| Message ID | 20250113003456.3904110-1-kuba@kernel.org (mailing list archive) |
|---|---|
| State | Mainlined |
| Delegated to: | Matthieu Baerts |
| Headers | show |
| Series | [net-next,v2,1/2] net: remove init_dummy_netdev() | expand |
| Context | Check | Description |
|---|---|---|
| matttbe/build | success | Build and static analysis OK |
| matttbe/checkpatch | success | total: 0 errors, 0 warnings, 0 checks, 79 lines checked |
| matttbe/shellcheck | success | MPTCP selftests files have not been modified |
From: Jakub Kicinski <kuba@kernel.org> Date: Sun, 12 Jan 2025 16:34:55 -0800 > init_dummy_netdev() can initialize statically declared or embedded > net_devices. Such netdevs did not come from alloc_netdev_mqs(). > After recent work by Breno, there are the only two cases where > we have do that. > > Switch those cases to alloc_netdev_mqs() and delete init_dummy_netdev(). > Dealing with static netdevs is not worth the maintenance burden. Series: Reviewed-by: Alexander Lobakin <aleksander.lobakin@intel.com> > > Signed-off-by: Jakub Kicinski <kuba@kernel.org> > --- > v2: change of plan, delete init_dummy_netdev() completely > v1: https://lore.kernel.org/20250111065955.3698801-1-kuba@kernel.org Thanks, Olek
Hi Jakub, On 13/01/2025 01:34, Jakub Kicinski wrote: > init_dummy_netdev() can initialize statically declared or embedded > net_devices. Such netdevs did not come from alloc_netdev_mqs(). > After recent work by Breno, there are the only two cases where > we have do that. > > Switch those cases to alloc_netdev_mqs() and delete init_dummy_netdev(). > Dealing with static netdevs is not worth the maintenance burden. Thank you for the patch! (...) > diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c > index 1b2e7cbb577f..c44c89ecaca6 100644 > --- a/net/mptcp/protocol.c > +++ b/net/mptcp/protocol.c > @@ -47,7 +47,7 @@ static void __mptcp_destroy_sock(struct sock *sk); > static void mptcp_check_send_data_fin(struct sock *sk); > > DEFINE_PER_CPU(struct mptcp_delegated_action, mptcp_delegated_actions); > -static struct net_device mptcp_napi_dev; > +static struct net_device *mptcp_napi_dev; > > /* Returns end sequence number of the receiver's advertised window */ > static u64 mptcp_wnd_end(const struct mptcp_sock *msk) > @@ -4147,11 +4147,13 @@ void __init mptcp_proto_init(void) > if (percpu_counter_init(&mptcp_sockets_allocated, 0, GFP_KERNEL)) > panic("Failed to allocate MPTCP pcpu counter\n"); > > - init_dummy_netdev(&mptcp_napi_dev); > + mptcp_napi_dev = alloc_netdev_dummy(0); This seems to initialise quite a few things that are not needed for the MPTCP case. But it doesn't seem to hurt, and the dummy net device doesn't seem to be exposed to the userspace (net devices, sysfs). So if it helps with the maintenance, it looks fine to do that. For the modifications in MPTCP code: Reviewed-by: Matthieu Baerts (NGI0) <matttbe@kernel.org> > + if (!mptcp_napi_dev) > + panic("Failed to allocate MPTCP dummy netdev\n"); > for_each_possible_cpu(cpu) { > delegated = per_cpu_ptr(&mptcp_delegated_actions, cpu); > INIT_LIST_HEAD(&delegated->head); > - netif_napi_add_tx(&mptcp_napi_dev, &delegated->napi, > + netif_napi_add_tx(mptcp_napi_dev, &delegated->napi, > mptcp_napi_poll); > napi_enable(&delegated->napi); > } (...) Cheers, Matt
On Sun, Jan 12, 2025 at 04:34:55PM -0800, Jakub Kicinski wrote: > init_dummy_netdev() can initialize statically declared or embedded > net_devices. Such netdevs did not come from alloc_netdev_mqs(). > After recent work by Breno, there are the only two cases where > we have do that. > > Switch those cases to alloc_netdev_mqs() and delete init_dummy_netdev(). > Dealing with static netdevs is not worth the maintenance burden. > > Signed-off-by: Jakub Kicinski <kuba@kernel.org> > --- > v2: change of plan, delete init_dummy_netdev() completely > v1: https://lore.kernel.org/20250111065955.3698801-1-kuba@kernel.org > --- > CC: matttbe@kernel.org > CC: martineau@kernel.org > CC: geliang@kernel.org > CC: steffen.klassert@secunet.com > CC: herbert@gondor.apana.org.au > CC: mptcp@lists.linux.dev > --- > include/linux/netdevice.h | 1 - > net/core/dev.c | 22 ---------------------- > net/mptcp/protocol.c | 8 +++++--- > net/xfrm/xfrm_input.c | 9 ++++++--- > 4 files changed, 11 insertions(+), 29 deletions(-) Reviewed-by: Joe Damato <jdamato@fastly.com>
Hello: This series was applied to netdev/net-next.git (main) by Jakub Kicinski <kuba@kernel.org>: On Sun, 12 Jan 2025 16:34:55 -0800 you wrote: > init_dummy_netdev() can initialize statically declared or embedded > net_devices. Such netdevs did not come from alloc_netdev_mqs(). > After recent work by Breno, there are the only two cases where > we have do that. > > Switch those cases to alloc_netdev_mqs() and delete init_dummy_netdev(). > Dealing with static netdevs is not worth the maintenance burden. > > [...] Here is the summary with links: - [net-next,v2,1/2] net: remove init_dummy_netdev() https://git.kernel.org/netdev/net-next/c/f835bdae7167 - [net-next,v2,2/2] net: cleanup init_dummy_netdev_core() https://git.kernel.org/netdev/net-next/c/37adf101f6f7 You are awesome, thank you!
diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h index aeb4a6cff171..dd8f6f8991fe 100644 --- a/include/linux/netdevice.h +++ b/include/linux/netdevice.h @@ -3238,7 +3238,6 @@ static inline void unregister_netdevice(struct net_device *dev) int netdev_refcnt_read(const struct net_device *dev); void free_netdev(struct net_device *dev); -void init_dummy_netdev(struct net_device *dev); struct net_device *netdev_get_xmit_slave(struct net_device *dev, struct sk_buff *skb, diff --git a/net/core/dev.c b/net/core/dev.c index 1a90ed8cc6cc..c9abc9fc770e 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -10762,28 +10762,6 @@ static void init_dummy_netdev_core(struct net_device *dev) */ } -/** - * init_dummy_netdev - init a dummy network device for NAPI - * @dev: device to init - * - * This takes a network device structure and initializes the minimum - * amount of fields so it can be used to schedule NAPI polls without - * registering a full blown interface. This is to be used by drivers - * that need to tie several hardware interfaces to a single NAPI - * poll scheduler due to HW limitations. - */ -void init_dummy_netdev(struct net_device *dev) -{ - /* Clear everything. Note we don't initialize spinlocks - * as they aren't supposed to be taken by any of the - * NAPI code and this dummy netdev is supposed to be - * only ever used for NAPI polls - */ - memset(dev, 0, sizeof(struct net_device)); - init_dummy_netdev_core(dev); -} -EXPORT_SYMBOL_GPL(init_dummy_netdev); - /** * register_netdev - register a network device * @dev: device to register diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c index 1b2e7cbb577f..c44c89ecaca6 100644 --- a/net/mptcp/protocol.c +++ b/net/mptcp/protocol.c @@ -47,7 +47,7 @@ static void __mptcp_destroy_sock(struct sock *sk); static void mptcp_check_send_data_fin(struct sock *sk); DEFINE_PER_CPU(struct mptcp_delegated_action, mptcp_delegated_actions); -static struct net_device mptcp_napi_dev; +static struct net_device *mptcp_napi_dev; /* Returns end sequence number of the receiver's advertised window */ static u64 mptcp_wnd_end(const struct mptcp_sock *msk) @@ -4147,11 +4147,13 @@ void __init mptcp_proto_init(void) if (percpu_counter_init(&mptcp_sockets_allocated, 0, GFP_KERNEL)) panic("Failed to allocate MPTCP pcpu counter\n"); - init_dummy_netdev(&mptcp_napi_dev); + mptcp_napi_dev = alloc_netdev_dummy(0); + if (!mptcp_napi_dev) + panic("Failed to allocate MPTCP dummy netdev\n"); for_each_possible_cpu(cpu) { delegated = per_cpu_ptr(&mptcp_delegated_actions, cpu); INIT_LIST_HEAD(&delegated->head); - netif_napi_add_tx(&mptcp_napi_dev, &delegated->napi, + netif_napi_add_tx(mptcp_napi_dev, &delegated->napi, mptcp_napi_poll); napi_enable(&delegated->napi); } diff --git a/net/xfrm/xfrm_input.c b/net/xfrm/xfrm_input.c index 2c4ae61e7e3a..7e6a71b9d6a3 100644 --- a/net/xfrm/xfrm_input.c +++ b/net/xfrm/xfrm_input.c @@ -48,7 +48,7 @@ static DEFINE_SPINLOCK(xfrm_input_afinfo_lock); static struct xfrm_input_afinfo const __rcu *xfrm_input_afinfo[2][AF_INET6 + 1]; static struct gro_cells gro_cells; -static struct net_device xfrm_napi_dev; +static struct net_device *xfrm_napi_dev; static DEFINE_PER_CPU(struct xfrm_trans_tasklet, xfrm_trans_tasklet); @@ -825,8 +825,11 @@ void __init xfrm_input_init(void) int err; int i; - init_dummy_netdev(&xfrm_napi_dev); - err = gro_cells_init(&gro_cells, &xfrm_napi_dev); + xfrm_napi_dev = alloc_netdev_dummy(0); + if (!xfrm_napi_dev) + panic("Failed to allocate XFRM dummy netdev\n"); + + err = gro_cells_init(&gro_cells, xfrm_napi_dev); if (err) gro_cells.cells = NULL;
init_dummy_netdev() can initialize statically declared or embedded net_devices. Such netdevs did not come from alloc_netdev_mqs(). After recent work by Breno, there are the only two cases where we have do that. Switch those cases to alloc_netdev_mqs() and delete init_dummy_netdev(). Dealing with static netdevs is not worth the maintenance burden. Signed-off-by: Jakub Kicinski <kuba@kernel.org> --- v2: change of plan, delete init_dummy_netdev() completely v1: https://lore.kernel.org/20250111065955.3698801-1-kuba@kernel.org --- CC: matttbe@kernel.org CC: martineau@kernel.org CC: geliang@kernel.org CC: steffen.klassert@secunet.com CC: herbert@gondor.apana.org.au CC: mptcp@lists.linux.dev --- include/linux/netdevice.h | 1 - net/core/dev.c | 22 ---------------------- net/mptcp/protocol.c | 8 +++++--- net/xfrm/xfrm_input.c | 9 ++++++--- 4 files changed, 11 insertions(+), 29 deletions(-)