domenico/openwrt_NSS_ACW
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

ipq806x: NSS Hardware Offloading Shortcut SFE Patches

Browse Source
This commit is contained in:
ACwifidude
2021-02-26 18:10:12 -06:00
parent 5eec0bc9c2
commit aba34a3693
3 changed files with 489 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

144
target/linux/ipq806x/patches-5.4/999-01-Revert-ARM-dma-mapping-remove-dmac_clean_range-and-d.patch Normal file
View File

@@ -0,0 +1,144 @@
From 733a75729c1fbb478caaed875dd9c09a878a553d Mon Sep 17 00:00:00 2001
From: Robert Marko <robimarko@gmail.com>
Date: Fri, 5 Jun 2020 11:44:27 +0200
Subject: [PATCH] Revert "ARM: dma-mapping: remove dmac_clean_range and
dmac_inv_range"
This partially reverts 'commit 702b94bff3c505 ("ARM: dma-mapping:
remove dmac_clean_range and dmac_inv_range")'
Some MSM drivers still use the dmac_clean and dmac_inv_range APIs.
Bring back the defines and exports for v7 CPUs.
Signed-off-by: Rohit Vaswani <rvaswani@codeaurora.org>
Signed-off-by: Abhimanyu Kapur <abhimany@codeaurora.org>
[sramana: resolved minor merge conflicts]
Signed-off-by: Srinivas Ramana <sramana@codeaurora.org>
(cherry picked from commit d6118c0a9f7ab2b131ca36dd3dbd5634603d14fe)
Change-Id: Ib2ddb4452711c5c2013bf29f0b5d8a3572b10357
Signed-off-by: Manoharan Vijaya Raghavan <mraghava@codeaurora.org>
Signed-off-by: Robert Marko <robimarko@gmail.com>
---
arch/arm/include/asm/cacheflush.h | 21 +++++++++++++++++++++
arch/arm/include/asm/glue-cache.h | 2 ++
arch/arm/mm/cache-v7.S | 6 ++++--
arch/arm/mm/proc-macros.S | 2 ++
arch/arm/mm/proc-syms.c | 3 +++
5 files changed, 32 insertions(+), 2 deletions(-)
--- a/arch/arm/include/asm/cacheflush.h
+++ b/arch/arm/include/asm/cacheflush.h
@@ -91,6 +91,21 @@
* DMA Cache Coherency
* ===================
*
+ * dma_inv_range(start, end)
+ *
+ * Invalidate (discard) the specified virtual address range.
+ * May not write back any entries. If 'start' or 'end'
+ * are not cache line aligned, those lines must be written
+ * back.
+ * - start - virtual start address
+ * - end - virtual end address
+ *
+ * dma_clean_range(start, end)
+ *
+ * Clean (write back) the specified virtual address range.
+ * - start - virtual start address
+ * - end - virtual end address
+ *
* dma_flush_range(start, end)
*
* Clean and invalidate the specified virtual address range.
@@ -112,6 +127,8 @@ struct cpu_cache_fns {
void (*dma_map_area)(const void *, size_t, int);
void (*dma_unmap_area)(const void *, size_t, int);
+ void (*dma_inv_range)(const void *, const void *);
+ void (*dma_clean_range)(const void *, const void *);
void (*dma_flush_range)(const void *, const void *);
} __no_randomize_layout;
@@ -137,6 +154,8 @@ extern struct cpu_cache_fns cpu_cache;
* is visible to DMA, or data written by DMA to system memory is
* visible to the CPU.
*/
+#define dmac_inv_range cpu_cache.dma_inv_range
+#define dmac_clean_range cpu_cache.dma_clean_range
#define dmac_flush_range cpu_cache.dma_flush_range
#else
@@ -156,6 +175,8 @@ extern void __cpuc_flush_dcache_area(voi
* is visible to DMA, or data written by DMA to system memory is
* visible to the CPU.
*/
+extern void dmac_inv_range(const void *, const void *);
+extern void dmac_clean_range(const void *, const void *);
extern void dmac_flush_range(const void *, const void *);
#endif
--- a/arch/arm/include/asm/glue-cache.h
+++ b/arch/arm/include/asm/glue-cache.h
@@ -156,6 +156,8 @@ static inline void nop_dma_unmap_area(co
#define __cpuc_flush_dcache_area __glue(_CACHE,_flush_kern_dcache_area)
#define dmac_flush_range __glue(_CACHE,_dma_flush_range)
+#define dmac_inv_range __glue(_CACHE, _dma_inv_range)
+#define dmac_clean_range __glue(_CACHE, _dma_clean_range)
#endif
#endif
--- a/arch/arm/mm/cache-v7.S
+++ b/arch/arm/mm/cache-v7.S
@@ -363,7 +363,7 @@ ENDPROC(v7_flush_kern_dcache_area)
* - start - virtual start address of region
* - end - virtual end address of region
*/
-v7_dma_inv_range:
+ENTRY(v7_dma_inv_range)
dcache_line_size r2, r3
sub r3, r2, #1
tst r0, r3
@@ -393,7 +393,7 @@ ENDPROC(v7_dma_inv_range)
* - start - virtual start address of region
* - end - virtual end address of region
*/
-v7_dma_clean_range:
+ENTRY(v7_dma_clean_range)
dcache_line_size r2, r3
sub r3, r2, #1
bic r0, r0, r3
@@ -479,6 +479,8 @@ ENDPROC(v7_dma_unmap_area)
globl_equ b15_dma_map_area, v7_dma_map_area
globl_equ b15_dma_unmap_area, v7_dma_unmap_area
+ globl_equ b15_dma_inv_range, v7_dma_inv_range
+ globl_equ b15_dma_clean_range, v7_dma_clean_range
globl_equ b15_dma_flush_range, v7_dma_flush_range
define_cache_functions b15
--- a/arch/arm/mm/proc-macros.S
+++ b/arch/arm/mm/proc-macros.S
@@ -335,6 +335,8 @@ ENTRY(\name\()_cache_fns)
.long \name\()_flush_kern_dcache_area
.long \name\()_dma_map_area
.long \name\()_dma_unmap_area
+ .long \name\()_dma_inv_range
+ .long \name\()_dma_clean_range
.long \name\()_dma_flush_range
.size \name\()_cache_fns, . - \name\()_cache_fns
.endm
--- a/arch/arm/mm/proc-syms.c
+++ b/arch/arm/mm/proc-syms.c
@@ -27,6 +27,9 @@ EXPORT_SYMBOL(__cpuc_flush_user_all);
EXPORT_SYMBOL(__cpuc_flush_user_range);
EXPORT_SYMBOL(__cpuc_coherent_kern_range);
EXPORT_SYMBOL(__cpuc_flush_dcache_area);
+EXPORT_SYMBOL(dmac_inv_range);
+EXPORT_SYMBOL(dmac_clean_range);
+EXPORT_SYMBOL(dmac_flush_range);
#else
EXPORT_SYMBOL(cpu_cache);
#endif

263
target/linux/ipq806x/patches-5.4/999-01a-nss_sfe-support.patch Normal file
View File

@@ -0,0 +1,263 @@
--- a/include/linux/timer.h
+++ b/include/linux/timer.h
@@ -17,6 +17,7 @@ struct timer_list {
unsigned long expires;
void (*function)(struct timer_list *);
u32 flags;
+ unsigned long cust_data;
#ifdef CONFIG_LOCKDEP
struct lockdep_map lockdep_map;
--- a/include/linux/if_bridge.h
+++ b/include/linux/if_bridge.h
@@ -51,6 +51,8 @@
#define BR_DEFAULT_AGEING_TIME (300 * HZ)
extern void brioctl_set(int (*ioctl_hook)(struct net *, unsigned int, void __user *));
+extern void br_dev_update_stats(struct net_device *dev,
+ struct rtnl_link_stats64 *nlstats);
#if IS_ENABLED(CONFIG_BRIDGE) && IS_ENABLED(CONFIG_BRIDGE_IGMP_SNOOPING)
int br_multicast_list_adjacent(struct net_device *dev,
--- a/include/net/netns/conntrack.h
+++ b/include/net/netns/conntrack.h
@@ -112,6 +112,9 @@ struct netns_ct {
struct ct_pcpu __percpu *pcpu_lists;
struct ip_conntrack_stat __percpu *stat;
+#ifdef CONFIG_NF_CONNTRACK_CHAIN_EVENTS
+ struct atomic_notifier_head nf_conntrack_chain;
+#endif
struct nf_ct_event_notifier __rcu *nf_conntrack_event_cb;
struct nf_exp_event_notifier __rcu *nf_expect_event_cb;
struct nf_ip_net nf_ct_proto;
--- a/include/net/netfilter/nf_conntrack_ecache.h
+++ b/include/net/netfilter/nf_conntrack_ecache.h
@@ -72,6 +72,11 @@ struct nf_ct_event {
int report;
};
+#ifdef CONFIG_NF_CONNTRACK_CHAIN_EVENTS
+extern int nf_conntrack_register_chain_notifier(struct net *net, struct notifier_block *nb);
+extern int nf_conntrack_unregister_chain_notifier(struct net *net, struct notifier_block *nb);
+#endif
+
struct nf_ct_event_notifier {
int (*fcn)(unsigned int events, struct nf_ct_event *item);
};
@@ -105,11 +110,13 @@ static inline void
nf_conntrack_event_cache(enum ip_conntrack_events event, struct nf_conn *ct)
{
#ifdef CONFIG_NF_CONNTRACK_EVENTS
- struct net *net = nf_ct_net(ct);
struct nf_conntrack_ecache *e;
+#ifndef CONFIG_NF_CONNTRACK_CHAIN_EVENTS
+ struct net *net = nf_ct_net(ct);
if (!rcu_access_pointer(net->ct.nf_conntrack_event_cb))
return;
+#endif
e = nf_ct_ecache_find(ct);
if (e == NULL)
@@ -124,10 +131,12 @@ nf_conntrack_event_report(enum ip_conntr
u32 portid, int report)
{
#ifdef CONFIG_NF_CONNTRACK_EVENTS
+#ifndef CONFIG_NF_CONNTRACK_CHAIN_EVENTS
const struct net *net = nf_ct_net(ct);
if (!rcu_access_pointer(net->ct.nf_conntrack_event_cb))
return 0;
+#endif
return nf_conntrack_eventmask_report(1 << event, ct, portid, report);
#else
@@ -139,10 +148,12 @@ static inline int
nf_conntrack_event(enum ip_conntrack_events event, struct nf_conn *ct)
{
#ifdef CONFIG_NF_CONNTRACK_EVENTS
+#ifndef CONFIG_NF_CONNTRACK_CHAIN_EVENTS
const struct net *net = nf_ct_net(ct);
if (!rcu_access_pointer(net->ct.nf_conntrack_event_cb))
return 0;
+#endif
return nf_conntrack_eventmask_report(1 << event, ct, 0, 0);
#else
--- a/net/netfilter/nf_conntrack_proto_tcp.c
+++ b/net/netfilter/nf_conntrack_proto_tcp.c
@@ -33,11 +33,13 @@
/* Do not check the TCP window for incoming packets */
-static int nf_ct_tcp_no_window_check __read_mostly = 1;
+int nf_ct_tcp_no_window_check __read_mostly = 1;
+EXPORT_SYMBOL_GPL(nf_ct_tcp_no_window_check);
/* "Be conservative in what you do,
be liberal in what you accept from others."
If it's non-zero, we mark only out of window RST segments as INVALID. */
-static int nf_ct_tcp_be_liberal __read_mostly = 0;
+int nf_ct_tcp_be_liberal __read_mostly = 0;
+EXPORT_SYMBOL_GPL(nf_ct_tcp_be_liberal);
/* If it is set to zero, we disable picking up already established
connections. */
--- a/net/bridge/br_if.c
+++ b/net/bridge/br_if.c
@@ -757,3 +757,26 @@
return p->flags & flag;
}
EXPORT_SYMBOL_GPL(br_port_flag_is_set);
+
+/* Update bridge statistics for bridge packets processed by offload engines */
+void br_dev_update_stats(struct net_device *dev,
+ struct rtnl_link_stats64 *nlstats)
+{
+ struct net_bridge *br;
+ struct pcpu_sw_netstats *stats;
+
+ /* Is this a bridge? */
+ if (!(dev->priv_flags & IFF_EBRIDGE))
+ return;
+
+ br = netdev_priv(dev);
+ stats = this_cpu_ptr(br->stats);
+
+ u64_stats_update_begin(&stats->syncp);
+ stats->rx_packets += nlstats->rx_packets;
+ stats->rx_bytes += nlstats->rx_bytes;
+ stats->tx_packets += nlstats->tx_packets;
+ stats->tx_bytes += nlstats->tx_bytes;
+ u64_stats_update_end(&stats->syncp);
+}
+EXPORT_SYMBOL_GPL(br_dev_update_stats);
--- a/net/netfilter/Kconfig
+++ b/net/netfilter/Kconfig
@@ -158,6 +158,14 @@
If unsure, say `N'.
+config NF_CONNTRACK_CHAIN_EVENTS
+ bool "Register multiple callbacks to ct events"
+ depends on NF_CONNTRACK_EVENTS
+ help
+ Support multiple registrations.
+
+ If unsure, say `N'.
+
config NF_CONNTRACK_TIMESTAMP
bool 'Connection tracking timestamping'
depends on NETFILTER_ADVANCED
--- a/net/netfilter/nf_conntrack_core.c
+++ b/net/netfilter/nf_conntrack_core.c
@@ -2588,6 +2588,9 @@ int nf_conntrack_init_net(struct net *ne
nf_conntrack_ecache_pernet_init(net);
nf_conntrack_helper_pernet_init(net);
nf_conntrack_proto_pernet_init(net);
+#ifdef CONFIG_NF_CONNTRACK_CHAIN_EVENTS
+ ATOMIC_INIT_NOTIFIER_HEAD(&net->ct.nf_conntrack_chain);
+#endif
return 0;
--- a/net/netfilter/nf_conntrack_ecache.c
+++ b/net/netfilter/nf_conntrack_ecache.c
@@ -17,6 +17,9 @@
#include <linux/stddef.h>
#include <linux/err.h>
#include <linux/percpu.h>
+#ifdef CONFIG_NF_CONNTRACK_CHAIN_EVENTS
+#include <linux/notifier.h>
+#endif
#include <linux/kernel.h>
#include <linux/netdevice.h>
#include <linux/slab.h>
@@ -127,7 +130,11 @@ int nf_conntrack_eventmask_report(unsign
rcu_read_lock();
notify = rcu_dereference(net->ct.nf_conntrack_event_cb);
+#ifdef CONFIG_NF_CONNTRACK_CHAIN_EVENTS
+ if (!notify && !rcu_dereference_raw(net->ct.nf_conntrack_chain.head))
+#else
if (!notify)
+#endif
goto out_unlock;
e = nf_ct_ecache_find(ct);
@@ -146,7 +153,15 @@ int nf_conntrack_eventmask_report(unsign
if (!((eventmask | missed) & e->ctmask))
goto out_unlock;
+#ifdef CONFIG_NF_CONNTRACK_CHAIN_EVENTS
+ ret = atomic_notifier_call_chain(&net->ct.nf_conntrack_chain,
+ eventmask | missed, &item);
+
+ if (notify)
+ ret = notify->fcn(eventmask | missed, &item);
+#else
ret = notify->fcn(eventmask | missed, &item);
+#endif
if (unlikely(ret < 0 || missed)) {
spin_lock_bh(&ct->lock);
if (ret < 0) {
@@ -186,7 +201,11 @@ void nf_ct_deliver_cached_events(struct
rcu_read_lock();
notify = rcu_dereference(net->ct.nf_conntrack_event_cb);
+#ifdef CONFIG_NF_CONNTRACK_CHAIN_EVENTS
+ if ((notify == NULL) && !rcu_dereference_raw(net->ct.nf_conntrack_chain.head))
+#else
if (notify == NULL)
+#endif
goto out_unlock;
e = nf_ct_ecache_find(ct);
@@ -210,7 +229,16 @@ void nf_ct_deliver_cached_events(struct
item.portid = 0;
item.report = 0;
+#ifdef CONFIG_NF_CONNTRACK_CHAIN_EVENTS
+ ret = atomic_notifier_call_chain(&net->ct.nf_conntrack_chain,
+ events | missed,
+ &item);
+
+ if (notify != NULL)
+ ret = notify->fcn(events | missed, &item);
+#else
ret = notify->fcn(events | missed, &item);
+#endif
if (likely(ret == 0 && !missed))
goto out_unlock;
@@ -257,6 +285,14 @@ out_unlock:
rcu_read_unlock();
}
+#ifdef CONFIG_NF_CONNTRACK_CHAIN_EVENTS
+int nf_conntrack_register_chain_notifier(struct net *net, struct notifier_block *nb)
+{
+ return atomic_notifier_chain_register(&net->ct.nf_conntrack_chain, nb);
+}
+EXPORT_SYMBOL_GPL(nf_conntrack_register_chain_notifier);
+#endif
+
int nf_conntrack_register_notifier(struct net *net,
struct nf_ct_event_notifier *new)
{
@@ -279,6 +315,14 @@ out_unlock:
}
EXPORT_SYMBOL_GPL(nf_conntrack_register_notifier);
+#ifdef CONFIG_NF_CONNTRACK_CHAIN_EVENTS
+int nf_conntrack_unregister_chain_notifier(struct net *net, struct notifier_block *nb)
+{
+ return atomic_notifier_chain_unregister(&net->ct.nf_conntrack_chain, nb);
+}
+EXPORT_SYMBOL_GPL(nf_conntrack_unregister_chain_notifier);
+#endif
+
void nf_conntrack_unregister_notifier(struct net *net,
struct nf_ct_event_notifier *new)
{

82
target/linux/ipq806x/patches-5.4/999-01b-shortcut-fe-support.patch Normal file
View File

@@ -0,0 +1,82 @@
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -825,6 +825,9 @@
__u8 decrypted:1;
#endif
__u8 gro_skip:1;
+#ifdef CONFIG_SHORTCUT_FE
+ __u8 fast_forwarded:1;
+#endif
#ifdef CONFIG_NET_SCHED
__u16 tc_index; /* traffic control index */
#if IS_ENABLED(CONFIG_BRIDGE) && IS_ENABLED(CONFIG_BRIDGE_IGMP_SNOOPING)
int br_multicast_list_adjacent(struct net_device *dev,
--- a/net/Kconfig
+++ b/net/Kconfig
@@ -473,3 +473,6 @@ config HAVE_CBPF_JIT
# Extended BPF JIT (eBPF)
config HAVE_EBPF_JIT
bool
+
+config SHORTCUT_FE
+ bool "Enables kernel network stack path for Shortcut Forwarding Engine
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -3192,8 +3192,17 @@ static int xmit_one(struct sk_buff *skb,
unsigned int len;
int rc;
+#ifdef CONFIG_SHORTCUT_FE
+ /* If this skb has been fast forwarded then we don't want it to
+ * go to any taps (by definition we're trying to bypass them).
+ */
+ if (!skb->fast_forwarded) {
+#endif
if (dev_nit_active(dev))
dev_queue_xmit_nit(skb, dev);
+#ifdef CONFIG_SHORTCUT_FE
+ }
+#endif
#ifdef CONFIG_ETHERNET_PACKET_MANGLE
if (!dev->eth_mangle_tx ||
@@ -4684,6 +4693,11 @@ void netdev_rx_handler_unregister(struct
}
EXPORT_SYMBOL_GPL(netdev_rx_handler_unregister);
+#ifdef CONFIG_SHORTCUT_FE
+int (*athrs_fast_nat_recv)(struct sk_buff *skb) __rcu __read_mostly;
+EXPORT_SYMBOL_GPL(athrs_fast_nat_recv);
+#endif
+
/*
* Limit the use of PFMEMALLOC reserves to those protocols that implement
* the special handling of PFMEMALLOC skbs.
@@ -4733,6 +4747,9 @@ static int __netif_receive_skb_core(stru
bool deliver_exact = false;
int ret = NET_RX_DROP;
__be16 type;
+#ifdef CONFIG_SHORTCUT_FE
+ int (*fast_recv)(struct sk_buff *skb);
+#endif
net_timestamp_check(!netdev_tstamp_prequeue, skb);
@@ -4773,6 +4790,16 @@ another_round:
goto out;
}
+#ifdef CONFIG_SHORTCUT_FE
+ fast_recv = rcu_dereference(athrs_fast_nat_recv);
+ if (fast_recv) {
+ if (fast_recv(skb)) {
+ ret = NET_RX_SUCCESS;
+ goto out;
+ }
+ }
+#endif
+
if (skb_skip_tc_classify(skb))
goto skip_classify;