();
for_each_net(net)
rt_genid_bump_all(net);
- up_read(&net_rwsem);
+ rtnl_unlock();
}
#else
static inline int selinux_xfrm_enabled(void)
^ permalink raw reply related
* Re: [PATCH net-next] dt-bindings: net: renesas-ravb: Add support for r8a77470 SoC
From: Geert Uytterhoeven @ 2018-03-30 9:26 UTC (permalink / raw)
To: Biju Das
Cc: David S . Miller, Rob Herring, Mark Rutland, Russell King,
Geert Uytterhoeven, Sergei Shtylyov, Simon Horman, Magnus Damm,
Chris Paterson, Fabrizio Castro,
open list:OPEN FIRMWARE AND FLATTENED DEVICE TREE BINDINGS,
Linux-Renesas, Linux ARM, netdev
In-Reply-To: <1522317775-35671-1-git-send-email-biju.das@bp.renesas.com>
On Thu, Mar 29, 2018 at 12:02 PM, Biju Das <biju.das@bp.renesas.com> wrote:
> Add a new compatible string for the RZ/G1C (R8A77470) SoC.
>
> Signed-off-by: Biju Das <biju.das@bp.renesas.com>
> Reviewed-by: Fabrizio Castro <fabrizio.castro@bp.renesas.com>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
^ permalink raw reply
* Re: [PATCH net-next 2/3] netfilter: Rework xt_TEE netdevice notifier
From: Pablo Neira Ayuso @ 2018-03-30 9:47 UTC (permalink / raw)
To: Kirill Tkhai
Cc: davem, steffen.klassert, herbert, kadlec, fw, daniel,
jakub.kicinski, ast, brouer, linux, john.fastabend, dsahern,
netdev, netfilter-devel, coreteam
In-Reply-To: <152233221526.1654.14062155454284081692.stgit@localhost.localdomain>
On Thu, Mar 29, 2018 at 05:03:35PM +0300, Kirill Tkhai wrote:
> Register netdevice notifier for every iptable entry
> is not good, since this breaks modularity, and
> the hidden synchronization is based on rtnl_lock().
>
> This patch reworks the synchronization via new lock,
> while the rest of logic remains as it was before.
> This is required for the next patch.
>
> Tested via:
>
> while :; do
> unshare -n iptables -t mangle -A OUTPUT -j TEE --gateway 1.1.1.2 --oif lo;
> done
>
> Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Acked-by: Pablo Neira Ayuso <pablo@netfilter.org>
^ permalink raw reply
* Re: [PATCH 2/2] net: mvneta: improve suspend/resume
From: Thomas Petazzoni @ 2018-03-30 9:49 UTC (permalink / raw)
To: Jisheng Zhang; +Cc: David Miller, netdev, linux-arm-kernel, linux-kernel
In-Reply-To: <20180330171547.5f96bbff@xhacker.debian>
Hello,
On Fri, 30 Mar 2018 17:15:47 +0800, Jisheng Zhang wrote:
> > > + cpuhp_state_remove_instance_nocalls(online_hpstate,
> > > + &pp->node_online);
> > > + cpuhp_state_remove_instance_nocalls(CPUHP_NET_MVNETA_DEAD,
> > > + &pp->node_dead);
> >
> > Do we need to remove/add those CPU notifiers when suspending/resuming ?
>
> Take mvneta_cpu_online() as an example, if we don't remove it during
> suspend, when system is resume back, it will touch mac when secondary
> cpu is ON, but at this point the mvneta isn't resumed, this is not safe.
Hm. I'm still a bit confused by this new CPU hotplug API. I understand
the issue you have and indeed unregistering the CPU hotplug callbacks
is a way to solve the problem, but I find it weird that we have to do
this.
Anyway, it's OK to do it, because it's anyway what was done so far. It
is just annoying that there is a duplication of the logic between
mvneta_suspend() and mvneta_stop() on one side, and duplication between
mvneta_resume() and mvnete_start() on the other side.
> > > + for (queue = 0; queue < rxq_number; queue++) {
> > > + struct mvneta_rx_queue *rxq = &pp->rxqs[queue];
> > > +
> > > + mvneta_rxq_drop_pkts(pp, rxq);
> > > + }
> >
> > Wouldn't it make sense to have
> > mvneta_rxq_sw_deinit/mvneta_rxq_hw_deinit(), like you did for the
> > initialization ?
>
> For rxq deinit, we'd like to drop rx pkts, this is both HW and SW operation.
> So we reuse mvneta_rxq_drop_pkts() here.
Hum, OK, indeed. It would have been nicer to have something symmetric,
with the hw/sw parts split in a similar way for the init and deinit of
both txqs and rxqs, but I agree that dropping the RX packets before
going into suspend involves both HW and SW operations.
Thanks!
Thomas Petazzoni
--
Thomas Petazzoni, CTO, Bootlin (formerly Free Electrons)
Embedded Linux and Kernel engineering
https://bootlin.com
^ permalink raw reply
* Re: [PATCH] net: netfilter: Merge assignment with return
From: Pablo Neira Ayuso @ 2018-03-30 9:51 UTC (permalink / raw)
To: Arushi Singhal
Cc: outreachy-kernel, Jozsef Kadlecsik, Florian Westphal,
David S. Miller, netfilter-devel, coreteam, netdev, linux-kernel
In-Reply-To: <20180328190950.GA3151@seema-Inspiron-15-3567>
On Thu, Mar 29, 2018 at 12:39:50AM +0530, Arushi Singhal wrote:
> Merge assignment with return statement to directly return the value.
Applied, thanks Arushi.
^ permalink raw reply
* Re: [PATCH] bridge: netfilter: ebt_stp: Use generic functions for comparisons
From: Pablo Neira Ayuso @ 2018-03-30 9:52 UTC (permalink / raw)
To: Joe Perches
Cc: Jozsef Kadlecsik, Florian Westphal, Stephen Hemminger,
David S. Miller, netfilter-devel, coreteam, bridge, netdev,
linux-kernel
In-Reply-To: <c22b7398816c4533a65ff57aaa44629729fc7254.1521630169.git.joe@perches.com>
On Wed, Mar 21, 2018 at 04:03:22AM -0700, Joe Perches wrote:
> Instead of unnecessary const declarations, use the generic functions to
> save a little object space.
>
> $ size net/bridge/netfilter/ebt_stp.o*
> text data bss dec hex filename
> 1250 144 0 1394 572 net/bridge/netfilter/ebt_stp.o.new
> 1344 144 0 1488 5d0 net/bridge/netfilter/ebt_stp.o.old
Applied, thanks Joe.
^ permalink raw reply
* Re: [PATCH net-next 0/9] devlink: Add support for region access
From: Rahul Lakkireddy @ 2018-03-30 9:51 UTC (permalink / raw)
To: Andrew Lunn
Cc: Alex Vesker, David S. Miller, netdev@vger.kernel.org,
Tariq Toukan, Jiri Pirko
In-Reply-To: <20180329182343.GB12150@lunn.ch>
On Thursday, March 03/29/18, 2018 at 23:53:43 +0530, Andrew Lunn wrote:
> On Thu, Mar 29, 2018 at 07:07:43PM +0300, Alex Vesker wrote:
> > This is a proposal which will allow access to driver defined address
> > regions using devlink. Each device can create its supported address
> > regions and register them. A device which exposes a region will allow
> > access to it using devlink.
>
> Hi Alex
>
> Did you see the work Rahul Lakkireddy has been doing?
>
> https://patchwork.kernel.org/patch/10305935/
>
> It seems like these are similar, or at least overlapping. We probably
> want one solution for both.
>
We're already collecting hardware snapshots when system is live with
ethtool --getdump (which devlink tool is now trying to do).
We are now in the process of adding support to collect hardware
snapshots during kernel panic.
Thanks,
Rahul
^ permalink raw reply
* Re: [PATCH] netfilter: ipset: Use is_zero_ether_addr instead of static and memcmp
From: Pablo Neira Ayuso @ 2018-03-30 9:54 UTC (permalink / raw)
To: Joe Perches
Cc: linux-kernel, Jozsef Kadlecsik, Florian Westphal, David S. Miller,
netfilter-devel, coreteam, netdev
In-Reply-To: <687a4edaed07441681d3bfe3b63aa4feb55b9d61.1521567319.git.joe@perches.com>
On Tue, Mar 20, 2018 at 10:35:47AM -0700, Joe Perches wrote:
> To make the test a bit clearer and to reduce object size a little.
>
> Miscellanea:
>
> o remove now unnecessary static const array
>
> $ size ip_set_hash_mac.o*
> text data bss dec hex filename
> 22822 4619 64 27505 6b71 ip_set_hash_mac.o.allyesconfig.new
> 22932 4683 64 27679 6c1f ip_set_hash_mac.o.allyesconfig.old
> 10443 1040 0 11483 2cdb ip_set_hash_mac.o.defconfig.new
> 10507 1040 0 11547 2d1b ip_set_hash_mac.o.defconfig.old
Applied, thanks Joe.
^ permalink raw reply
* Re: [PATCH net-next] net: Revert net_rwsem
From: Kirill Tkhai @ 2018-03-30 9:57 UTC (permalink / raw)
To: davem, netdev
In-Reply-To: <152240169991.24910.6978707493362813761.stgit@localhost.localdomain>
On 30.03.2018 12:23, Kirill Tkhai wrote:
> This reverts:
>
> 152f253152cc net: Remove rtnl_lock() in nf_ct_iterate_destroy()
> ec9c780925c5 ovs: Remove rtnl_lock() from ovs_exit_net()
> 350311aab4c0 security: Remove rtnl_lock() in selinux_xfrm_notify_policyload()
> 10256debb918 net: Don't take rtnl_lock() in wireless_nlevent_flush()
> f0b07bb151b0 net: Introduce net_rwsem to protect net_namespace_list
>
> There are missed, that down_read() can't be taken recursive.
> This is because of rw_semaphore design, which prevents it
> to be occupied by only readers forever.
>
> So, we can't take it in register_netdevice_notifier(), as it's also
> taken in wext_netdev_notifier_call()->wireless_nlevent_flush().
>
> The solution is to protect net_namespace_list modifications
> in register_netdevice_notifier() via pernet_ops_rwsem, as it's
> made in:
>
> https://patchwork.ozlabs.org/project/netdev/list/?series=36495
>
> Then, we won't have to take net_rwsem in call_netevent_notifiers().
>
> But since the patchset is not in kernel, let's just revert net_rwsem
> for now, and I'll resubmit it later (after the above patchset).
>
> Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Or, if https://patchwork.ozlabs.org/project/netdev/list/?series=36495
is ok enough to go in kernel in a little while, I'll make another fix
(removing down_read(&net_rwsem) from {,un}register_netdevice_notifiers()).
Please, let me know if some other actions are required.
Thanks,
Kirill
> ---
> drivers/infiniband/core/roce_gid_mgmt.c | 2 --
> include/linux/rtnetlink.h | 1 -
> include/net/net_namespace.h | 1 -
> net/core/dev.c | 5 -----
> net/core/fib_notifier.c | 2 --
> net/core/net_namespace.c | 18 +++++-------------
> net/core/rtnetlink.c | 5 -----
> net/netfilter/nf_conntrack_core.c | 4 ++--
> net/openvswitch/datapath.c | 4 ++--
> net/wireless/wext-core.c | 6 ++++--
> security/selinux/include/xfrm.h | 4 ++--
> 11 files changed, 15 insertions(+), 37 deletions(-)
>
> diff --git a/drivers/infiniband/core/roce_gid_mgmt.c b/drivers/infiniband/core/roce_gid_mgmt.c
> index cc2966380c0c..5a52ec77940a 100644
> --- a/drivers/infiniband/core/roce_gid_mgmt.c
> +++ b/drivers/infiniband/core/roce_gid_mgmt.c
> @@ -403,12 +403,10 @@ static void enum_all_gids_of_dev_cb(struct ib_device *ib_dev,
> * our feet
> */
> rtnl_lock();
> - down_read(&net_rwsem);
> for_each_net(net)
> for_each_netdev(net, ndev)
> if (is_eth_port_of_netdev(ib_dev, port, rdma_ndev, ndev))
> add_netdev_ips(ib_dev, port, rdma_ndev, ndev);
> - up_read(&net_rwsem);
> rtnl_unlock();
> }
>
> diff --git a/include/linux/rtnetlink.h b/include/linux/rtnetlink.h
> index 5225832bd6ff..c7d1e4689325 100644
> --- a/include/linux/rtnetlink.h
> +++ b/include/linux/rtnetlink.h
> @@ -37,7 +37,6 @@ extern int rtnl_lock_killable(void);
>
> extern wait_queue_head_t netdev_unregistering_wq;
> extern struct rw_semaphore pernet_ops_rwsem;
> -extern struct rw_semaphore net_rwsem;
>
> #ifdef CONFIG_PROVE_LOCKING
> extern bool lockdep_rtnl_is_held(void);
> diff --git a/include/net/net_namespace.h b/include/net/net_namespace.h
> index 47e35cce3b64..1ab4f920f109 100644
> --- a/include/net/net_namespace.h
> +++ b/include/net/net_namespace.h
> @@ -291,7 +291,6 @@ static inline struct net *read_pnet(const possible_net_t *pnet)
> #endif
> }
>
> -/* Protected by net_rwsem */
> #define for_each_net(VAR) \
> list_for_each_entry(VAR, &net_namespace_list, list)
>
> diff --git a/net/core/dev.c b/net/core/dev.c
> index eca5458b2753..e13807b5c84d 100644
> --- a/net/core/dev.c
> +++ b/net/core/dev.c
> @@ -1629,7 +1629,6 @@ int register_netdevice_notifier(struct notifier_block *nb)
> goto unlock;
> if (dev_boot_phase)
> goto unlock;
> - down_read(&net_rwsem);
> for_each_net(net) {
> for_each_netdev(net, dev) {
> err = call_netdevice_notifier(nb, NETDEV_REGISTER, dev);
> @@ -1643,7 +1642,6 @@ int register_netdevice_notifier(struct notifier_block *nb)
> call_netdevice_notifier(nb, NETDEV_UP, dev);
> }
> }
> - up_read(&net_rwsem);
>
> unlock:
> rtnl_unlock();
> @@ -1666,7 +1664,6 @@ int register_netdevice_notifier(struct notifier_block *nb)
> }
>
> outroll:
> - up_read(&net_rwsem);
> raw_notifier_chain_unregister(&netdev_chain, nb);
> goto unlock;
> }
> @@ -1697,7 +1694,6 @@ int unregister_netdevice_notifier(struct notifier_block *nb)
> if (err)
> goto unlock;
>
> - down_read(&net_rwsem);
> for_each_net(net) {
> for_each_netdev(net, dev) {
> if (dev->flags & IFF_UP) {
> @@ -1708,7 +1704,6 @@ int unregister_netdevice_notifier(struct notifier_block *nb)
> call_netdevice_notifier(nb, NETDEV_UNREGISTER, dev);
> }
> }
> - up_read(&net_rwsem);
> unlock:
> rtnl_unlock();
> return err;
> diff --git a/net/core/fib_notifier.c b/net/core/fib_notifier.c
> index 13a40b831d6d..b793b523aba3 100644
> --- a/net/core/fib_notifier.c
> +++ b/net/core/fib_notifier.c
> @@ -39,7 +39,6 @@ static unsigned int fib_seq_sum(void)
> struct net *net;
>
> rtnl_lock();
> - down_read(&net_rwsem);
> for_each_net(net) {
> rcu_read_lock();
> list_for_each_entry_rcu(ops, &net->fib_notifier_ops, list) {
> @@ -50,7 +49,6 @@ static unsigned int fib_seq_sum(void)
> }
> rcu_read_unlock();
> }
> - up_read(&net_rwsem);
> rtnl_unlock();
>
> return fib_seq;
> diff --git a/net/core/net_namespace.c b/net/core/net_namespace.c
> index 7fdf321d4997..b5796d17a302 100644
> --- a/net/core/net_namespace.c
> +++ b/net/core/net_namespace.c
> @@ -33,10 +33,6 @@ static struct list_head *first_device = &pernet_list;
> LIST_HEAD(net_namespace_list);
> EXPORT_SYMBOL_GPL(net_namespace_list);
>
> -/* Protects net_namespace_list. Nests iside rtnl_lock() */
> -DECLARE_RWSEM(net_rwsem);
> -EXPORT_SYMBOL_GPL(net_rwsem);
> -
> struct net init_net = {
> .count = REFCOUNT_INIT(1),
> .dev_base_head = LIST_HEAD_INIT(init_net.dev_base_head),
> @@ -313,9 +309,9 @@ static __net_init int setup_net(struct net *net, struct user_namespace *user_ns)
> if (error < 0)
> goto out_undo;
> }
> - down_write(&net_rwsem);
> + rtnl_lock();
> list_add_tail_rcu(&net->list, &net_namespace_list);
> - up_write(&net_rwsem);
> + rtnl_unlock();
> out:
> return error;
>
> @@ -454,7 +450,7 @@ static void unhash_nsid(struct net *net, struct net *last)
> * and this work is the only process, that may delete
> * a net from net_namespace_list. So, when the below
> * is executing, the list may only grow. Thus, we do not
> - * use for_each_net_rcu() or net_rwsem.
> + * use for_each_net_rcu() or rtnl_lock().
> */
> for_each_net(tmp) {
> int id;
> @@ -489,7 +485,7 @@ static void cleanup_net(struct work_struct *work)
> down_read(&pernet_ops_rwsem);
>
> /* Don't let anyone else find us. */
> - down_write(&net_rwsem);
> + rtnl_lock();
> llist_for_each_entry(net, net_kill_list, cleanup_list)
> list_del_rcu(&net->list);
> /* Cache last net. After we unlock rtnl, no one new net
> @@ -503,7 +499,7 @@ static void cleanup_net(struct work_struct *work)
> * useless anyway, as netns_ids are destroyed there.
> */
> last = list_last_entry(&net_namespace_list, struct net, list);
> - up_write(&net_rwsem);
> + rtnl_unlock();
>
> llist_for_each_entry(net, net_kill_list, cleanup_list) {
> unhash_nsid(net, last);
> @@ -904,9 +900,6 @@ static int __register_pernet_operations(struct list_head *list,
>
> list_add_tail(&ops->list, list);
> if (ops->init || (ops->id && ops->size)) {
> - /* We held write locked pernet_ops_rwsem, and parallel
> - * setup_net() and cleanup_net() are not possible.
> - */
> for_each_net(net) {
> error = ops_init(ops, net);
> if (error)
> @@ -930,7 +923,6 @@ static void __unregister_pernet_operations(struct pernet_operations *ops)
> LIST_HEAD(net_exit_list);
>
> list_del(&ops->list);
> - /* See comment in __register_pernet_operations() */
> for_each_net(net)
> list_add_tail(&net->exit_list, &net_exit_list);
> ops_exit_list(ops, &net_exit_list);
> diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
> index e86b28482ca7..2d3949789cef 100644
> --- a/net/core/rtnetlink.c
> +++ b/net/core/rtnetlink.c
> @@ -418,11 +418,9 @@ void __rtnl_link_unregister(struct rtnl_link_ops *ops)
> {
> struct net *net;
>
> - down_read(&net_rwsem);
> for_each_net(net) {
> __rtnl_kill_links(net, ops);
> }
> - up_read(&net_rwsem);
> list_del(&ops->list);
> }
> EXPORT_SYMBOL_GPL(__rtnl_link_unregister);
> @@ -440,9 +438,6 @@ static void rtnl_lock_unregistering_all(void)
> for (;;) {
> unregistering = false;
> rtnl_lock();
> - /* We held write locked pernet_ops_rwsem, and parallel
> - * setup_net() and cleanup_net() are not possible.
> - */
> for_each_net(net) {
> if (net->dev_unreg_count > 0) {
> unregistering = true;
> diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c
> index 41ff04ee2554..705198de671d 100644
> --- a/net/netfilter/nf_conntrack_core.c
> +++ b/net/netfilter/nf_conntrack_core.c
> @@ -1763,14 +1763,14 @@ nf_ct_iterate_destroy(int (*iter)(struct nf_conn *i, void *data), void *data)
> {
> struct net *net;
>
> - down_read(&net_rwsem);
> + rtnl_lock();
> for_each_net(net) {
> if (atomic_read(&net->ct.count) == 0)
> continue;
> __nf_ct_unconfirmed_destroy(net);
> nf_queue_nf_hook_drop(net);
> }
> - up_read(&net_rwsem);
> + rtnl_unlock();
>
> /* Need to wait for netns cleanup worker to finish, if its
> * running -- it might have deleted a net namespace from
> diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
> index 015e24e08909..ef38e5aecd28 100644
> --- a/net/openvswitch/datapath.c
> +++ b/net/openvswitch/datapath.c
> @@ -2363,10 +2363,10 @@ static void __net_exit ovs_exit_net(struct net *dnet)
> list_for_each_entry_safe(dp, dp_next, &ovs_net->dps, list_node)
> __dp_destroy(dp);
>
> - down_read(&net_rwsem);
> + rtnl_lock();
> for_each_net(net)
> list_vports_from_net(net, dnet, &head);
> - up_read(&net_rwsem);
> + rtnl_unlock();
>
> /* Detach all vports from given namespace. */
> list_for_each_entry_safe(vport, vport_next, &head, detach_list) {
> diff --git a/net/wireless/wext-core.c b/net/wireless/wext-core.c
> index 5e677dac2a0c..9efbfc753347 100644
> --- a/net/wireless/wext-core.c
> +++ b/net/wireless/wext-core.c
> @@ -347,13 +347,13 @@ void wireless_nlevent_flush(void)
> struct sk_buff *skb;
> struct net *net;
>
> - down_read(&net_rwsem);
> + ASSERT_RTNL();
> +
> for_each_net(net) {
> while ((skb = skb_dequeue(&net->wext_nlevents)))
> rtnl_notify(skb, net, 0, RTNLGRP_LINK, NULL,
> GFP_KERNEL);
> }
> - up_read(&net_rwsem);
> }
> EXPORT_SYMBOL_GPL(wireless_nlevent_flush);
>
> @@ -410,7 +410,9 @@ subsys_initcall(wireless_nlevent_init);
> /* Process events generated by the wireless layer or the driver. */
> static void wireless_nlevent_process(struct work_struct *work)
> {
> + rtnl_lock();
> wireless_nlevent_flush();
> + rtnl_unlock();
> }
>
> static DECLARE_WORK(wireless_nlevent_work, wireless_nlevent_process);
> diff --git a/security/selinux/include/xfrm.h b/security/selinux/include/xfrm.h
> index a0b465316292..1f173a7a4daa 100644
> --- a/security/selinux/include/xfrm.h
> +++ b/security/selinux/include/xfrm.h
> @@ -47,10 +47,10 @@ static inline void selinux_xfrm_notify_policyload(void)
> {
> struct net *net;
>
> - down_read(&net_rwsem);
> + rtnl_lock();
> for_each_net(net)
> rt_genid_bump_all(net);
> - up_read(&net_rwsem);
> + rtnl_unlock();
> }
> #else
> static inline int selinux_xfrm_enabled(void)
>
^ permalink raw reply
* [PATCH] net: mvneta: fix enable of all initialized RXQs
From: Gregory CLEMENT @ 2018-03-30 10:05 UTC (permalink / raw)
To: David S. Miller, linux-kernel, netdev
Cc: Thomas Petazzoni, Antoine Tenart, Miquèl Raynal,
Maxime Chevallier, stefanc, ymarkman, Yelena Krivosheev,
Gregory CLEMENT
From: Yelena Krivosheev <yelena@marvell.com>
In mvneta_port_up() we enable relevant RX and TX port queues by write
queues bit map to an appropriate register.
q_map must be ZERO in the beginning of this process.
Signed-off-by: Yelena Krivosheev <yelena@marvell.com>
Signed-off-by: Gregory CLEMENT <gregory.clement@bootlin.com>
---
drivers/net/ethernet/marvell/mvneta.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/net/ethernet/marvell/mvneta.c b/drivers/net/ethernet/marvell/mvneta.c
index 25e9a551cc8c..3f6fb635738c 100644
--- a/drivers/net/ethernet/marvell/mvneta.c
+++ b/drivers/net/ethernet/marvell/mvneta.c
@@ -1132,6 +1132,7 @@ static void mvneta_port_up(struct mvneta_port *pp)
}
mvreg_write(pp, MVNETA_TXQ_CMD, q_map);
+ q_map = 0;
/* Enable all initialized RXQs. */
for (queue = 0; queue < rxq_number; queue++) {
struct mvneta_rx_queue *rxq = &pp->rxqs[queue];
--
2.16.2
^ permalink raw reply related
* Re: [PATCH net v3 1/3] ipv6: move ip6_dst_store() calls with flowi6 checks to a wrapper
From: kbuild test robot @ 2018-03-30 10:14 UTC (permalink / raw)
To: Alexey Kodanev
Cc: kbuild-all, netdev, Eric Dumazet, Martin KaFai Lau, David Miller,
Alexey Kodanev
In-Reply-To: <1522345042-26646-2-git-send-email-alexey.kodanev@oracle.com>
[-- Attachment #1: Type: text/plain, Size: 2262 bytes --]
Hi Alexey,
I love your patch! Perhaps something to improve:
[auto build test WARNING on net/master]
url: https://github.com/0day-ci/linux/commits/Alexey-Kodanev/ipv6-move-ip6_dst_store-calls-with-flowi6-checks-to-a-wrapper/20180330-173050
config: ia64-defconfig (attached as .config)
compiler: ia64-linux-gcc (GCC) 7.2.0
reproduce:
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
chmod +x ~/bin/make.cross
# save the attached .config to linux build tree
make.cross ARCH=ia64
All warnings (new ones prefixed by >>):
In file included from include/linux/tcp.h:23:0,
from include/linux/ipv6.h:87,
from include/net/ipv6.h:16,
from include/net/inetpeer.h:16,
from include/net/route.h:28,
from drivers/infiniband/core/addr.c:43:
include/net/ip6_route.h: In function 'ip6_dst_store_flow':
include/net/sock.h:349:34: error: 'struct sock_common' has no member named 'skc_v6_daddr'; did you mean 'skc_daddr'?
#define sk_v6_daddr __sk_common.skc_v6_daddr
^
>> include/net/ip6_route.h:221:43: note: in expansion of macro 'sk_v6_daddr'
ipv6_addr_equal(&fl6->daddr, &sk->sk_v6_daddr) ?
^~~~~~~~~~~
include/net/sock.h:349:34: error: 'struct sock_common' has no member named 'skc_v6_daddr'; did you mean 'skc_daddr'?
#define sk_v6_daddr __sk_common.skc_v6_daddr
^
include/net/ip6_route.h:222:14: note: in expansion of macro 'sk_v6_daddr'
&sk->sk_v6_daddr : NULL,
^~~~~~~~~~~
vim +/sk_v6_daddr +221 include/net/ip6_route.h
219
220 ip6_dst_store(sk, dst,
> 221 ipv6_addr_equal(&fl6->daddr, &sk->sk_v6_daddr) ?
222 &sk->sk_v6_daddr : NULL,
223 #ifdef CONFIG_IPV6_SUBTREES
224 ipv6_addr_equal(&fl6->saddr, &np->saddr) ?
225 &np->saddr :
226 #endif
227 NULL);
228 }
229
---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
[-- Attachment #2: .config.gz --]
[-- Type: application/gzip, Size: 18690 bytes --]
^ permalink raw reply
* Re: [PATCH net-next 0/9] devlink: Add support for region access
From: Jiri Pirko @ 2018-03-30 10:21 UTC (permalink / raw)
To: Andrew Lunn
Cc: Alex Vesker, David S. Miller, netdev, Tariq Toukan, Jiri Pirko
In-Reply-To: <20180329195154.GB15565@lunn.ch>
Thu, Mar 29, 2018 at 09:51:54PM CEST, andrew@lunn.ch wrote:
>> >>Show all of the exposed regions with region sizes:
>> >>$ devlink region show
>> >>pci/0000:00:05.0/cr-space: size 1048576 snapshot [1 2]
>> >So you have 2Mbytes of snapshot data. Is this held in the device, or
>> >kernel memory?
>> This is allocated in devlink, the maximum number of snapshots is set by the
>> driver.
>
>And it seems to want contiguous pages. How well does that work after
>the system has been running for a while and memory is fragmented?
>
>> >>Dump a snapshot:
>> >>$ devlink region dump pci/0000:00:05.0/fw-health snapshot 1
>> >>0000000000000000 0014 95dc 0014 9514 0035 1670 0034 db30
>> >>0000000000000010 0000 0000 ffff ff04 0029 8c00 0028 8cc8
>> >>0000000000000020 0016 0bb8 0016 1720 0000 0000 c00f 3ffc
>> >>0000000000000030 bada cce5 bada cce5 bada cce5 bada cce5
>> >>
>> >>Read a specific part of a snapshot:
>> >>$ devlink region read pci/0000:00:05.0/fw-health snapshot 1 address 0
>> >> length 16
>> >>0000000000000000 0014 95dc 0014 9514 0035 1670 0034 db30
>> >Why a separate command? It seems to be just a subset of dump.
>>
>> This is useful when debugging values on specific addresses, this also
>> brings the API one step closer for a read and write API.
>
>The functionality is useful, yes. But why two commands? Why not one
>command, dump, which takes optional parameters?
Between userspace and kernel, this is implemented as a single command.
So this is just a userspace wrapper. I think it is nice to provide clear
commands to the user so he is not confused about what is he doing. Also,
as Alex mentioned, we plan to have write command which will have same
command like args as read. These 2 should be in sync.
>
>Also, i doubt write support will be accepted. That sounds like the
>start of an API to allow a user space driver.
We discussed that on netconf in Seoul and was agreed it is needed.
We have 2 options: Some out of tree crap utils and access via /dev/mem,
or something which is well defined and implemented by in-tree drivers.
Writes will serve for debugging purposes, even tuning and bug hunting
on in production. For that, we need a standard way to do it.
^ permalink raw reply
* Re: [PATCH] net: mvneta: fix enable of all initialized RXQs
From: Thomas Petazzoni @ 2018-03-30 10:22 UTC (permalink / raw)
To: Gregory CLEMENT
Cc: David S. Miller, linux-kernel, netdev, Antoine Tenart,
Miquèl Raynal, Maxime Chevallier, stefanc, ymarkman,
Yelena Krivosheev
In-Reply-To: <20180330100531.5995-1-gregory.clement@bootlin.com>
Hello,
On Fri, 30 Mar 2018 12:05:31 +0200, Gregory CLEMENT wrote:
> From: Yelena Krivosheev <yelena@marvell.com>
>
> In mvneta_port_up() we enable relevant RX and TX port queues by write
> queues bit map to an appropriate register.
>
> q_map must be ZERO in the beginning of this process.
>
> Signed-off-by: Yelena Krivosheev <yelena@marvell.com>
> Signed-off-by: Gregory CLEMENT <gregory.clement@bootlin.com>
Acked-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Thanks,
Thomas
--
Thomas Petazzoni, CTO, Bootlin (formerly Free Electrons)
Embedded Linux and Kernel engineering
https://bootlin.com
^ permalink raw reply
* Re: [PATCH net-next 0/9] devlink: Add support for region access
From: Jiri Pirko @ 2018-03-30 10:24 UTC (permalink / raw)
To: Rahul Lakkireddy
Cc: Andrew Lunn, Alex Vesker, David S. Miller, netdev@vger.kernel.org,
Tariq Toukan, Jiri Pirko
In-Reply-To: <20180330095156.GA24242@chelsio.com>
Fri, Mar 30, 2018 at 11:51:57AM CEST, rahul.lakkireddy@chelsio.com wrote:
>On Thursday, March 03/29/18, 2018 at 23:53:43 +0530, Andrew Lunn wrote:
>> On Thu, Mar 29, 2018 at 07:07:43PM +0300, Alex Vesker wrote:
>> > This is a proposal which will allow access to driver defined address
>> > regions using devlink. Each device can create its supported address
>> > regions and register them. A device which exposes a region will allow
>> > access to it using devlink.
>>
>> Hi Alex
>>
>> Did you see the work Rahul Lakkireddy has been doing?
>>
>> https://patchwork.kernel.org/patch/10305935/
>>
>> It seems like these are similar, or at least overlapping. We probably
>> want one solution for both.
>>
>
>We're already collecting hardware snapshots when system is live with
>ethtool --getdump (which devlink tool is now trying to do).
Ethtool is definitelly a wrong tool for this. It uses netdev as a
handle, but the dumps happen on a parent device - represented by a
devlink instance.
Also, in devlink we have notifications so a deamon can actually listen
on a socket to see if there is new dump available due to a critical
event etc.
>
>We are now in the process of adding support to collect hardware
>snapshots during kernel panic.
>
>Thanks,
>Rahul
^ permalink raw reply
* [PATCH v2 0/2] net: mvneta: improve suspend/resume
From: Jisheng Zhang @ 2018-03-30 10:33 UTC (permalink / raw)
To: David Miller, Thomas Petazzoni; +Cc: netdev, linux-arm-kernel, linux-kernel
This series tries to optimize the mvneta's suspend/resume
implementation by only taking necessary actions.
Since v1:
- unify ret check
- try best to keep the suspend/resume behavior
- split txq deinit into sw/hw parts as well
- adjust mvneta_stop_dev() location
I didn't add Thomas's Ack tag to patch1, because in v2, I added new code
to split the txq deinit into two parts.
Jisheng Zhang (2):
net: mvneta: split rxq/txq init and txq deinit into SW and HW parts
net: mvneta: improve suspend/resume
drivers/net/ethernet/marvell/mvneta.c | 156 +++++++++++++++++++++++++++-------
1 file changed, 127 insertions(+), 29 deletions(-)
--
2.16.3
^ permalink raw reply
* [PATCH v2 1/2] net: mvneta: split rxq/txq init and txq deinit into SW and HW parts
From: Jisheng Zhang @ 2018-03-30 10:34 UTC (permalink / raw)
To: David Miller, Thomas Petazzoni; +Cc: netdev, linux-arm-kernel, linux-kernel
In-Reply-To: <20180330183332.3e180a1a@xhacker.debian>
This is to prepare the suspend/resume improvement in next patch. The
SW parts can be optimized out during resume.
As for rxq handling during suspend, we'd like to drop packets by
calling mvneta_rxq_drop_pkts() which is both SW and HW operation,
so we don't split rxq deinit.
Signed-off-by: Jisheng Zhang <Jisheng.Zhang@synaptics.com>
---
drivers/net/ethernet/marvell/mvneta.c | 85 +++++++++++++++++++++++++++--------
1 file changed, 66 insertions(+), 19 deletions(-)
diff --git a/drivers/net/ethernet/marvell/mvneta.c b/drivers/net/ethernet/marvell/mvneta.c
index 30aab9bf77cc..f96815853108 100644
--- a/drivers/net/ethernet/marvell/mvneta.c
+++ b/drivers/net/ethernet/marvell/mvneta.c
@@ -2796,10 +2796,8 @@ static void mvneta_rx_reset(struct mvneta_port *pp)
/* Rx/Tx queue initialization/cleanup methods */
-/* Create a specified RX queue */
-static int mvneta_rxq_init(struct mvneta_port *pp,
- struct mvneta_rx_queue *rxq)
-
+static int mvneta_rxq_sw_init(struct mvneta_port *pp,
+ struct mvneta_rx_queue *rxq)
{
rxq->size = pp->rx_ring_size;
@@ -2812,6 +2810,12 @@ static int mvneta_rxq_init(struct mvneta_port *pp,
rxq->last_desc = rxq->size - 1;
+ return 0;
+}
+
+static void mvneta_rxq_hw_init(struct mvneta_port *pp,
+ struct mvneta_rx_queue *rxq)
+{
/* Set Rx descriptors queue starting address */
mvreg_write(pp, MVNETA_RXQ_BASE_ADDR_REG(rxq->id), rxq->descs_phys);
mvreg_write(pp, MVNETA_RXQ_SIZE_REG(rxq->id), rxq->size);
@@ -2835,6 +2839,20 @@ static int mvneta_rxq_init(struct mvneta_port *pp,
mvneta_rxq_short_pool_set(pp, rxq);
mvneta_rxq_non_occup_desc_add(pp, rxq, rxq->size);
}
+}
+
+/* Create a specified RX queue */
+static int mvneta_rxq_init(struct mvneta_port *pp,
+ struct mvneta_rx_queue *rxq)
+
+{
+ int ret;
+
+ ret = mvneta_rxq_sw_init(pp, rxq);
+ if (ret < 0)
+ return ret;
+
+ mvneta_rxq_hw_init(pp, rxq);
return 0;
}
@@ -2857,9 +2875,8 @@ static void mvneta_rxq_deinit(struct mvneta_port *pp,
rxq->descs_phys = 0;
}
-/* Create and initialize a tx queue */
-static int mvneta_txq_init(struct mvneta_port *pp,
- struct mvneta_tx_queue *txq)
+static int mvneta_txq_sw_init(struct mvneta_port *pp,
+ struct mvneta_tx_queue *txq)
{
int cpu;
@@ -2872,7 +2889,6 @@ static int mvneta_txq_init(struct mvneta_port *pp,
txq->tx_stop_threshold = txq->size - MVNETA_MAX_SKB_DESCS;
txq->tx_wake_threshold = txq->tx_stop_threshold / 2;
-
/* Allocate memory for TX descriptors */
txq->descs = dma_alloc_coherent(pp->dev->dev.parent,
txq->size * MVNETA_DESC_ALIGNED_SIZE,
@@ -2882,14 +2898,6 @@ static int mvneta_txq_init(struct mvneta_port *pp,
txq->last_desc = txq->size - 1;
- /* Set maximum bandwidth for enabled TXQs */
- mvreg_write(pp, MVETH_TXQ_TOKEN_CFG_REG(txq->id), 0x03ffffff);
- mvreg_write(pp, MVETH_TXQ_TOKEN_COUNT_REG(txq->id), 0x3fffffff);
-
- /* Set Tx descriptors queue starting address */
- mvreg_write(pp, MVNETA_TXQ_BASE_ADDR_REG(txq->id), txq->descs_phys);
- mvreg_write(pp, MVNETA_TXQ_SIZE_REG(txq->id), txq->size);
-
txq->tx_skb = kmalloc_array(txq->size, sizeof(*txq->tx_skb),
GFP_KERNEL);
if (!txq->tx_skb) {
@@ -2910,7 +2918,6 @@ static int mvneta_txq_init(struct mvneta_port *pp,
txq->descs, txq->descs_phys);
return -ENOMEM;
}
- mvneta_tx_done_pkts_coal_set(pp, txq, txq->done_pkts_coal);
/* Setup XPS mapping */
if (txq_number > 1)
@@ -2923,9 +2930,38 @@ static int mvneta_txq_init(struct mvneta_port *pp,
return 0;
}
+static void mvneta_txq_hw_init(struct mvneta_port *pp,
+ struct mvneta_tx_queue *txq)
+{
+ /* Set maximum bandwidth for enabled TXQs */
+ mvreg_write(pp, MVETH_TXQ_TOKEN_CFG_REG(txq->id), 0x03ffffff);
+ mvreg_write(pp, MVETH_TXQ_TOKEN_COUNT_REG(txq->id), 0x3fffffff);
+
+ /* Set Tx descriptors queue starting address */
+ mvreg_write(pp, MVNETA_TXQ_BASE_ADDR_REG(txq->id), txq->descs_phys);
+ mvreg_write(pp, MVNETA_TXQ_SIZE_REG(txq->id), txq->size);
+
+ mvneta_tx_done_pkts_coal_set(pp, txq, txq->done_pkts_coal);
+}
+
+/* Create and initialize a tx queue */
+static int mvneta_txq_init(struct mvneta_port *pp,
+ struct mvneta_tx_queue *txq)
+{
+ int ret;
+
+ ret = mvneta_txq_sw_init(pp, txq);
+ if (ret < 0)
+ return ret;
+
+ mvneta_txq_hw_init(pp, txq);
+
+ return 0;
+}
+
/* Free allocated resources when mvneta_txq_init() fails to allocate memory*/
-static void mvneta_txq_deinit(struct mvneta_port *pp,
- struct mvneta_tx_queue *txq)
+static void mvneta_txq_sw_deinit(struct mvneta_port *pp,
+ struct mvneta_tx_queue *txq)
{
struct netdev_queue *nq = netdev_get_tx_queue(pp->dev, txq->id);
@@ -2946,7 +2982,11 @@ static void mvneta_txq_deinit(struct mvneta_port *pp,
txq->last_desc = 0;
txq->next_desc_to_proc = 0;
txq->descs_phys = 0;
+}
+static void mvneta_txq_hw_deinit(struct mvneta_port *pp,
+ struct mvneta_tx_queue *txq)
+{
/* Set minimum bandwidth for disabled TXQs */
mvreg_write(pp, MVETH_TXQ_TOKEN_CFG_REG(txq->id), 0);
mvreg_write(pp, MVETH_TXQ_TOKEN_COUNT_REG(txq->id), 0);
@@ -2956,6 +2996,13 @@ static void mvneta_txq_deinit(struct mvneta_port *pp,
mvreg_write(pp, MVNETA_TXQ_SIZE_REG(txq->id), 0);
}
+static void mvneta_txq_deinit(struct mvneta_port *pp,
+ struct mvneta_tx_queue *txq)
+{
+ mvneta_txq_sw_deinit(pp, txq);
+ mvneta_txq_hw_deinit(pp, txq);
+}
+
/* Cleanup all Tx queues */
static void mvneta_cleanup_txqs(struct mvneta_port *pp)
{
--
2.16.3
^ permalink raw reply related
* [PATCH v2 2/2] net: mvneta: improve suspend/resume
From: Jisheng Zhang @ 2018-03-30 10:36 UTC (permalink / raw)
To: David Miller, Thomas Petazzoni; +Cc: netdev, linux-arm-kernel, linux-kernel
In-Reply-To: <20180330183332.3e180a1a@xhacker.debian>
Current suspend/resume implementation reuses the mvneta_open() and
mvneta_close(), but it could be optimized to take only necessary
actions during suspend/resume.
One obvious problem of current implementation is: after hundreds of
system suspend/resume cycles, the resume of mvneta could fail due to
fragmented dma coherent memory. After this patch, the non-necessary
memory alloc/free is optimized out.
Signed-off-by: Jisheng Zhang <Jisheng.Zhang@synaptics.com>
---
drivers/net/ethernet/marvell/mvneta.c | 71 ++++++++++++++++++++++++++++++-----
1 file changed, 61 insertions(+), 10 deletions(-)
diff --git a/drivers/net/ethernet/marvell/mvneta.c b/drivers/net/ethernet/marvell/mvneta.c
index f96815853108..cb7fce99ed6d 100644
--- a/drivers/net/ethernet/marvell/mvneta.c
+++ b/drivers/net/ethernet/marvell/mvneta.c
@@ -4586,16 +4586,43 @@ static int mvneta_remove(struct platform_device *pdev)
#ifdef CONFIG_PM_SLEEP
static int mvneta_suspend(struct device *device)
{
+ int queue;
struct net_device *dev = dev_get_drvdata(device);
struct mvneta_port *pp = netdev_priv(dev);
- rtnl_lock();
- if (netif_running(dev))
- mvneta_stop(dev);
- rtnl_unlock();
+ if (!netif_running(dev))
+ goto clean_exit;
+
+ if (!pp->neta_armada3700) {
+ spin_lock(&pp->lock);
+ pp->is_stopped = true;
+ spin_unlock(&pp->lock);
+
+ cpuhp_state_remove_instance_nocalls(online_hpstate,
+ &pp->node_online);
+ cpuhp_state_remove_instance_nocalls(CPUHP_NET_MVNETA_DEAD,
+ &pp->node_dead);
+ }
+
+ mvneta_stop_dev(pp);
+
+ for (queue = 0; queue < rxq_number; queue++) {
+ struct mvneta_rx_queue *rxq = &pp->rxqs[queue];
+
+ mvneta_rxq_drop_pkts(pp, rxq);
+ }
+
+ for (queue = 0; queue < txq_number; queue++) {
+ struct mvneta_tx_queue *txq = &pp->txqs[queue];
+
+ mvneta_txq_hw_deinit(pp, txq);
+ }
+
+clean_exit:
netif_device_detach(dev);
clk_disable_unprepare(pp->clk_bus);
clk_disable_unprepare(pp->clk);
+
return 0;
}
@@ -4604,7 +4631,7 @@ static int mvneta_resume(struct device *device)
struct platform_device *pdev = to_platform_device(device);
struct net_device *dev = dev_get_drvdata(device);
struct mvneta_port *pp = netdev_priv(dev);
- int err;
+ int err, queue;
clk_prepare_enable(pp->clk);
if (!IS_ERR(pp->clk_bus))
@@ -4626,12 +4653,36 @@ static int mvneta_resume(struct device *device)
}
netif_device_attach(dev);
- rtnl_lock();
- if (netif_running(dev)) {
- mvneta_open(dev);
- mvneta_set_rx_mode(dev);
+
+ if (!netif_running(dev))
+ return 0;
+
+ for (queue = 0; queue < rxq_number; queue++) {
+ struct mvneta_rx_queue *rxq = &pp->rxqs[queue];
+
+ rxq->next_desc_to_proc = 0;
+ mvneta_rxq_hw_init(pp, rxq);
}
- rtnl_unlock();
+
+ for (queue = 0; queue < txq_number; queue++) {
+ struct mvneta_tx_queue *txq = &pp->txqs[queue];
+
+ txq->next_desc_to_proc = 0;
+ mvneta_txq_hw_init(pp, txq);
+ }
+
+ if (!pp->neta_armada3700) {
+ spin_lock(&pp->lock);
+ pp->is_stopped = false;
+ spin_unlock(&pp->lock);
+ cpuhp_state_add_instance_nocalls(online_hpstate,
+ &pp->node_online);
+ cpuhp_state_add_instance_nocalls(CPUHP_NET_MVNETA_DEAD,
+ &pp->node_dead);
+ }
+
+ mvneta_start_dev(pp);
+ mvneta_set_rx_mode(dev);
return 0;
}
--
2.16.3
^ permalink raw reply related
* Re: [PATCH net-next v2 1/2] fs/crashdd: add API to collect hardware dump in second kernel
From: Jiri Pirko @ 2018-03-30 10:39 UTC (permalink / raw)
To: Rahul Lakkireddy
Cc: netdev, linux-fsdevel, kexec, linux-kernel, davem, viro, ebiederm,
stephen, akpm, torvalds, ganeshgr, nirranjan, indranil
In-Reply-To: <296ffbd47fd4f30238689e636bd2480683224227.1521888444.git.rahul.lakkireddy@chelsio.com>
Sat, Mar 24, 2018 at 11:56:33AM CET, rahul.lakkireddy@chelsio.com wrote:
>Add a new module crashdd that exports the /sys/kernel/crashdd/
>directory in second kernel, containing collected hardware/firmware
>dumps.
>
>The sequence of actions done by device drivers to append their device
>specific hardware/firmware logs to /sys/kernel/crashdd/ directory are
>as follows:
>
>1. During probe (before hardware is initialized), device drivers
>register to the crashdd module (via crashdd_add_dump()), with
>callback function, along with buffer size and log name needed for
>firmware/hardware log collection.
>
>2. Crashdd creates a driver's directory under
>/sys/kernel/crashdd/<driver>. Then, it allocates the buffer with
This smells. I need to identify the exact ASIC instance that produced
the dump. To identify by driver name does not help me if I have multiple
instances of the same driver. This looks wrong to me. This looks like
a job for devlink where you have 1 devlink instance per 1 ASIC instance.
Please see:
http://patchwork.ozlabs.org/project/netdev/list/?series=36524
I bevieve that the solution in the patchset could be used for
your usecase too.
>requested size and invokes the device driver's registered callback
>function.
>
>3. Device driver collects all hardware/firmware logs into the buffer
>and returns control back to crashdd.
>
>4. Crashdd exposes the buffer as a binary file via
>/sys/kernel/crashdd/<driver>/<dump_file>.
>
^ permalink raw reply
* Re: [PATCH v2 2/2] net: mvneta: improve suspend/resume
From: Russell King - ARM Linux @ 2018-03-30 10:43 UTC (permalink / raw)
To: Jisheng Zhang
Cc: David Miller, Thomas Petazzoni, netdev, linux-kernel,
linux-arm-kernel
In-Reply-To: <20180330183615.6b34d153@xhacker.debian>
On Fri, Mar 30, 2018 at 06:36:15PM +0800, Jisheng Zhang wrote:
> Current suspend/resume implementation reuses the mvneta_open() and
> mvneta_close(), but it could be optimized to take only necessary
> actions during suspend/resume.
>
> One obvious problem of current implementation is: after hundreds of
> system suspend/resume cycles, the resume of mvneta could fail due to
> fragmented dma coherent memory. After this patch, the non-necessary
> memory alloc/free is optimized out.
I don't think you've properly tested this. Please ensure that you test
patches with the appropriate debug options enabled.
> @@ -4586,16 +4586,43 @@ static int mvneta_remove(struct platform_device *pdev)
> #ifdef CONFIG_PM_SLEEP
> static int mvneta_suspend(struct device *device)
> {
> + int queue;
> struct net_device *dev = dev_get_drvdata(device);
> struct mvneta_port *pp = netdev_priv(dev);
>
> - rtnl_lock();
> - if (netif_running(dev))
> - mvneta_stop(dev);
> - rtnl_unlock();
...
> + mvneta_stop_dev(pp);
You're removing the rtnl_lock() that I introduced in 3b8bc67413de
("net: mvneta: ensure PM paths take the rtnl lock") which is necessary
to provide phylink with consistent locking. mvneta_stop_dev() calls
phylink_stop() which will check that the rtnl lock is held, and will
print a warning if it isn't.
Your patch will cause a regression here.
> +
> + for (queue = 0; queue < rxq_number; queue++) {
> + struct mvneta_rx_queue *rxq = &pp->rxqs[queue];
> +
> + mvneta_rxq_drop_pkts(pp, rxq);
> + }
> +
> + for (queue = 0; queue < txq_number; queue++) {
> + struct mvneta_tx_queue *txq = &pp->txqs[queue];
> +
> + mvneta_txq_hw_deinit(pp, txq);
> + }
> +
> +clean_exit:
> netif_device_detach(dev);
> clk_disable_unprepare(pp->clk_bus);
> clk_disable_unprepare(pp->clk);
> +
> return 0;
> }
>
> @@ -4604,7 +4631,7 @@ static int mvneta_resume(struct device *device)
> struct platform_device *pdev = to_platform_device(device);
> struct net_device *dev = dev_get_drvdata(device);
> struct mvneta_port *pp = netdev_priv(dev);
> - int err;
> + int err, queue;
>
> clk_prepare_enable(pp->clk);
> if (!IS_ERR(pp->clk_bus))
> @@ -4626,12 +4653,36 @@ static int mvneta_resume(struct device *device)
> }
>
> netif_device_attach(dev);
> - rtnl_lock();
> - if (netif_running(dev)) {
> - mvneta_open(dev);
> - mvneta_set_rx_mode(dev);
...
> }
> - rtnl_unlock();
...
> + mvneta_start_dev(pp);
Same applies here.
--
RMK's Patch system: http://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line in suburbia: sync at 8.8Mbps down 630kbps up
According to speedtest.net: 8.21Mbps down 510kbps up
^ permalink raw reply
* [PATCH net-next 0/2] net: bridge: MTU handling changes
From: Nikolay Aleksandrov @ 2018-03-30 10:46 UTC (permalink / raw)
To: netdev; +Cc: Nikolay Aleksandrov, roopa, bridge, 3chas3, davem
Hi,
As previously discussed the recent changes break some setups and could lead
to packet drops. Thus the first patch reverts the behaviour for the bridge
to follow the minimum MTU but also keeps the ability to set the MTU to the
maximum (out of all ports) if vlan filtering is enabled. Patch 02 is the
bigger change in behaviour - we've always had trouble when configuring
bridges and their MTU which is auto tuning on port events
(add/del/changemtu), which means config software needs to chase it and fix
it after each such event, after patch 02 we allow the user to configure any
MTU (ETH_MIN/MAX limited) but once that is done the bridge stops auto
tuning and relies on the user to keep the MTU correct.
This should be compatible with cases that don't touch the MTU (or set it
to the same value), while allowing to configure the MTU and not worry
about it changing afterwards.
The patches are intentionally split like this, so that if they get accepted
and there are any complaints patch 02 can be reverted.
Thanks,
Nik
Nikolay Aleksandrov (2):
net: bridge: set min MTU on port events and allow user to set max
net: bridge: disable bridge MTU auto tuning if it was set manually
net/bridge/br.c | 2 +-
net/bridge/br_device.c | 4 ++--
net/bridge/br_if.c | 49 ++++++++++++++++++++-----------------------------
net/bridge/br_private.h | 3 ++-
4 files changed, 25 insertions(+), 33 deletions(-)
--
2.11.0
^ permalink raw reply
* [PATCH net-next 1/2] net: bridge: set min MTU on port events and allow user to set max
From: Nikolay Aleksandrov @ 2018-03-30 10:46 UTC (permalink / raw)
To: netdev
Cc: roopa, davem, makita.toshiaki, 3chas3, stephen, bridge,
Nikolay Aleksandrov
In-Reply-To: <20180330104619.31479-1-nikolay@cumulusnetworks.com>
Recently the bridge was changed to automatically set maximum MTU on port
events (add/del/changemtu) when vlan filtering is enabled, but that
actually changes behaviour in a way which breaks some setups and can lead
to packet drops. In order to still allow that maximum to be set while being
compatible, we add the ability for the user to tune the bridge MTU up to
the maximum when vlan filtering is enabled, but that has to be done
explicitly and all port events (add/del/changemtu) lead to resetting that
MTU to the minimum as before.
Suggested-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
---
net/bridge/br.c | 2 +-
net/bridge/br_device.c | 3 ++-
net/bridge/br_if.c | 43 ++++++++++++++-----------------------------
net/bridge/br_private.h | 2 +-
4 files changed, 18 insertions(+), 32 deletions(-)
diff --git a/net/bridge/br.c b/net/bridge/br.c
index 26e1616b2c90..565ff055813b 100644
--- a/net/bridge/br.c
+++ b/net/bridge/br.c
@@ -52,7 +52,7 @@ static int br_device_event(struct notifier_block *unused, unsigned long event, v
switch (event) {
case NETDEV_CHANGEMTU:
- dev_set_mtu(br->dev, br_mtu(br));
+ dev_set_mtu(br->dev, br_mtu(br, false));
break;
case NETDEV_CHANGEADDR:
diff --git a/net/bridge/br_device.c b/net/bridge/br_device.c
index 278fc999d355..edb9967eb165 100644
--- a/net/bridge/br_device.c
+++ b/net/bridge/br_device.c
@@ -224,7 +224,8 @@ static void br_get_stats64(struct net_device *dev,
static int br_change_mtu(struct net_device *dev, int new_mtu)
{
struct net_bridge *br = netdev_priv(dev);
- if (new_mtu > br_mtu(br))
+
+ if (new_mtu > br_mtu(br, br_vlan_enabled(dev)))
return -EINVAL;
dev->mtu = new_mtu;
diff --git a/net/bridge/br_if.c b/net/bridge/br_if.c
index 87b2afd455c7..7d5dc5a91084 100644
--- a/net/bridge/br_if.c
+++ b/net/bridge/br_if.c
@@ -424,41 +424,26 @@ int br_del_bridge(struct net *net, const char *name)
return ret;
}
-static bool min_mtu(int a, int b)
-{
- return a < b ? 1 : 0;
-}
-
-static bool max_mtu(int a, int b)
-{
- return a > b ? 1 : 0;
-}
-
-/* MTU of the bridge pseudo-device: ETH_DATA_LEN or the minimum of the ports */
-static int __br_mtu(const struct net_bridge *br, bool (compare_fn)(int, int))
+/* MTU of the bridge pseudo-device: ETH_DATA_LEN if there are no ports, the
+ * minimum of the ports if @max is false or the maximum if it's true
+ */
+int br_mtu(const struct net_bridge *br, bool max)
{
const struct net_bridge_port *p;
- int mtu = 0;
+ int ret_mtu = 0;
ASSERT_RTNL();
- if (list_empty(&br->port_list))
- mtu = ETH_DATA_LEN;
- else {
- list_for_each_entry(p, &br->port_list, list) {
- if (!mtu || compare_fn(p->dev->mtu, mtu))
- mtu = p->dev->mtu;
+ list_for_each_entry(p, &br->port_list, list) {
+ if (!max) {
+ if (!ret_mtu || ret_mtu > p->dev->mtu)
+ ret_mtu = p->dev->mtu;
+ } else if (p->dev->mtu > ret_mtu) {
+ ret_mtu = p->dev->mtu;
}
}
- return mtu;
-}
-int br_mtu(const struct net_bridge *br)
-{
- if (br_vlan_enabled(br->dev))
- return __br_mtu(br, max_mtu);
- else
- return __br_mtu(br, min_mtu);
+ return ret_mtu ? ret_mtu : ETH_DATA_LEN;
}
static void br_set_gso_limits(struct net_bridge *br)
@@ -612,7 +597,7 @@ int br_add_if(struct net_bridge *br, struct net_device *dev,
if (changed_addr)
call_netdevice_notifiers(NETDEV_CHANGEADDR, br->dev);
- dev_set_mtu(br->dev, br_mtu(br));
+ dev_set_mtu(br->dev, br_mtu(br, false));
br_set_gso_limits(br);
kobject_uevent(&p->kobj, KOBJ_ADD);
@@ -659,7 +644,7 @@ int br_del_if(struct net_bridge *br, struct net_device *dev)
*/
del_nbp(p);
- dev_set_mtu(br->dev, br_mtu(br));
+ dev_set_mtu(br->dev, br_mtu(br, false));
br_set_gso_limits(br);
spin_lock_bh(&br->lock);
diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h
index 048d5b51813b..586f84b9670d 100644
--- a/net/bridge/br_private.h
+++ b/net/bridge/br_private.h
@@ -578,7 +578,7 @@ int br_del_bridge(struct net *net, const char *name);
int br_add_if(struct net_bridge *br, struct net_device *dev,
struct netlink_ext_ack *extack);
int br_del_if(struct net_bridge *br, struct net_device *dev);
-int br_mtu(const struct net_bridge *br);
+int br_mtu(const struct net_bridge *br, bool max);
netdev_features_t br_features_recompute(struct net_bridge *br,
netdev_features_t features);
void br_port_flags_change(struct net_bridge_port *port, unsigned long mask);
--
2.11.0
^ permalink raw reply related
* [PATCH net-next 2/2] net: bridge: disable bridge MTU auto tuning if it was set manually
From: Nikolay Aleksandrov @ 2018-03-30 10:46 UTC (permalink / raw)
To: netdev
Cc: roopa, davem, makita.toshiaki, 3chas3, stephen, bridge,
Nikolay Aleksandrov
In-Reply-To: <20180330104619.31479-1-nikolay@cumulusnetworks.com>
As Roopa noted today the biggest source of problems when configuring
bridge and ports is that the bridge MTU keeps changing automatically on
port events (add/del/changemtu). That leads to inconsistent behaviour
and network config software needs to chase the MTU and fix it on each
such event. Let's improve on that situation and allow for the user to
set any MTU within ETH_MIN/MAX limits, but once manually configured it
is the user's responsibility to keep it correct afterwards.
In case the MTU isn't manually set - the behaviour reverts to the
previous and the bridge follows the minimum MTU.
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
---
net/bridge/br.c | 2 +-
net/bridge/br_device.c | 5 ++---
net/bridge/br_if.c | 36 +++++++++++++++++++++---------------
net/bridge/br_private.h | 3 ++-
4 files changed, 26 insertions(+), 20 deletions(-)
diff --git a/net/bridge/br.c b/net/bridge/br.c
index 565ff055813b..671d13c10f6f 100644
--- a/net/bridge/br.c
+++ b/net/bridge/br.c
@@ -52,7 +52,7 @@ static int br_device_event(struct notifier_block *unused, unsigned long event, v
switch (event) {
case NETDEV_CHANGEMTU:
- dev_set_mtu(br->dev, br_mtu(br, false));
+ br_mtu_auto_adjust(br);
break;
case NETDEV_CHANGEADDR:
diff --git a/net/bridge/br_device.c b/net/bridge/br_device.c
index edb9967eb165..e682a668ce57 100644
--- a/net/bridge/br_device.c
+++ b/net/bridge/br_device.c
@@ -225,11 +225,10 @@ static int br_change_mtu(struct net_device *dev, int new_mtu)
{
struct net_bridge *br = netdev_priv(dev);
- if (new_mtu > br_mtu(br, br_vlan_enabled(dev)))
- return -EINVAL;
-
dev->mtu = new_mtu;
+ /* this flag will be cleared if the MTU was automatically adjusted */
+ br->mtu_set_by_user = true;
#if IS_ENABLED(CONFIG_BRIDGE_NETFILTER)
/* remember the MTU in the rtable for PMTU */
dst_metric_set(&br->fake_rtable.dst, RTAX_MTU, new_mtu);
diff --git a/net/bridge/br_if.c b/net/bridge/br_if.c
index 7d5dc5a91084..82c1a6f430b3 100644
--- a/net/bridge/br_if.c
+++ b/net/bridge/br_if.c
@@ -424,28 +424,34 @@ int br_del_bridge(struct net *net, const char *name)
return ret;
}
-/* MTU of the bridge pseudo-device: ETH_DATA_LEN if there are no ports, the
- * minimum of the ports if @max is false or the maximum if it's true
- */
-int br_mtu(const struct net_bridge *br, bool max)
+/* MTU of the bridge pseudo-device: ETH_DATA_LEN or the minimum of the ports */
+static int br_mtu_min(const struct net_bridge *br)
{
const struct net_bridge_port *p;
int ret_mtu = 0;
- ASSERT_RTNL();
-
- list_for_each_entry(p, &br->port_list, list) {
- if (!max) {
- if (!ret_mtu || ret_mtu > p->dev->mtu)
- ret_mtu = p->dev->mtu;
- } else if (p->dev->mtu > ret_mtu) {
+ list_for_each_entry(p, &br->port_list, list)
+ if (!ret_mtu || ret_mtu > p->dev->mtu)
ret_mtu = p->dev->mtu;
- }
- }
return ret_mtu ? ret_mtu : ETH_DATA_LEN;
}
+void br_mtu_auto_adjust(struct net_bridge *br)
+{
+ ASSERT_RTNL();
+
+ /* if the bridge MTU was manually configured don't mess with it */
+ if (br->mtu_set_by_user)
+ return;
+
+ /* change to the minimum MTU and clear the flag which was set by
+ * the bridge ndo_change_mtu callback
+ */
+ dev_set_mtu(br->dev, br_mtu_min(br));
+ br->mtu_set_by_user = false;
+}
+
static void br_set_gso_limits(struct net_bridge *br)
{
unsigned int gso_max_size = GSO_MAX_SIZE;
@@ -597,7 +603,7 @@ int br_add_if(struct net_bridge *br, struct net_device *dev,
if (changed_addr)
call_netdevice_notifiers(NETDEV_CHANGEADDR, br->dev);
- dev_set_mtu(br->dev, br_mtu(br, false));
+ br_mtu_auto_adjust(br);
br_set_gso_limits(br);
kobject_uevent(&p->kobj, KOBJ_ADD);
@@ -644,7 +650,7 @@ int br_del_if(struct net_bridge *br, struct net_device *dev)
*/
del_nbp(p);
- dev_set_mtu(br->dev, br_mtu(br, false));
+ br_mtu_auto_adjust(br);
br_set_gso_limits(br);
spin_lock_bh(&br->lock);
diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h
index 586f84b9670d..a7cb3ece5031 100644
--- a/net/bridge/br_private.h
+++ b/net/bridge/br_private.h
@@ -410,6 +410,7 @@ struct net_bridge {
int offload_fwd_mark;
#endif
bool neigh_suppress_enabled;
+ bool mtu_set_by_user;
struct hlist_head fdb_list;
};
@@ -578,7 +579,7 @@ int br_del_bridge(struct net *net, const char *name);
int br_add_if(struct net_bridge *br, struct net_device *dev,
struct netlink_ext_ack *extack);
int br_del_if(struct net_bridge *br, struct net_device *dev);
-int br_mtu(const struct net_bridge *br, bool max);
+void br_mtu_auto_adjust(struct net_bridge *br);
netdev_features_t br_features_recompute(struct net_bridge *br,
netdev_features_t features);
void br_port_flags_change(struct net_bridge_port *port, unsigned long mask);
--
2.11.0
^ permalink raw reply related
* Re: [PATCH net-next v2 1/2] fs/crashdd: add API to collect hardware dump in second kernel
From: Rahul Lakkireddy @ 2018-03-30 10:51 UTC (permalink / raw)
To: Jiri Pirko
Cc: Indranil Choudhury,
netdev-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
Nirranjan Kirubaharan,
kexec-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org,
linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
stephen-OTpzqLSitTUnbdJkjeBofR2eb7JE58TQ@public.gmane.org,
Ganesh GR, viro-RmSDqhL/yNMiFSDQTTA3OLVCufUGDwFn@public.gmane.org,
linux-fsdevel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
akpm-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b@public.gmane.org,
torvalds-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b@public.gmane.org,
davem-fT/PcQaiUtIeIZ0/mPfg9Q@public.gmane.org,
ebiederm-aS9lmoZGLiVWk0Htik3J/w@public.gmane.org
In-Reply-To: <20180330103907.GC3313@nanopsycho>
On Friday, March 03/30/18, 2018 at 16:09:07 +0530, Jiri Pirko wrote:
> Sat, Mar 24, 2018 at 11:56:33AM CET, rahul.lakkireddy-ut6Up61K2wZBDgjK7y7TUQ@public.gmane.org wrote:
> >Add a new module crashdd that exports the /sys/kernel/crashdd/
> >directory in second kernel, containing collected hardware/firmware
> >dumps.
> >
> >The sequence of actions done by device drivers to append their device
> >specific hardware/firmware logs to /sys/kernel/crashdd/ directory are
> >as follows:
> >
> >1. During probe (before hardware is initialized), device drivers
> >register to the crashdd module (via crashdd_add_dump()), with
> >callback function, along with buffer size and log name needed for
> >firmware/hardware log collection.
> >
> >2. Crashdd creates a driver's directory under
> >/sys/kernel/crashdd/<driver>. Then, it allocates the buffer with
>
> This smells. I need to identify the exact ASIC instance that produced
> the dump. To identify by driver name does not help me if I have multiple
> instances of the same driver. This looks wrong to me. This looks like
> a job for devlink where you have 1 devlink instance per 1 ASIC instance.
>
> Please see:
> http://patchwork.ozlabs.org/project/netdev/list/?series=36524
>
> I bevieve that the solution in the patchset could be used for
> your usecase too.
>
>
The sysfs approach proposed here had been dropped in favour exporting
the dumps as ELF notes in /proc/vmcore.
Will be posting the new patches soon.
> >requested size and invokes the device driver's registered callback
> >function.
> >
> >3. Device driver collects all hardware/firmware logs into the buffer
> >and returns control back to crashdd.
> >
> >4. Crashdd exposes the buffer as a binary file via
> >/sys/kernel/crashdd/<driver>/<dump_file>.
> >
^ permalink raw reply
* Re: [PATCH net v3 1/3] ipv6: move ip6_dst_store() calls with flowi6 checks to a wrapper
From: Alexey Kodanev @ 2018-03-30 11:36 UTC (permalink / raw)
To: kbuild test robot
Cc: kbuild-all, netdev, Eric Dumazet, Martin KaFai Lau, David Miller
In-Reply-To: <201803301853.dWQl77c0%fengguang.wu@intel.com>
On 03/30/2018 01:14 PM, kbuild test robot wrote:
> Hi Alexey,
>
> I love your patch! Perhaps something to improve:
>
> [auto build test WARNING on net/master]
>
> url: https://github.com/0day-ci/linux/commits/Alexey-Kodanev/ipv6-move-ip6_dst_store-calls-with-flowi6-checks-to-a-wrapper/20180330-173050
> config: ia64-defconfig (attached as .config)
> compiler: ia64-linux-gcc (GCC) 7.2.0
> reproduce:
> wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
> chmod +x ~/bin/make.cross
> # save the attached .config to linux build tree
> make.cross ARCH=ia64
>
> All warnings (new ones prefixed by >>):
>
> In file included from include/linux/tcp.h:23:0,
> from include/linux/ipv6.h:87,
> from include/net/ipv6.h:16,
> from include/net/inetpeer.h:16,
> from include/net/route.h:28,
> from drivers/infiniband/core/addr.c:43:
> include/net/ip6_route.h: In function 'ip6_dst_store_flow':
> include/net/sock.h:349:34: error: 'struct sock_common' has no member named 'skc_v6_daddr'; did you mean 'skc_daddr'?
> #define sk_v6_daddr __sk_common.skc_v6_daddr
> ^
>>> include/net/ip6_route.h:221:43: note: in expansion of macro 'sk_v6_daddr'
> ipv6_addr_equal(&fl6->daddr, &sk->sk_v6_daddr) ?
> ^~~~~~~~~~~
> include/net/sock.h:349:34: error: 'struct sock_common' has no member named 'skc_v6_daddr'; did you mean 'skc_daddr'?
> #define sk_v6_daddr __sk_common.skc_v6_daddr
> ^
> include/net/ip6_route.h:222:14: note: in expansion of macro 'sk_v6_daddr'
> &sk->sk_v6_daddr : NULL
> ^~~~~~~~~~~
This is because CONFIG_IPV6 is not enabled, no sk_v6_daddr member,
I'll fix it in the new version.
Thanks,
Alexey
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox