Netdev List
 help / color / mirror / Atom feed
* Thoughts on staging and on fixing up drivers?
From: Dan Carpenter @ 2017-08-17 20:46 UTC (permalink / raw)
  To: Larry Finger
  Cc: devel, Ping-Ke Shih, Yan-Hsuan Chuang, gregkh, Shaofu,
	Birming Chiu, netdev, Steven Ting
In-Reply-To: <20170817174652.17656-1-Larry.Finger@lwfinger.net>

I'm excited to see all the @realtek.com addresses.  :)

Larry, you've migrated a bunch of staging code, and tried various
approaches.  Do you have any lessons on what has worked and what hasn't
and if there is anything we can do to make the process better?

regards,
dan carpenter

^ permalink raw reply

* [PATCH iproute2] libnetlink: Fix extack attribute parsing
From: David Ahern @ 2017-08-17 20:43 UTC (permalink / raw)
  To: netdev, stephen; +Cc: David Ahern

Initialize tb in nl_dump_ext_err since not all attributes will be
sent in the messages.

Add error checking on mnl_attr_parse and print messages on the off
chance the ext ack attributes fail to validate.

Signed-off-by: David Ahern <dsahern@gmail.com>
---
 lib/libnetlink.c | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/lib/libnetlink.c b/lib/libnetlink.c
index 81a344abff27..874e660be7eb 100644
--- a/lib/libnetlink.c
+++ b/lib/libnetlink.c
@@ -49,13 +49,17 @@ static int err_attr_cb(const struct nlattr *attr, void *data)
 	const struct nlattr **tb = data;
 	uint16_t type;
 
-	if (mnl_attr_type_valid(attr, NLMSGERR_ATTR_MAX) < 0)
+	if (mnl_attr_type_valid(attr, NLMSGERR_ATTR_MAX) < 0) {
+		fprintf(stderr, "Invalid extack attribute\n");
 		return MNL_CB_ERROR;
+	}
 
 	type = mnl_attr_get_type(attr);
-	if (mnl_attr_validate(attr, extack_policy[type]) < 0)
+	if (mnl_attr_validate(attr, extack_policy[type]) < 0) {
+		fprintf(stderr, "extack attribute %d failed validation\n",
+			type);
 		return MNL_CB_ERROR;
-
+	}
 
 	tb[type] = attr;
 	return MNL_CB_OK;
@@ -64,7 +68,7 @@ static int err_attr_cb(const struct nlattr *attr, void *data)
 /* dump netlink extended ack error message */
 static int nl_dump_ext_err(const struct nlmsghdr *nlh, nl_ext_ack_fn_t errfn)
 {
-	struct nlattr *tb[NLMSGERR_ATTR_MAX + 1];
+	struct nlattr *tb[NLMSGERR_ATTR_MAX + 1] = {};
 	const struct nlmsgerr *err = mnl_nlmsg_get_payload(nlh);
 	const struct nlmsghdr *err_nlh = NULL;
 	unsigned int hlen = sizeof(*err);
@@ -79,7 +83,8 @@ static int nl_dump_ext_err(const struct nlmsghdr *nlh, nl_ext_ack_fn_t errfn)
 	if (!(nlh->nlmsg_flags & NLM_F_CAPPED))
 		hlen += mnl_nlmsg_get_payload_len(&err->msg);
 
-	mnl_attr_parse(nlh, hlen, err_attr_cb, tb);
+	if (mnl_attr_parse(nlh, hlen, err_attr_cb, tb) != MNL_CB_OK)
+		return 0;
 
 	if (tb[NLMSGERR_ATTR_MSG])
 		errmsg = mnl_attr_get_str(tb[NLMSGERR_ATTR_MSG]);
-- 
2.1.4

^ permalink raw reply related

* Re: [iproute PATCH v2 1/7] devlink: No need for this self-assignment
From: Jiri Pirko @ 2017-08-17 19:48 UTC (permalink / raw)
  To: Phil Sutter; +Cc: Stephen Hemminger, netdev
In-Reply-To: <20170817170931.24089-2-phil@nwl.cc>

Thu, Aug 17, 2017 at 07:09:25PM CEST, phil@nwl.cc wrote:
>dl_argv_handle_both() will either assign to handle_bit or error out in
>which case the variable is not used by the caller.

I'm pretty sure that I did this to silence the compiler. If the compiler
bug is fixed now, good.

Acked-by: Jiri Pirko <jiri@mellanox.com>

^ permalink raw reply

* Re: [PATCH 2/2] xdp: adjust xdp redirect tracepoint to include return error code
From: John Fastabend @ 2017-08-17 19:43 UTC (permalink / raw)
  To: Jesper Dangaard Brouer; +Cc: netdev
In-Reply-To: <20170817212857.1de3f557@redhat.com>

On 08/17/2017 12:28 PM, Jesper Dangaard Brouer wrote:
> 
> On Thu, 17 Aug 2017 11:46:10 -0700 John Fastabend <john.fastabend@gmail.com> wrote:
> 
>> On 08/17/2017 09:22 AM, Jesper Dangaard Brouer wrote:
>>> The return error code need to be included in the tracepoint
>>> xdp:xdp_redirect, else its not possible to distinguish successful or
>>> failed XDP_REDIRECT transmits.
>>>
>>> XDP have no queuing mechanism. Thus, it is fairly easily to overrun a
>>> NIC transmit queue.  The eBPF program invoking helpers (bpf_redirect
>>> or bpf_redirect_map) to redirect a packet doesn't get any feedback
>>> whether the packet was actually transmitted.
>>>
>>> Info on failed transmits in the tracepoint xdp:xdp_redirect, is
>>> interesting as this opens for providing a feedback-loop to the
>>> receiving XDP program.
>>>
>>> Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
>>> ---  
>>
>> [...]
>>
>>> @@ -2532,12 +2535,14 @@ int xdp_do_redirect(struct net_device *dev, struct xdp_buff *xdp,
>>>  	ri->map = NULL;
>>>  	if (unlikely(!fwd)) {
>>>  		bpf_warn_invalid_xdp_redirect(index);

I think we should drop the warn_invalid now that we have a tracepoint.
The tracepoint is much nicer for debugging vs a warning for what might
be a valid case depending on xdp program.

>>> -		return -EINVAL;
>>> +		err = -EINVAL;
>>> +		goto out;  
>>
>> It doesn't look like there is a check in trace_xdp_redirect to
>> avoid dereferencing a NULL fwd pointer here (*to in trace code
>> path). Did I miss something?
> 
> Nice that you spotted this in your review, but the __string() macro
> used in trace code already takes case of this, see output:
> 
>  xdp:xdp_redirect: prog=39cf08f65683838a from=ixgbe2 to=(null) action=REDIRECT err=-22

Great thanks.

^ permalink raw reply

* Re: [PATCH V7 net-next 00/22] Huawei HiNIC Ethernet Driver
From: David Miller @ 2017-08-17 19:33 UTC (permalink / raw)
  To: aviad.krawczyk
  Cc: linux-kernel, netdev, bc.y, victor.gissin, zhaochen6, tony.qu
In-Reply-To: <cover.1502996306.git.aviad.krawczyk@huawei.com>


You've posted this series 3 times today.

That's way too fast.

You must wait at least one full day for more feedback to come
your way.

If you just repsin your series for every little small change nobody is
going to perform a thorough review of your patches because you keep
respining them so often that people feel like their review efforts are
going to be wasted.

You must be patient.  I can see that the reason you are respinning so
often if because you want your patches integrated more quickly.

But what you are doing is having the opposite effect.  It is draining
on people and will make your changes take longer to be integrated.

^ permalink raw reply

* Re: [PATCH 2/2] xdp: adjust xdp redirect tracepoint to include return error code
From: Jesper Dangaard Brouer @ 2017-08-17 19:28 UTC (permalink / raw)
  To: John Fastabend; +Cc: netdev, brouer
In-Reply-To: <5995E472.70609@gmail.com>


On Thu, 17 Aug 2017 11:46:10 -0700 John Fastabend <john.fastabend@gmail.com> wrote:

> On 08/17/2017 09:22 AM, Jesper Dangaard Brouer wrote:
> > The return error code need to be included in the tracepoint
> > xdp:xdp_redirect, else its not possible to distinguish successful or
> > failed XDP_REDIRECT transmits.
> > 
> > XDP have no queuing mechanism. Thus, it is fairly easily to overrun a
> > NIC transmit queue.  The eBPF program invoking helpers (bpf_redirect
> > or bpf_redirect_map) to redirect a packet doesn't get any feedback
> > whether the packet was actually transmitted.
> > 
> > Info on failed transmits in the tracepoint xdp:xdp_redirect, is
> > interesting as this opens for providing a feedback-loop to the
> > receiving XDP program.
> > 
> > Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
> > ---  
> 
> [...]
> 
> > @@ -2532,12 +2535,14 @@ int xdp_do_redirect(struct net_device *dev, struct xdp_buff *xdp,
> >  	ri->map = NULL;
> >  	if (unlikely(!fwd)) {
> >  		bpf_warn_invalid_xdp_redirect(index);
> > -		return -EINVAL;
> > +		err = -EINVAL;
> > +		goto out;  
> 
> It doesn't look like there is a check in trace_xdp_redirect to
> avoid dereferencing a NULL fwd pointer here (*to in trace code
> path). Did I miss something?

Nice that you spotted this in your review, but the __string() macro
used in trace code already takes case of this, see output:

 xdp:xdp_redirect: prog=39cf08f65683838a from=ixgbe2 to=(null) action=REDIRECT err=-22
 
> >  	}
> >  
> > -	trace_xdp_redirect(dev, fwd, xdp_prog, XDP_REDIRECT);
> > -
> > -	return __bpf_tx_xdp(fwd, NULL, xdp, 0);
> > +	err = __bpf_tx_xdp(fwd, NULL, xdp, 0);
> > +out:
> > +	trace_xdp_redirect(dev, fwd, xdp_prog, XDP_REDIRECT, err);
> > +	return err;
> >  }
> >  EXPORT_SYMBOL_GPL(xdp_do_redirect);

-- 
Best regards,
  Jesper Dangaard Brouer
  MSc.CS, Principal Kernel Engineer at Red Hat
  LinkedIn: http://www.linkedin.com/in/brouer

^ permalink raw reply

* skb_over_panic when sending esp traffic from a vmware guest
From: Nick Huber @ 2017-08-17 19:19 UTC (permalink / raw)
  To: netdev@vger.kernel.org

I've been experience the following traceback since upgrading from the 4.9 kernel to the 4.11 branch. I've only seen this in a few VMWare guests and I haven't been able to narrow down what exactly is causing it. I'm not familiar with kernel debugging but any help in tracing this down would be appreciated.

skbuff: skb_over_panic: text:ffffffff8a53e2a7 len:464 put:4 head:ffff904d7682b400 data:ffff904d7682b4f4 tail:0x2c4 end:0x2c0 dev:chwk-pwr186
------------[ cut here ]------------
kernel BUG at net/core/skbuff.c:105!
invalid opcode: 0000 [#1] SMP
Modules linked in: sch_sfq xt_TEE nf_dup_ipv6 nf_dup_ipv4 xt_set ip_set_hash_net ip_set nfnetlink xt_socket nf_socket_ipv4 nf_socket_ipv6 xt_REDIRECT nf_nat_redirect sch_htb sit tunnel4 veth drbg ansi_cprng authenc seqiv esp4 xfrm4_mode_transport binfmt_misc xt_connbytes xt_hashlimit xt_dscp xt_CLASSIFY xt_length xt_TCPMSS iptable_raw xt_nat xt_connmark xt_mark iptable_nat nf_nat_ipv4 iptable_mangle netconsole ip6t_REJECT nf_reject_ipv6 ipt_REJECT nf_reject_ipv4 nf_conntrack_ipv6 nf_defrag_ipv6 nf_conntrack_ipv4 nf_defrag_ipv4 xt_conntrack xt_comment ip6table_filter ip6_tables xt_multiport xt_tcpudp arptable_filter arp_tables iptable_filter ip_tables x_tables configfs mptctl nfsd auth_rpcgss nfs_acl nfs lockd grace fscache sunrpc sb_edac edac_core coretemp crct10dif_pclmul crc32_pclmul
 ghash_clmulni_intel deflate pcbc evdev ppdev snd_pcm snd_timer vmw_balloon aesni_intel snd serio_raw soundcore pcspkr ctr aes_x86_64 crypto_simd twofish_generic twofish_avx_x86_64 twofish_x86_64_3way twofish_x86_64 twofish_common sg camellia_generic battery camellia_aesni_avx_x86_64 camellia_x86_64 vmwgfx parport_pc ttm parport serpent_avx_x86_64 shpchp drm_kms_helper drm vmw_vmci serpent_sse2_x86_64 xts serpent_generic lrw gf128mul glue_helper ac button blowfish_generic blowfish_x86_64 blowfish_common cast5_avx_x86_64 cast5_generic cast_common ablk_helper cryptd des_generic cbc cmac xcbc rmd160 sha512_ssse3 sha512_generic af_key xfrm_algo loop nf_nat_pptp nf_nat_proto_gre nf_nat_ftp nf_nat nf_conntrack_pptp nf_conntrack_proto_gre nf_conntrack_ftp nf_conntrack libcrc32c crc32c_generic
 ip_gre ip_tunnel gre 8021q garp mrp stp llc tun autofs4 ext4 crc16 jbd2 fscrypto mbcache sr_mod cdrom sd_mod ata_generic crc32c_intel psmouse ata_piix mptspi scsi_transport_spi mptscsih vmxnet3 mptbase i2c_piix4 libata scsi_mod floppy [last unloaded: netconsole]
CPU: 3 PID: 2365 Comm: bridge Not tainted 4.11.12 #2
Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 04/14/2014
task: ffff904d76efa040 task.stack: ffffb92fc0d60000
RIP: 0010:skb_panic+0x51/0x60
RSP: 0018:ffffb92fc0d63658 EFLAGS: 00010292
RAX: 000000000000008c RBX: ffff904d7a331800 RCX: 0000000000000000
RDX: ffff904d7fd95e28 RSI: ffff904d7fd8df88 RDI: ffff904d7fd8df88
RBP: ffffb92fc0d63678 R08: 0000000135219082 R09: 0000000000000585
R10: 0000000000000004 R11: 3039666666663a64 R12: 0000000000000001
R13: ffff904d786f8480 R14: ffff904d75b87a00 R15: ffff904d7682b4e4
FS:  00007fc0a42c1700(0000) GS:ffff904d7fd80000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f5173a8a8b0 CR3: 0000000136471000 CR4: 00000000000406e0
Call Trace:
 skb_put+0x4d/0x50
 pskb_put+0x27/0x30
 esp_output+0x1e7/0xace [esp4]
 ? ip_finish_output+0x180/0x280
 xfrm_output_resume+0x13c/0x390
 ? xfrm_output+0x82/0xe0
 ? skb_release_all+0x24/0x30
 xfrm_output+0xb3/0xe0
 xfrm4_output_finish+0x2b/0x30
 __xfrm4_output+0x34/0x40
 xfrm4_output+0x40/0xa0
 ? xfrm4_udp_encap_rcv+0x1b0/0x1b0
 ip_local_out+0x35/0x40
 iptunnel_xmit+0x13d/0x1b0
 ip_tunnel_xmit+0x388/0x14b0 [ip_tunnel]
 __gre_xmit+0x174/0x210 [ip_gre]
 ipgre_xmit+0x1d4/0x220 [ip_gre]
 dev_hard_start_xmit+0xa1/0x200
 __dev_queue_xmit+0x595/0x670
 dev_queue_xmit+0x10/0x20
 neigh_direct_output+0x11/0x20
 ip_finish_output2+0x16b/0x360
 ip_finish_output+0x180/0x280
 ip_output+0x67/0xd0
 ? ip_fragment.constprop.49+0x80/0x80
 ip_local_out+0x35/0x40
 ip_queue_xmit+0x158/0x3e0
 ? tcp_v4_md5_lookup+0x13/0x20
 tcp_transmit_skb+0x56d/0x950
 tcp_write_xmit+0x191/0xf40
 __tcp_push_pending_frames+0x31/0xd0
 tcp_push+0xef/0x120
 tcp_sendmsg+0x432/0xbf0
 inet_sendmsg+0x31/0xb0
 sock_sendmsg+0x38/0x50
 SYSC_sendto+0xef/0x170
 SyS_sendto+0xe/0x10
 entry_SYSCALL_64_fastpath+0x1e/0xad
RIP: 0033:0x7fc0a3967ead
RSP: 002b:00007fff42176488 EFLAGS: 00000246 ORIG_RAX: 000000000000002c
RAX: ffffffffffffffda RBX: 00007fc09c2fdc48 RCX: 00007fc0a3967ead
RDX: 000000000000070a RSI: 0000000002452bd0 RDI: 0000000000000048
RBP: 00007fc09c3948b8 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 00007fc09f01e5c0
R13: 0000000000000000 R14: 00007fc09c3948b8 R15: 00007fc0a3c03b40
Code: 00 50 8b 87 cc 00 00 00 50 8b 87 c8 00 00 00 50 ff b7 d8 00 00 00 31 c0 4c 8b 8f d0 00 00 00 48 c7 c7 e8 b0 a3 8a e8 d5 97 c5 ff <0f> 0b 66 66 66 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 48 
RIP: skb_panic+0x51/0x60 RSP: ffffb92fc0d63658

^ permalink raw reply

* [PATCH v2 net-next] net: ipv6: put host and anycast routes on device with address
From: David Ahern @ 2017-08-17 19:17 UTC (permalink / raw)
  To: netdev; +Cc: yoshfuji, hannes, David Ahern

One nagging difference between ipv4 and ipv6 is host routes for ipv6
addresses are installed using the loopback device or VRF / L3 Master
device. e.g.,

    2001:db8:1::/120 dev veth0 proto kernel metric 256 pref medium
    local 2001:db8:1::1 dev lo table local proto kernel metric 0 pref medium

Using the loopback device is convenient -- necessary for local tx, but
has some nasty side effects, most notably setting the 'lo' device down
causes all host routes for all local IPv6 address to be removed from the
FIB and completely breaks IPv6 networking across all interfaces.

This patch puts FIB entries for IPv6 routes against the device. This
simplifies the routes in the FIB, for example by making dst->dev and
rt6i_idev->dev the same (a future patch can look at removing the device
reference taken for rt6i_idev for FIB entries).

When copies are made on FIB lookups, the cloned route has dst->dev
set to loopback (or the L3 master device). This is needed for the
local Tx of packets to local addresses.

With fib entries allocated against the real network device, the addrconf
code that reinserts host routes on admin up of 'lo' is no longer needed.

Signed-off-by: David Ahern <dsahern@gmail.com>
---
v2
- update icmp6_send to use rt6i_idev for linklocal addresses if
  skb dev is loopback. ICMP unreachables were hitting the wrong
  route in response to UDP messages to local linklocal addresses
  with no server listening

 net/ipv6/addrconf.c | 42 ------------------------------------------
 net/ipv6/icmp.c     | 15 +++++++++++++--
 net/ipv6/route.c    | 46 ++++++++++++++++++++++++++++++++++------------
 3 files changed, 47 insertions(+), 56 deletions(-)

diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
index 640792e1ecb7..45d0a24644de 100644
--- a/net/ipv6/addrconf.c
+++ b/net/ipv6/addrconf.c
@@ -3030,9 +3030,6 @@ static void sit_add_v4_addrs(struct inet6_dev *idev)
 static void init_loopback(struct net_device *dev)
 {
 	struct inet6_dev  *idev;
-	struct net_device *sp_dev;
-	struct inet6_ifaddr *sp_ifa;
-	struct rt6_info *sp_rt;
 
 	/* ::1 */
 
@@ -3045,45 +3042,6 @@ static void init_loopback(struct net_device *dev)
 	}
 
 	add_addr(idev, &in6addr_loopback, 128, IFA_HOST);
-
-	/* Add routes to other interface's IPv6 addresses */
-	for_each_netdev(dev_net(dev), sp_dev) {
-		if (!strcmp(sp_dev->name, dev->name))
-			continue;
-
-		idev = __in6_dev_get(sp_dev);
-		if (!idev)
-			continue;
-
-		read_lock_bh(&idev->lock);
-		list_for_each_entry(sp_ifa, &idev->addr_list, if_list) {
-
-			if (sp_ifa->flags & (IFA_F_DADFAILED | IFA_F_TENTATIVE))
-				continue;
-
-			if (sp_ifa->rt) {
-				/* This dst has been added to garbage list when
-				 * lo device down, release this obsolete dst and
-				 * reallocate a new router for ifa.
-				 */
-				if (!sp_ifa->rt->rt6i_node) {
-					ip6_rt_put(sp_ifa->rt);
-					sp_ifa->rt = NULL;
-				} else {
-					continue;
-				}
-			}
-
-			sp_rt = addrconf_dst_alloc(idev, &sp_ifa->addr, false);
-
-			/* Failure cases are ignored */
-			if (!IS_ERR(sp_rt)) {
-				sp_ifa->rt = sp_rt;
-				ip6_ins_rt(sp_rt);
-			}
-		}
-		read_unlock_bh(&idev->lock);
-	}
 }
 
 void addrconf_add_linklocal(struct inet6_dev *idev,
diff --git a/net/ipv6/icmp.c b/net/ipv6/icmp.c
index 8d7b113958b1..4f82830fc068 100644
--- a/net/ipv6/icmp.c
+++ b/net/ipv6/icmp.c
@@ -459,9 +459,20 @@ static void icmp6_send(struct sk_buff *skb, u8 type, u8 code, __u32 info,
 	 *	Source addr check
 	 */
 
-	if (__ipv6_addr_needs_scope_id(addr_type))
+	if (__ipv6_addr_needs_scope_id(addr_type)) {
 		iif = skb->dev->ifindex;
-	else {
+
+		/* for local packets, get the real device index */
+		if (iif == LOOPBACK_IFINDEX) {
+			dst = skb_dst(skb);
+			if (dst) {
+				struct rt6_info *rt;
+
+				rt = container_of(dst, struct rt6_info, dst);
+				iif = rt->rt6i_idev->dev->ifindex;
+			}
+		}
+	} else {
 		dst = skb_dst(skb);
 		iif = l3mdev_master_ifindex(dst ? dst->dev : skb->dev);
 	}
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index dc021ed6dd37..1864effcaf00 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -958,10 +958,34 @@ int ip6_ins_rt(struct rt6_info *rt)
 	return __ip6_ins_rt(rt, &info, &mxc, NULL);
 }
 
+/* called with rcu_lock held */
+static struct net_device *ip6_rt_get_dev_rcu(struct rt6_info *rt)
+{
+	struct net_device *dev = rt->dst.dev;
+
+	if (rt->rt6i_flags & RTF_LOCAL) {
+		/* for copies of local routes, dst->dev needs to be the
+		 * device if it is a master device, the master device if
+		 * device is enslaved, and the loopback as the default
+		 */
+		if (netif_is_l3_slave(dev) &&
+		    !rt6_need_strict(&rt->rt6i_dst.addr))
+			dev = l3mdev_master_dev_rcu(dev);
+		else if (!netif_is_l3_master(dev))
+			dev = dev_net(dev)->loopback_dev;
+		/* last case is netif_is_l3_master(dev) is true in which
+		 * case we want dev returned to be dev
+		 */
+	}
+
+	return dev;
+}
+
 static struct rt6_info *ip6_rt_cache_alloc(struct rt6_info *ort,
 					   const struct in6_addr *daddr,
 					   const struct in6_addr *saddr)
 {
+	struct net_device *dev;
 	struct rt6_info *rt;
 
 	/*
@@ -971,8 +995,10 @@ static struct rt6_info *ip6_rt_cache_alloc(struct rt6_info *ort,
 	if (ort->rt6i_flags & (RTF_CACHE | RTF_PCPU))
 		ort = (struct rt6_info *)ort->dst.from;
 
-	rt = __ip6_dst_alloc(dev_net(ort->dst.dev), ort->dst.dev, 0);
-
+	rcu_read_lock();
+	dev = ip6_rt_get_dev_rcu(ort);
+	rt = __ip6_dst_alloc(dev_net(dev), dev, 0);
+	rcu_read_unlock();
 	if (!rt)
 		return NULL;
 
@@ -1000,11 +1026,13 @@ static struct rt6_info *ip6_rt_cache_alloc(struct rt6_info *ort,
 
 static struct rt6_info *ip6_rt_pcpu_alloc(struct rt6_info *rt)
 {
+	struct net_device *dev;
 	struct rt6_info *pcpu_rt;
 
-	pcpu_rt = __ip6_dst_alloc(dev_net(rt->dst.dev),
-				  rt->dst.dev, rt->dst.flags);
-
+	rcu_read_lock();
+	dev = ip6_rt_get_dev_rcu(rt);
+	pcpu_rt = __ip6_dst_alloc(dev_net(dev), dev, rt->dst.flags);
+	rcu_read_unlock();
 	if (!pcpu_rt)
 		return NULL;
 	ip6_rt_copy_init(pcpu_rt, rt);
@@ -2688,15 +2716,9 @@ struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
 {
 	u32 tb_id;
 	struct net *net = dev_net(idev->dev);
-	struct net_device *dev = net->loopback_dev;
+	struct net_device *dev = idev->dev;
 	struct rt6_info *rt;
 
-	/* use L3 Master device as loopback for host routes if device
-	 * is enslaved and address is not link local or multicast
-	 */
-	if (!rt6_need_strict(addr))
-		dev = l3mdev_master_dev_rcu(idev->dev) ? : dev;
-
 	rt = ip6_dst_alloc(net, dev, DST_NOCOUNT);
 	if (!rt)
 		return ERR_PTR(-ENOMEM);
-- 
2.1.4

^ permalink raw reply related

* Re: [PATCH V4 net 0/2] ipv6: fix flowlabel issue for reset packet
From: Tom Herbert @ 2017-08-17 19:07 UTC (permalink / raw)
  To: Shaohua Li
  Cc: Linux Kernel Network Developers, David S. Miller,
	Martin KaFai Lau
In-Reply-To: <20170817172602.mmv73kjxuwwimxo6@kernel.org>

On Thu, Aug 17, 2017 at 10:26 AM, Shaohua Li <shli@kernel.org> wrote:
> On Tue, Aug 15, 2017 at 05:15:46PM -0700, Tom Herbert wrote:
>> On Tue, Aug 15, 2017 at 3:42 PM, Shaohua Li <shli@kernel.org> wrote:
>> > On Tue, Aug 15, 2017 at 07:08:31AM -0700, Tom Herbert wrote:
>> >> On Mon, Aug 14, 2017 at 7:52 PM, Shaohua Li <shli@kernel.org> wrote:
>> >> > On Fri, Aug 11, 2017 at 06:00:20PM -0700, Tom Herbert wrote:
>> >> >> On Thu, Aug 10, 2017 at 12:13 PM, Shaohua Li <shli@kernel.org> wrote:
>> >> >> > On Thu, Aug 10, 2017 at 11:30:51AM -0700, Tom Herbert wrote:
>> >> >> >> On Thu, Aug 10, 2017 at 9:30 AM, Shaohua Li <shli@kernel.org> wrote:
>> >> >> >> > On Wed, Aug 09, 2017 at 09:40:08AM -0700, Tom Herbert wrote:
>> >> >> >> >> On Mon, Jul 31, 2017 at 3:19 PM, Shaohua Li <shli@kernel.org> wrote:
>> >> >> >> >> > From: Shaohua Li <shli@fb.com>
>> >> >> >> >> >
>> >> >> >> >> > Please see below tcpdump output:
>> >> >> >> >> > 21:00:48.109122 IP6 (flowlabel 0x43304, hlim 64, next-header TCP (6) payload length: 40) fec0::5054:ff:fe12:3456.55804 > fec0::5054:ff:fe12:3456.5555: Flags [S], cksum 0x0529 (incorrect -> 0xf56c), seq 3282214508, win 43690, options [mss 65476,sackOK,TS val 2500903437 ecr 0,nop,wscale 7], length 0
>> >> >> >> >> > 21:00:48.109381 IP6 (flowlabel 0xd827f, hlim 64, next-header TCP (6) payload length: 40) fec0::5054:ff:fe12:3456.5555 > fec0::5054:ff:fe12:3456.55804: Flags [S.], cksum 0x0529 (incorrect -> 0x49ad), seq 1923801573, ack 3282214509, win 43690, options [mss 65476,sackOK,TS val 2500903437 ecr 2500903437,nop,wscale 7], length 0
>> >> >> >> >> > 21:00:48.109548 IP6 (flowlabel 0x43304, hlim 64, next-header TCP (6) payload length: 32) fec0::5054:ff:fe12:3456.55804 > fec0::5054:ff:fe12:3456.5555: Flags [.], cksum 0x0521 (incorrect -> 0x1bdf), seq 1, ack 1, win 342, options [nop,nop,TS val 2500903437 ecr 2500903437], length 0
>> >> >> >> >> > 21:00:48.109823 IP6 (flowlabel 0x43304, hlim 64, next-header TCP (6) payload length: 62) fec0::5054:ff:fe12:3456.55804 > fec0::5054:ff:fe12:3456.5555: Flags [P.], cksum 0x053f (incorrect -> 0xb8b1), seq 1:31, ack 1, win 342, options [nop,nop,TS val 2500903437 ecr 2500903437], length 30
>> >> >> >> >> > 21:00:48.109910 IP6 (flowlabel 0xd827f, hlim 64, next-header TCP (6) payload length: 32) fec0::5054:ff:fe12:3456.5555 > fec0::5054:ff:fe12:3456.55804: Flags [.], cksum 0x0521 (incorrect -> 0x1bc1), seq 1, ack 31, win 342, options [nop,nop,TS val 2500903437 ecr 2500903437], length 0
>> >> >> >> >> > 21:00:48.110043 IP6 (flowlabel 0xd827f, hlim 64, next-header TCP (6) payload length: 56) fec0::5054:ff:fe12:3456.5555 > fec0::5054:ff:fe12:3456.55804: Flags [P.], cksum 0x0539 (incorrect -> 0xb726), seq 1:25, ack 31, win 342, options [nop,nop,TS val 2500903438 ecr 2500903437], length 24
>> >> >> >> >> > 21:00:48.110173 IP6 (flowlabel 0x43304, hlim 64, next-header TCP (6) payload length: 32) fec0::5054:ff:fe12:3456.55804 > fec0::5054:ff:fe12:3456.5555: Flags [.], cksum 0x0521 (incorrect -> 0x1ba7), seq 31, ack 25, win 342, options [nop,nop,TS val 2500903438 ecr 2500903438], length 0
>> >> >> >> >> > 21:00:48.110211 IP6 (flowlabel 0xd827f, hlim 64, next-header TCP (6) payload length: 32) fec0::5054:ff:fe12:3456.5555 > fec0::5054:ff:fe12:3456.55804: Flags [F.], cksum 0x0521 (incorrect -> 0x1ba7), seq 25, ack 31, win 342, options [nop,nop,TS val 2500903438 ecr 2500903437], length 0
>> >> >> >> >> > 21:00:48.151099 IP6 (flowlabel 0x43304, hlim 64, next-header TCP (6) payload length: 32) fec0::5054:ff:fe12:3456.55804 > fec0::5054:ff:fe12:3456.5555: Flags [.], cksum 0x0521 (incorrect -> 0x1ba6), seq 31, ack 26, win 342, options [nop,nop,TS val 2500903438 ecr 2500903438], length 0
>> >> >> >> >> > 21:00:49.110524 IP6 (flowlabel 0x43304, hlim 64, next-header TCP (6) payload length: 56) fec0::5054:ff:fe12:3456.55804 > fec0::5054:ff:fe12:3456.5555: Flags [P.], cksum 0x0539 (incorrect -> 0xb324), seq 31:55, ack 26, win 342, options [nop,nop,TS val 2500904438 ecr 2500903438], length 24
>> >> >> >> >> > 21:00:49.110637 IP6 (flowlabel 0xb34d5, hlim 64, next-header TCP (6) payload length: 20) fec0::5054:ff:fe12:3456.5555 > fec0::5054:ff:fe12:3456.55804: Flags [R], cksum 0x0515 (incorrect -> 0x668c), seq 1923801599, win 0, length 0
>> >> >> >> >> >
>> >> >> >> >> > The flowlabel of reset packet (0xb34d5) and flowlabel of normal packet
>> >> >> >> >> > (0xd827f) are different. This causes our router doesn't correctly close tcp
>> >> >> >> >> > connection. The patches try to fix the issue.
>> >> >> >> >> >
>> >> >> >> >> Shaohua,
>> >> >> >> >>
>> >> >> >> >> Can you give some more detail about what the router doesn't close the
>> >> >> >> >> TCP connection means? I'm guessing the problem is either: 1) the
>> >> >> >> >> router is maintaining connection state that includes the flow label in
>> >> >> >> >> a connection tuple. 2) some router in the path is maintaining
>> >> >> >> >> connection state, but when the flow label changes the flow's packet
>> >> >> >> >> are routed through a different router that doesn't have a state for
>> >> >> >> >> the flow it drops the packet. #1 should be easily fix in the router,
>> >> >> >> >> flow labels cannot be used as state. #2 is the known problem that
>> >> >> >> >> stateful firewalls have killed our ability to use multihoming.
>> >> >> >> >
>> >> >> >> > The #2 is exactly the problem we saw.
>> >> >> >> >
>> >> >> >> >> Another consideration is that sk_txhash is also used in routing
>> >> >> >> >> decisions by the local host (flow label is normally derived from
>> >> >> >> >> txhash). If you want to ensure that connections are routed
>> >> >> >> >> consistently for timewait state you might need sk_txhash saved also.
>> >> >> >> >
>> >> >> >> > As far as I understood, we don't use sk_txhash for routing selection. The code
>> >> >> >> > does routing selection with flowlabel user configured, at that time we don't
>> >> >> >> > derive fl6.flowlabel from skb->hash (which is from sk_txhash). The code always
>> >> >> >> > does routing selection first and then uses ip6_make_flowlabel to build packet
>> >> >> >> > data where we derive flowlabel from skb->hash.
>> >> >> >> >
>> >> >> >> That is assuming one particular use case. Generally, if you want to
>> >> >> >> ensure all packets for a flow take the same path you'll need tx_hash
>> >> >> >> and make it persistent (disable flow bender). For instance, if you
>> >> >> >> were doing UDP encapsulation like in VXLAN the UDP source port
>> >> >> >> selection is unaffected by saved flow label for the lifetime of the
>> >> >> >> flow. So we would still hit #2 in that case and the stateful device
>> >> >> >> doesn't see whole flow. It might be just as easy to move tx_hash in
>> >> >> >> skc_common so that it's available in TW state for this purpose. Then
>> >> >> >> when moving to TW state just copy the tx_hash.
>> >> >> >
>> >> >> > Hi Tom,
>> >> >> >
>> >> >> > My original implementation is to add a tx_hash in tw sock, we then copy sock's
>> >> >> > tx_hash to the tw tx_hash. This does makes things simplier. One concern from
>> >> >> > Eric is this will increase the size of tw sock. If we move tx_hash to
>> >> >> > skc_common, all sock size will increase, is this acceptable?
>> >> >>
>> >> >> I think that can only be measured by how critical it is to
>> >> >> persistently route all packets the same exact way for every
>> >> >> connection. Page one of the IP book clearly states that IP packets can
>> >> >> be dropped, duplicated, or received out of order. Received OOO implies
>> >> >> that packet for the same flow are allowed to take different paths. The
>> >> >> requirement that packets for the same flow must always take the same
>> >> >> path through the network was created by stateful middleboxes-- it's
>> >> >> not inherent in the architecture of IP networking. Unfortunately,
>> >> >> we're seeing this become more and more of a problem as more devices
>> >> >> are multi-homed (like smart phones) and these network requirement
>> >> >> cripple our ability to take advantage of features like that.
>> >> >>
>> >> >> Personally, I wish the middleboxes fix the problem they created, but I
>> >> >> suppose we need to be pragmatic at least in the short term.
>> >> >
>> >> > Hmm, I still hesitate to add a new field in skc_common. Fixing current problem
>> >> > looks propriate in current stage. I'd defer fixing the generic issue till it's
>> >> > necessary.
>> >> >
>> >> Shaohua,
>> >>
>> >> An alternative would be to not initialize sk_txhash, but instead defer
>> >> hash computation to use flow dissector in the TX path when the hash is
>> >> needed (to get flow label, src port for UDP encap, route for
>> >> multipath, etc.). At the first hash computation in TX path  the result
>> >> in sk_txhash. TW state there is no socket so flow dissector is
>> >> always used but that should yield the same hash. No extra fields would
>> >> be needed and additional cost is negligible.
>> >
>> > Hi Tom,
>> >
>> > Did you mean revert 877d1f6291f8(net: Set sk_txhash from a random number)? This
>> > could fix the issue. So in normal case we calculate the sk_txhash using flow
>> > dissector but in negative routing case we use the random hash, is this what you
>> > want?
>> >
>> No. What you'd want is something like a sysctl that sets an alternate
>> mode for sk_txhash processing. sk_txhash is derived from flow
>> dissector for the first TX packet and then it's never allowed to
>> change. Maybe this should be called persistent-hash mode.
>
> persistent-hash will almostly mean disabling auto flowlabel. We'd prefer
> disabling auto flowlabel rather than the persistent-hash. But I think we do
> want to use the auto flowlabel for load balancing, so the persistent-hash isn't
> a go for us.
>
Those are two separate features here. Auto flow label means
automatically generating flow labels for packets based on some hash. A
persistent hash means that once a hash is determined for a connection
it never changes, so the flow label derived from the hash never
changes.

> Let's take the other way. Your main concern is the multipath selection. I think
> this can be fixed. We can still use the tw->tw_flowlabel to store the auto
> created flowlabel like what my patch does, but we use an extra bit in tw_pad of
> inet_timewait_sock to indicate this flowlabel is auto created. In the reset
> packet send path, we don't use the auto created flowlabel for route slection,
> but just for packet data. This will keep current multipath selection behavior
> and fix the reset packet flowlabel issue.
>
>> > There seems to have other bug in this side. From my understanding, commit
>> > 265f94ff54d6(net: Recompute sk_txhash on negative routing advice) tries to
>> > select a different route. But the multipath selection code
>> > (rt6_multipath_select) doesn't use sk_txhash or skb->hash, it does use
>> > fl6.flowlabel, but that is the flowlabel user sets. So looks like the commit
>> > doesn't change anything.
>> >
>> The routing functions typically don't use sock of skbuff, but use flow
>> structs instead. It may be reasonable to add a hash to those.
>
> I'll look at this issue later. It's not related to current flowlabel problem.
>
As I mentioned already, this is not a problem specific to flowlabels.
This is a problem about getting all the elements of the system work
together to ensure that all packets of flow are routed they same way
in order to meet a requirement imposed by middleboxes. You can solve
this problem just to make flow labels work, but IMO that's just
kicking the can down the road for someone else to have to debug and
solve when they hit the same underlying issue but in a different
context.

Tom

^ permalink raw reply

* [PATCH V7 net-next 22/22] net-next/hinic: Add Maintainer
From: Aviad Krawczyk @ 2017-08-17 19:04 UTC (permalink / raw)
  To: davem
  Cc: linux-kernel, netdev, bc.y, victor.gissin, aviad.krawczyk,
	zhaochen6, tony.qu
In-Reply-To: <cover.1502996306.git.aviad.krawczyk@huawei.com>

Update MAINTAINERS file

Signed-off-by: Aviad Krawczyk <aviad.krawczyk@huawei.com>
Signed-off-by: Zhao Chen <zhaochen6@huawei.com>
---
 MAINTAINERS | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/MAINTAINERS b/MAINTAINERS
index 0e967b3..8f9ea9b 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -6240,6 +6240,13 @@ L:	linux-input@vger.kernel.org
 S:	Maintained
 F:	drivers/input/touchscreen/htcpen.c
 
+HUAWEI ETHERNET DRIVER
+M:	Aviad Krawczyk <aviad.krawczyk@huawei.com>
+L:	netdev@vger.kernel.org
+S:	Supported
+F:	Documentation/networking/hinic.txt
+F:	drivers/net/ethernet/huawei/hinic/
+
 HUGETLB FILESYSTEM
 M:	Nadia Yvette Chambers <nyc@holomorphy.com>
 S:	Maintained
-- 
1.9.1

^ permalink raw reply related

* [PATCH V7 net-next 21/22] net-next/hinic: Add netpoll
From: Aviad Krawczyk @ 2017-08-17 19:04 UTC (permalink / raw)
  To: davem
  Cc: linux-kernel, netdev, bc.y, victor.gissin, aviad.krawczyk,
	zhaochen6, tony.qu
In-Reply-To: <cover.1502996306.git.aviad.krawczyk@huawei.com>

Add more netdev operation - netpoll.

Signed-off-by: Aviad Krawczyk <aviad.krawczyk@huawei.com>
Signed-off-by: Zhao Chen <zhaochen6@huawei.com>
---
 drivers/net/ethernet/huawei/hinic/hinic_main.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)

diff --git a/drivers/net/ethernet/huawei/hinic/hinic_main.c b/drivers/net/ethernet/huawei/hinic/hinic_main.c
index a77a7f8..59f3358 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_main.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_main.c
@@ -787,6 +787,23 @@ static void hinic_get_stats64(struct net_device *netdev,
 	stats->tx_errors  = nic_tx_stats->tx_dropped;
 }
 
+#ifdef CONFIG_NET_POLL_CONTROLLER
+static void hinic_netpoll(struct net_device *netdev)
+{
+	struct hinic_dev *nic_dev = netdev_priv(netdev);
+	int i, num_qps;
+
+	num_qps = hinic_hwdev_num_qps(nic_dev->hwdev);
+	for (i = 0; i < num_qps; i++) {
+		struct hinic_txq *txq = &nic_dev->txqs[i];
+		struct hinic_rxq *rxq = &nic_dev->rxqs[i];
+
+		napi_schedule(&txq->napi);
+		napi_schedule(&rxq->napi);
+	}
+}
+#endif
+
 static const struct net_device_ops hinic_netdev_ops = {
 	.ndo_open = hinic_open,
 	.ndo_stop = hinic_close,
@@ -799,6 +816,9 @@ static void hinic_get_stats64(struct net_device *netdev,
 	.ndo_start_xmit = hinic_xmit_frame,
 	.ndo_tx_timeout = hinic_tx_timeout,
 	.ndo_get_stats64 = hinic_get_stats64,
+#ifdef CONFIG_NET_POLL_CONTROLLER
+	.ndo_poll_controller = hinic_netpoll,
+#endif
 };
 
 static void netdev_features_init(struct net_device *netdev)
-- 
1.9.1

^ permalink raw reply related

* [PATCH V7 net-next 20/22] net-next/hinic: Add ethtool and stats
From: Aviad Krawczyk @ 2017-08-17 19:04 UTC (permalink / raw)
  To: davem
  Cc: linux-kernel, netdev, bc.y, victor.gissin, aviad.krawczyk,
	zhaochen6, tony.qu
In-Reply-To: <cover.1502996306.git.aviad.krawczyk@huawei.com>

Add ethtool operations and statistics operations.

Signed-off-by: Aviad Krawczyk <aviad.krawczyk@huawei.com>
Signed-off-by: Zhao Chen <zhaochen6@huawei.com>
---
 drivers/net/ethernet/huawei/hinic/hinic_dev.h  |   3 +
 drivers/net/ethernet/huawei/hinic/hinic_main.c | 218 ++++++++++++++++++++++++-
 drivers/net/ethernet/huawei/hinic/hinic_port.c |  31 ++++
 drivers/net/ethernet/huawei/hinic/hinic_port.h |  45 +++++
 drivers/net/ethernet/huawei/hinic/hinic_rx.c   |  19 +++
 drivers/net/ethernet/huawei/hinic/hinic_rx.h   |   2 +
 drivers/net/ethernet/huawei/hinic/hinic_tx.c   |  22 +++
 drivers/net/ethernet/huawei/hinic/hinic_tx.h   |   2 +
 8 files changed, 341 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/huawei/hinic/hinic_dev.h b/drivers/net/ethernet/huawei/hinic/hinic_dev.h
index 15d0c2e..5186cc9 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_dev.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_dev.h
@@ -56,6 +56,9 @@ struct hinic_dev {
 
 	struct hinic_txq                *txqs;
 	struct hinic_rxq                *rxqs;
+
+	struct hinic_txq_stats          tx_stats;
+	struct hinic_rxq_stats          rx_stats;
 };
 
 #endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_main.c b/drivers/net/ethernet/huawei/hinic/hinic_main.c
index 4a5f23f..a77a7f8 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_main.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_main.c
@@ -69,6 +69,186 @@
 
 static int change_mac_addr(struct net_device *netdev, const u8 *addr);
 
+static void set_link_speed(struct ethtool_link_ksettings *link_ksettings,
+			   enum hinic_speed speed)
+{
+	switch (speed) {
+	case HINIC_SPEED_10MB_LINK:
+		link_ksettings->base.speed = SPEED_10;
+		break;
+
+	case HINIC_SPEED_100MB_LINK:
+		link_ksettings->base.speed = SPEED_100;
+		break;
+
+	case HINIC_SPEED_1000MB_LINK:
+		link_ksettings->base.speed = SPEED_1000;
+		break;
+
+	case HINIC_SPEED_10GB_LINK:
+		link_ksettings->base.speed = SPEED_10000;
+		break;
+
+	case HINIC_SPEED_25GB_LINK:
+		link_ksettings->base.speed = SPEED_25000;
+		break;
+
+	case HINIC_SPEED_40GB_LINK:
+		link_ksettings->base.speed = SPEED_40000;
+		break;
+
+	case HINIC_SPEED_100GB_LINK:
+		link_ksettings->base.speed = SPEED_100000;
+		break;
+
+	default:
+		link_ksettings->base.speed = SPEED_UNKNOWN;
+		break;
+	}
+}
+
+static int hinic_get_link_ksettings(struct net_device *netdev,
+				    struct ethtool_link_ksettings
+				    *link_ksettings)
+{
+	struct hinic_dev *nic_dev = netdev_priv(netdev);
+	enum hinic_port_link_state link_state;
+	struct hinic_port_cap port_cap;
+	int err;
+
+	ethtool_link_ksettings_zero_link_mode(link_ksettings, advertising);
+	ethtool_link_ksettings_add_link_mode(link_ksettings, supported,
+					     Autoneg);
+
+	link_ksettings->base.speed   = SPEED_UNKNOWN;
+	link_ksettings->base.autoneg = AUTONEG_DISABLE;
+	link_ksettings->base.duplex  = DUPLEX_UNKNOWN;
+
+	err = hinic_port_get_cap(nic_dev, &port_cap);
+	if (err) {
+		netif_err(nic_dev, drv, netdev,
+			  "Failed to get port capabilities\n");
+		return err;
+	}
+
+	err = hinic_port_link_state(nic_dev, &link_state);
+	if (err) {
+		netif_err(nic_dev, drv, netdev,
+			  "Failed to get port link state\n");
+		return err;
+	}
+
+	if (link_state != HINIC_LINK_STATE_UP) {
+		netif_info(nic_dev, drv, netdev, "No link\n");
+		return err;
+	}
+
+	set_link_speed(link_ksettings, port_cap.speed);
+
+	if (!!(port_cap.autoneg_cap & HINIC_AUTONEG_SUPPORTED))
+		ethtool_link_ksettings_add_link_mode(link_ksettings,
+						     advertising, Autoneg);
+
+	if (port_cap.autoneg_state == HINIC_AUTONEG_ACTIVE)
+		link_ksettings->base.autoneg = AUTONEG_ENABLE;
+
+	link_ksettings->base.duplex = (port_cap.duplex == HINIC_DUPLEX_FULL) ?
+				       DUPLEX_FULL : DUPLEX_HALF;
+	return 0;
+}
+
+static void hinic_get_drvinfo(struct net_device *netdev,
+			      struct ethtool_drvinfo *info)
+{
+	struct hinic_dev *nic_dev = netdev_priv(netdev);
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	struct hinic_hwif *hwif = hwdev->hwif;
+
+	strlcpy(info->driver, HINIC_DRV_NAME, sizeof(info->driver));
+	strlcpy(info->bus_info, pci_name(hwif->pdev), sizeof(info->bus_info));
+}
+
+static void hinic_get_ringparam(struct net_device *netdev,
+				struct ethtool_ringparam *ring)
+{
+	ring->rx_max_pending = HINIC_RQ_DEPTH;
+	ring->tx_max_pending = HINIC_SQ_DEPTH;
+	ring->rx_pending = HINIC_RQ_DEPTH;
+	ring->tx_pending = HINIC_SQ_DEPTH;
+}
+
+static void hinic_get_channels(struct net_device *netdev,
+			       struct ethtool_channels *channels)
+{
+	struct hinic_dev *nic_dev = netdev_priv(netdev);
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+
+	channels->max_rx = hwdev->nic_cap.max_qps;
+	channels->max_tx = hwdev->nic_cap.max_qps;
+	channels->max_other    = 0;
+	channels->max_combined = 0;
+	channels->rx_count = hinic_hwdev_num_qps(hwdev);
+	channels->tx_count = hinic_hwdev_num_qps(hwdev);
+	channels->other_count    = 0;
+	channels->combined_count = 0;
+}
+
+static const struct ethtool_ops hinic_ethtool_ops = {
+	.get_link_ksettings = hinic_get_link_ksettings,
+	.get_drvinfo = hinic_get_drvinfo,
+	.get_link = ethtool_op_get_link,
+	.get_ringparam = hinic_get_ringparam,
+	.get_channels = hinic_get_channels,
+};
+
+static void update_rx_stats(struct hinic_dev *nic_dev, struct hinic_rxq *rxq)
+{
+	struct hinic_rxq_stats *nic_rx_stats = &nic_dev->rx_stats;
+	struct hinic_rxq_stats rx_stats;
+
+	u64_stats_init(&rx_stats.syncp);
+
+	hinic_rxq_get_stats(rxq, &rx_stats);
+
+	u64_stats_update_begin(&nic_rx_stats->syncp);
+	nic_rx_stats->bytes += rx_stats.bytes;
+	nic_rx_stats->pkts  += rx_stats.pkts;
+	u64_stats_update_end(&nic_rx_stats->syncp);
+
+	hinic_rxq_clean_stats(rxq);
+}
+
+static void update_tx_stats(struct hinic_dev *nic_dev, struct hinic_txq *txq)
+{
+	struct hinic_txq_stats *nic_tx_stats = &nic_dev->tx_stats;
+	struct hinic_txq_stats tx_stats;
+
+	u64_stats_init(&tx_stats.syncp);
+
+	hinic_txq_get_stats(txq, &tx_stats);
+
+	u64_stats_update_begin(&nic_tx_stats->syncp);
+	nic_tx_stats->bytes += tx_stats.bytes;
+	nic_tx_stats->pkts += tx_stats.pkts;
+	nic_tx_stats->tx_busy += tx_stats.tx_busy;
+	nic_tx_stats->tx_wake += tx_stats.tx_wake;
+	nic_tx_stats->tx_dropped += tx_stats.tx_dropped;
+	u64_stats_update_end(&nic_tx_stats->syncp);
+
+	hinic_txq_clean_stats(txq);
+}
+
+static void update_nic_stats(struct hinic_dev *nic_dev)
+{
+	int i, num_qps = hinic_hwdev_num_qps(nic_dev->hwdev);
+
+	for (i = 0; i < num_qps; i++)
+		update_rx_stats(nic_dev, &nic_dev->rxqs[i]);
+
+	for (i = 0; i < num_qps; i++)
+		update_tx_stats(nic_dev, &nic_dev->txqs[i]);
+}
+
 /**
  * create_txqs - Create the Logical Tx Queues of specific NIC device
  * @nic_dev: the specific NIC device
@@ -303,6 +483,8 @@ static int hinic_close(struct net_device *netdev)
 	netif_carrier_off(netdev);
 	netif_tx_disable(netdev);
 
+	update_nic_stats(nic_dev);
+
 	up(&nic_dev->mgmt_lock);
 
 	err = hinic_port_set_func_state(nic_dev, HINIC_FUNC_PORT_DISABLE);
@@ -580,6 +762,31 @@ static void hinic_tx_timeout(struct net_device *netdev)
 	netif_err(nic_dev, drv, netdev, "Tx timeout\n");
 }
 
+static void hinic_get_stats64(struct net_device *netdev,
+			      struct rtnl_link_stats64 *stats)
+{
+	struct hinic_dev *nic_dev = netdev_priv(netdev);
+	struct hinic_rxq_stats *nic_rx_stats;
+	struct hinic_txq_stats *nic_tx_stats;
+
+	nic_rx_stats = &nic_dev->rx_stats;
+	nic_tx_stats = &nic_dev->tx_stats;
+
+	down(&nic_dev->mgmt_lock);
+
+	if (nic_dev->flags & HINIC_INTF_UP)
+		update_nic_stats(nic_dev);
+
+	up(&nic_dev->mgmt_lock);
+
+	stats->rx_bytes   = nic_rx_stats->bytes;
+	stats->rx_packets = nic_rx_stats->pkts;
+
+	stats->tx_bytes   = nic_tx_stats->bytes;
+	stats->tx_packets = nic_tx_stats->pkts;
+	stats->tx_errors  = nic_tx_stats->tx_dropped;
+}
+
 static const struct net_device_ops hinic_netdev_ops = {
 	.ndo_open = hinic_open,
 	.ndo_stop = hinic_close,
@@ -591,7 +798,7 @@ static void hinic_tx_timeout(struct net_device *netdev)
 	.ndo_set_rx_mode = hinic_set_rx_mode,
 	.ndo_start_xmit = hinic_xmit_frame,
 	.ndo_tx_timeout = hinic_tx_timeout,
-	/* more operations should be filled */
+	.ndo_get_stats64 = hinic_get_stats64,
 };
 
 static void netdev_features_init(struct net_device *netdev)
@@ -663,6 +870,8 @@ static void link_status_event_handler(void *handle, void *buf_in, u16 in_size,
 static int nic_dev_init(struct pci_dev *pdev)
 {
 	struct hinic_rx_mode_work *rx_mode_work;
+	struct hinic_txq_stats *tx_stats;
+	struct hinic_rxq_stats *rx_stats;
 	struct hinic_dev *nic_dev;
 	struct net_device *netdev;
 	struct hinic_hwdev *hwdev;
@@ -689,6 +898,7 @@ static int nic_dev_init(struct pci_dev *pdev)
 	}
 
 	netdev->netdev_ops = &hinic_netdev_ops;
+	netdev->ethtool_ops = &hinic_ethtool_ops;
 
 	nic_dev = (struct hinic_dev *)netdev_priv(netdev);
 	nic_dev->netdev = netdev;
@@ -702,6 +912,12 @@ static int nic_dev_init(struct pci_dev *pdev)
 
 	sema_init(&nic_dev->mgmt_lock, 1);
 
+	tx_stats = &nic_dev->tx_stats;
+	rx_stats = &nic_dev->rx_stats;
+
+	u64_stats_init(&tx_stats->syncp);
+	u64_stats_init(&rx_stats->syncp);
+
 	nic_dev->vlan_bitmap = devm_kzalloc(&pdev->dev,
 					    VLAN_BITMAP_SIZE(nic_dev),
 					    GFP_KERNEL);
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_port.c b/drivers/net/ethernet/huawei/hinic/hinic_port.c
index 528ec6f..4d4e3f0 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_port.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_port.c
@@ -346,3 +346,34 @@ int hinic_port_set_func_state(struct hinic_dev *nic_dev,
 
 	return 0;
 }
+
+/**
+ * hinic_port_get_cap - get port capabilities
+ * @nic_dev: nic device
+ * @port_cap: returned port capabilities
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_port_get_cap(struct hinic_dev *nic_dev,
+		       struct hinic_port_cap *port_cap)
+{
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	u16 out_size;
+	int err;
+
+	port_cap->func_idx = HINIC_HWIF_FUNC_IDX(hwif);
+
+	err = hinic_port_msg_cmd(hwdev, HINIC_PORT_CMD_GET_CAP,
+				 port_cap, sizeof(*port_cap),
+				 port_cap, &out_size);
+	if (err || (out_size != sizeof(*port_cap)) || port_cap->status) {
+		dev_err(&pdev->dev,
+			"Failed to get port capabilities, ret = %d\n",
+			port_cap->status);
+		return -EINVAL;
+	}
+
+	return 0;
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_port.h b/drivers/net/ethernet/huawei/hinic/hinic_port.h
index 17f9d7fc..9404365 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_port.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_port.h
@@ -45,6 +45,33 @@ enum hinic_func_port_state {
 	HINIC_FUNC_PORT_ENABLE  = 2,
 };
 
+enum hinic_autoneg_cap {
+	HINIC_AUTONEG_UNSUPPORTED,
+	HINIC_AUTONEG_SUPPORTED,
+};
+
+enum hinic_autoneg_state {
+	HINIC_AUTONEG_DISABLED,
+	HINIC_AUTONEG_ACTIVE,
+};
+
+enum hinic_duplex {
+	HINIC_DUPLEX_HALF,
+	HINIC_DUPLEX_FULL,
+};
+
+enum hinic_speed {
+	HINIC_SPEED_10MB_LINK = 0,
+	HINIC_SPEED_100MB_LINK,
+	HINIC_SPEED_1000MB_LINK,
+	HINIC_SPEED_10GB_LINK,
+	HINIC_SPEED_25GB_LINK,
+	HINIC_SPEED_40GB_LINK,
+	HINIC_SPEED_100GB_LINK,
+
+	HINIC_SPEED_UNKNOWN = 0xFF,
+};
+
 struct hinic_port_mac_cmd {
 	u8              status;
 	u8              version;
@@ -125,6 +152,21 @@ struct hinic_port_func_state_cmd {
 	u8      rsvd2[3];
 };
 
+struct hinic_port_cap {
+	u8      status;
+	u8      version;
+	u8      rsvd0[6];
+
+	u16     func_idx;
+	u16     rsvd1;
+	u8      port_type;
+	u8      autoneg_cap;
+	u8      autoneg_state;
+	u8      duplex;
+	u8      speed;
+	u8      rsvd2[3];
+};
+
 int hinic_port_add_mac(struct hinic_dev *nic_dev, const u8 *addr,
 		       u16 vlan_id);
 
@@ -150,4 +192,7 @@ int hinic_port_set_state(struct hinic_dev *nic_dev,
 int hinic_port_set_func_state(struct hinic_dev *nic_dev,
 			      enum hinic_func_port_state state);
 
+int hinic_port_get_cap(struct hinic_dev *nic_dev,
+		       struct hinic_port_cap *port_cap);
+
 #endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_rx.c b/drivers/net/ethernet/huawei/hinic/hinic_rx.c
index 6367fa1..bbe5d7d 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_rx.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_rx.c
@@ -58,6 +58,25 @@ void hinic_rxq_clean_stats(struct hinic_rxq *rxq)
 }
 
 /**
+ * hinic_rxq_get_stats - get statistics of Rx Queue
+ * @rxq: Logical Rx Queue
+ * @stats: return updated stats here
+ **/
+void hinic_rxq_get_stats(struct hinic_rxq *rxq, struct hinic_rxq_stats *stats)
+{
+	struct hinic_rxq_stats *rxq_stats = &rxq->rxq_stats;
+	unsigned int start;
+
+	u64_stats_update_begin(&stats->syncp);
+	do {
+		start = u64_stats_fetch_begin(&rxq_stats->syncp);
+		stats->pkts = rxq_stats->pkts;
+		stats->bytes = rxq_stats->bytes;
+	} while (u64_stats_fetch_retry(&rxq_stats->syncp, start));
+	u64_stats_update_end(&stats->syncp);
+}
+
+/**
  * rxq_stats_init - Initialize the statistics of specific queue
  * @rxq: Logical Rx Queue
  **/
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_rx.h b/drivers/net/ethernet/huawei/hinic/hinic_rx.h
index 538c886..27c9af4 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_rx.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_rx.h
@@ -45,6 +45,8 @@ struct hinic_rxq {
 
 void hinic_rxq_clean_stats(struct hinic_rxq *rxq);
 
+void hinic_rxq_get_stats(struct hinic_rxq *rxq, struct hinic_rxq_stats *stats);
+
 int hinic_init_rxq(struct hinic_rxq *rxq, struct hinic_rq *rq,
 		   struct net_device *netdev);
 
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_tx.c b/drivers/net/ethernet/huawei/hinic/hinic_tx.c
index d0c539f..b5056c3 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_tx.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_tx.c
@@ -67,6 +67,28 @@ void hinic_txq_clean_stats(struct hinic_txq *txq)
 }
 
 /**
+ * hinic_txq_get_stats - get statistics of Tx Queue
+ * @txq: Logical Tx Queue
+ * @stats: return updated stats here
+ **/
+void hinic_txq_get_stats(struct hinic_txq *txq, struct hinic_txq_stats *stats)
+{
+	struct hinic_txq_stats *txq_stats = &txq->txq_stats;
+	unsigned int start;
+
+	u64_stats_update_begin(&stats->syncp);
+	do {
+		start = u64_stats_fetch_begin(&txq_stats->syncp);
+		stats->pkts = txq_stats->pkts;
+		stats->bytes = txq_stats->bytes;
+		stats->tx_busy = txq_stats->tx_busy;
+		stats->tx_wake = txq_stats->tx_wake;
+		stats->tx_dropped = txq_stats->tx_dropped;
+	} while (u64_stats_fetch_retry(&txq_stats->syncp, start));
+	u64_stats_update_end(&stats->syncp);
+}
+
+/**
  * txq_stats_init - Initialize the statistics of specific queue
  * @txq: Logical Tx Queue
  **/
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_tx.h b/drivers/net/ethernet/huawei/hinic/hinic_tx.h
index 7123c7f..1fa55dc 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_tx.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_tx.h
@@ -50,6 +50,8 @@ struct hinic_txq {
 
 void hinic_txq_clean_stats(struct hinic_txq *txq);
 
+void hinic_txq_get_stats(struct hinic_txq *txq, struct hinic_txq_stats *stats);
+
 netdev_tx_t hinic_xmit_frame(struct sk_buff *skb, struct net_device *netdev);
 
 int hinic_init_txq(struct hinic_txq *txq, struct hinic_sq *sq,
-- 
1.9.1

^ permalink raw reply related

* [PATCH V7 net-next 19/22] net-next/hinic: Add Tx operation
From: Aviad Krawczyk @ 2017-08-17 19:04 UTC (permalink / raw)
  To: davem
  Cc: linux-kernel, netdev, bc.y, victor.gissin, aviad.krawczyk,
	zhaochen6, tony.qu
In-Reply-To: <cover.1502996306.git.aviad.krawczyk@huawei.com>

Add transmit operation for sending data by qp operations.

Signed-off-by: Aviad Krawczyk <aviad.krawczyk@huawei.com>
Signed-off-by: Zhao Chen <zhaochen6@huawei.com>
---
 drivers/net/ethernet/huawei/hinic/hinic_dev.h    |   1 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c |  47 +++
 drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h |  22 ++
 drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c  | 257 ++++++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h  |  48 +++
 drivers/net/ethernet/huawei/hinic/hinic_main.c   |  12 +-
 drivers/net/ethernet/huawei/hinic/hinic_tx.c     | 406 +++++++++++++++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_tx.h     |  11 +
 8 files changed, 802 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/huawei/hinic/hinic_dev.h b/drivers/net/ethernet/huawei/hinic/hinic_dev.h
index 3d0f6cf..15d0c2e 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_dev.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_dev.h
@@ -43,6 +43,7 @@ struct hinic_dev {
 	struct hinic_hwdev              *hwdev;
 
 	u32                             msg_enable;
+	unsigned int                    tx_weight;
 	unsigned int                    rx_weight;
 
 	unsigned int                    flags;
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
index 03149e8..6606fba 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
@@ -40,6 +40,8 @@
 #define MAX_IRQS(max_qps, num_aeqs, num_ceqs)   \
 		 (2 * (max_qps) + (num_aeqs) + (num_ceqs))
 
+#define ADDR_IN_4BYTES(addr)            ((addr) >> 2)
+
 enum intr_type {
 	INTR_MSIX_TYPE,
 };
@@ -996,3 +998,48 @@ int hinic_hwdev_msix_set(struct hinic_hwdev *hwdev, u16 msix_index,
 				   lli_timer_cfg, lli_credit_limit,
 				   resend_timer);
 }
+
+/**
+ * hinic_hwdev_hw_ci_addr_set - set cons idx addr and attributes in HW for sq
+ * @hwdev: the NIC HW device
+ * @sq: send queue
+ * @pending_limit: the maximum pending update ci events (unit 8)
+ * @coalesc_timer: coalesc period for update ci (unit 8 us)
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_hwdev_hw_ci_addr_set(struct hinic_hwdev *hwdev, struct hinic_sq *sq,
+			       u8 pending_limit, u8 coalesc_timer)
+{
+	struct hinic_qp *qp = container_of(sq, struct hinic_qp, sq);
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_pfhwdev *pfhwdev;
+	struct hinic_cmd_hw_ci hw_ci;
+
+	if (!HINIC_IS_PF(hwif) && !HINIC_IS_PPF(hwif)) {
+		dev_err(&pdev->dev, "Unsupported PCI Function type\n");
+		return -EINVAL;
+	}
+
+	hw_ci.dma_attr_off  = 0;
+	hw_ci.pending_limit = pending_limit;
+	hw_ci.coalesc_timer  = coalesc_timer;
+
+	hw_ci.msix_en = 1;
+	hw_ci.msix_entry_idx = sq->msix_entry;
+
+	hw_ci.func_idx = HINIC_HWIF_FUNC_IDX(hwif);
+
+	hw_ci.sq_id = qp->q_id;
+
+	hw_ci.ci_addr = ADDR_IN_4BYTES(sq->hw_ci_dma_addr);
+
+	pfhwdev = container_of(hwdev, struct hinic_pfhwdev, hwdev);
+
+	return hinic_msg_to_mgmt(&pfhwdev->pf_to_mgmt,
+				 HINIC_MOD_COMM,
+				 HINIC_COMM_CMD_SQ_HI_CI_SET,
+				 &hw_ci, sizeof(hw_ci), NULL,
+				 NULL, HINIC_MGMT_MSG_SYNC);
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h
index e7277d1..0f5563f 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h
@@ -153,6 +153,25 @@ struct hinic_cmd_base_qpn {
 	u16     qpn;
 };
 
+struct hinic_cmd_hw_ci {
+	u8      status;
+	u8      version;
+	u8      rsvd0[6];
+
+	u16     func_idx;
+
+	u8      dma_attr_off;
+	u8      pending_limit;
+	u8      coalesc_timer;
+
+	u8      msix_en;
+	u16     msix_entry_idx;
+
+	u32     sq_id;
+	u32     rsvd1;
+	u64     ci_addr;
+};
+
 struct hinic_hwdev {
 	struct hinic_hwif               *hwif;
 	struct msix_entry               *msix_entries;
@@ -214,4 +233,7 @@ int hinic_hwdev_msix_set(struct hinic_hwdev *hwdev, u16 msix_index,
 			 u8 lli_timer_cfg, u8 lli_credit_limit,
 			 u8 resend_timer);
 
+int hinic_hwdev_hw_ci_addr_set(struct hinic_hwdev *hwdev, struct hinic_sq *sq,
+			       u8 pending_limit, u8 coalesc_timer);
+
 #endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c
index 6e540d2..97ee8eb 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c
@@ -23,6 +23,7 @@
 #include <linux/sizes.h>
 #include <linux/atomic.h>
 #include <linux/skbuff.h>
+#include <linux/io.h>
 #include <asm/barrier.h>
 #include <asm/byteorder.h>
 
@@ -32,6 +33,7 @@
 #include "hinic_hw_wq.h"
 #include "hinic_hw_qp_ctxt.h"
 #include "hinic_hw_qp.h"
+#include "hinic_hw_io.h"
 
 #define SQ_DB_OFF               SZ_2K
 
@@ -55,8 +57,24 @@
 
 #define SIZE_16BYTES(size)      (ALIGN(size, 16) >> 4)
 #define SIZE_8BYTES(size)       (ALIGN(size, 8) >> 3)
+#define SECT_SIZE_FROM_8BYTES(size)     ((size) << 3)
+
+#define SQ_DB_PI_HI_SHIFT       8
+#define SQ_DB_PI_HI(prod_idx)   ((prod_idx) >> SQ_DB_PI_HI_SHIFT)
+
+#define SQ_DB_PI_LOW_MASK       0xFF
+#define SQ_DB_PI_LOW(prod_idx)  ((prod_idx) & SQ_DB_PI_LOW_MASK)
+
+#define SQ_DB_ADDR(sq, pi)      ((u64 *)((sq)->db_base) + SQ_DB_PI_LOW(pi))
 
 #define RQ_MASKED_IDX(rq, idx)  ((idx) & (rq)->wq->mask)
+#define SQ_MASKED_IDX(sq, idx)  ((idx) & (sq)->wq->mask)
+
+#define TX_MAX_MSS_DEFAULT      0x3E00
+
+enum sq_wqe_type {
+	SQ_NORMAL_WQE = 0,
+};
 
 enum rq_completion_fmt {
 	RQ_COMPLETE_SGE = 1
@@ -435,6 +453,19 @@ void hinic_clean_rq(struct hinic_rq *rq)
 }
 
 /**
+ * hinic_get_sq_free_wqebbs - return number of free wqebbs for use
+ * @sq: send queue
+ *
+ * Return number of free wqebbs
+ **/
+int hinic_get_sq_free_wqebbs(struct hinic_sq *sq)
+{
+	struct hinic_wq *wq = sq->wq;
+
+	return atomic_read(&wq->delta) - 1;
+}
+
+/**
  * hinic_get_rq_free_wqebbs - return number of free wqebbs for use
  * @rq: recv queue
  *
@@ -447,6 +478,232 @@ int hinic_get_rq_free_wqebbs(struct hinic_rq *rq)
 	return atomic_read(&wq->delta) - 1;
 }
 
+static void sq_prepare_ctrl(struct hinic_sq_ctrl *ctrl, u16 prod_idx,
+			    int nr_descs)
+{
+	u32 ctrl_size, task_size, bufdesc_size;
+
+	ctrl_size = SIZE_8BYTES(sizeof(struct hinic_sq_ctrl));
+	task_size = SIZE_8BYTES(sizeof(struct hinic_sq_task));
+	bufdesc_size = nr_descs * sizeof(struct hinic_sq_bufdesc);
+	bufdesc_size = SIZE_8BYTES(bufdesc_size);
+
+	ctrl->ctrl_info = HINIC_SQ_CTRL_SET(bufdesc_size, BUFDESC_SECT_LEN) |
+			HINIC_SQ_CTRL_SET(task_size, TASKSECT_LEN)      |
+			HINIC_SQ_CTRL_SET(SQ_NORMAL_WQE, DATA_FORMAT)   |
+			HINIC_SQ_CTRL_SET(ctrl_size, LEN);
+
+	ctrl->queue_info = HINIC_SQ_CTRL_SET(TX_MAX_MSS_DEFAULT,
+					     QUEUE_INFO_MSS);
+}
+
+static void sq_prepare_task(struct hinic_sq_task *task)
+{
+	task->pkt_info0 =
+		HINIC_SQ_TASK_INFO0_SET(0, L2HDR_LEN) |
+		HINIC_SQ_TASK_INFO0_SET(HINIC_L4_OFF_DISABLE, L4_OFFLOAD) |
+		HINIC_SQ_TASK_INFO0_SET(HINIC_OUTER_L3TYPE_UNKNOWN,
+					INNER_L3TYPE) |
+		HINIC_SQ_TASK_INFO0_SET(HINIC_VLAN_OFF_DISABLE,
+					VLAN_OFFLOAD) |
+		HINIC_SQ_TASK_INFO0_SET(HINIC_PKT_NOT_PARSED, PARSE_FLAG);
+
+	task->pkt_info1 =
+		HINIC_SQ_TASK_INFO1_SET(HINIC_MEDIA_UNKNOWN, MEDIA_TYPE) |
+		HINIC_SQ_TASK_INFO1_SET(0, INNER_L4_LEN) |
+		HINIC_SQ_TASK_INFO1_SET(0, INNER_L3_LEN);
+
+	task->pkt_info2 =
+		HINIC_SQ_TASK_INFO2_SET(0, TUNNEL_L4_LEN) |
+		HINIC_SQ_TASK_INFO2_SET(0, OUTER_L3_LEN) |
+		HINIC_SQ_TASK_INFO2_SET(HINIC_TUNNEL_L4TYPE_UNKNOWN,
+					TUNNEL_L4TYPE) |
+		HINIC_SQ_TASK_INFO2_SET(HINIC_OUTER_L3TYPE_UNKNOWN,
+					OUTER_L3TYPE);
+
+	task->ufo_v6_identify = 0;
+
+	task->pkt_info4 = HINIC_SQ_TASK_INFO4_SET(HINIC_L2TYPE_ETH, L2TYPE);
+
+	task->zero_pad = 0;
+}
+
+/**
+ * hinic_sq_prepare_wqe - prepare wqe before insert to the queue
+ * @sq: send queue
+ * @prod_idx: pi value
+ * @sq_wqe: wqe to prepare
+ * @sges: sges for use by the wqe for send for buf addresses
+ * @nr_sges: number of sges
+ **/
+void hinic_sq_prepare_wqe(struct hinic_sq *sq, u16 prod_idx,
+			  struct hinic_sq_wqe *sq_wqe, struct hinic_sge *sges,
+			  int nr_sges)
+{
+	int i;
+
+	sq_prepare_ctrl(&sq_wqe->ctrl, prod_idx, nr_sges);
+
+	sq_prepare_task(&sq_wqe->task);
+
+	for (i = 0; i < nr_sges; i++)
+		sq_wqe->buf_descs[i].sge = sges[i];
+}
+
+/**
+ * sq_prepare_db - prepare doorbell to write
+ * @sq: send queue
+ * @prod_idx: pi value for the doorbell
+ * @cos: cos of the doorbell
+ *
+ * Return db value
+ **/
+static u32 sq_prepare_db(struct hinic_sq *sq, u16 prod_idx, unsigned int cos)
+{
+	struct hinic_qp *qp = container_of(sq, struct hinic_qp, sq);
+	int hi_prod_idx = SQ_DB_PI_HI(SQ_MASKED_IDX(sq, prod_idx));
+
+	/* Data should be written to HW in Big Endian Format */
+	return cpu_to_be32(HINIC_SQ_DB_INFO_SET(hi_prod_idx, PI_HI)     |
+			   HINIC_SQ_DB_INFO_SET(HINIC_DB_SQ_TYPE, TYPE) |
+			   HINIC_SQ_DB_INFO_SET(HINIC_DATA_PATH, PATH)  |
+			   HINIC_SQ_DB_INFO_SET(cos, COS)               |
+			   HINIC_SQ_DB_INFO_SET(qp->q_id, QID));
+}
+
+/**
+ * hinic_sq_write_db- write doorbell
+ * @sq: send queue
+ * @prod_idx: pi value for the doorbell
+ * @wqe_size: wqe size
+ * @cos: cos of the doorbell
+ **/
+void hinic_sq_write_db(struct hinic_sq *sq, u16 prod_idx, unsigned int wqe_size,
+		       unsigned int cos)
+{
+	struct hinic_wq *wq = sq->wq;
+
+	/* increment prod_idx to the next */
+	prod_idx += ALIGN(wqe_size, wq->wqebb_size) / wq->wqebb_size;
+
+	wmb();  /* Write all before the doorbell */
+
+	writel(sq_prepare_db(sq, prod_idx, cos), SQ_DB_ADDR(sq, prod_idx));
+}
+
+/**
+ * hinic_sq_get_wqe - get wqe ptr in the current pi and update the pi
+ * @sq: sq to get wqe from
+ * @wqe_size: wqe size
+ * @prod_idx: returned pi
+ *
+ * Return wqe pointer
+ **/
+struct hinic_sq_wqe *hinic_sq_get_wqe(struct hinic_sq *sq,
+				      unsigned int wqe_size, u16 *prod_idx)
+{
+	struct hinic_hw_wqe *hw_wqe = hinic_get_wqe(sq->wq, wqe_size,
+						    prod_idx);
+
+	if (IS_ERR(hw_wqe))
+		return NULL;
+
+	return &hw_wqe->sq_wqe;
+}
+
+/**
+ * hinic_sq_write_wqe - write the wqe to the sq
+ * @sq: send queue
+ * @prod_idx: pi of the wqe
+ * @sq_wqe: the wqe to write
+ * @skb: skb to save
+ * @wqe_size: the size of the wqe
+ **/
+void hinic_sq_write_wqe(struct hinic_sq *sq, u16 prod_idx,
+			struct hinic_sq_wqe *sq_wqe,
+			struct sk_buff *skb, unsigned int wqe_size)
+{
+	struct hinic_hw_wqe *hw_wqe = (struct hinic_hw_wqe *)sq_wqe;
+
+	sq->saved_skb[prod_idx] = skb;
+
+	/* The data in the HW should be in Big Endian Format */
+	hinic_cpu_to_be32(sq_wqe, wqe_size);
+
+	hinic_write_wqe(sq->wq, hw_wqe, wqe_size);
+}
+
+/**
+ * hinic_sq_read_wqe - read wqe ptr in the current ci and update the ci
+ * @sq: send queue
+ * @skb: return skb that was saved
+ * @wqe_size: the size of the wqe
+ * @cons_idx: consumer index of the wqe
+ *
+ * Return wqe in ci position
+ **/
+struct hinic_sq_wqe *hinic_sq_read_wqe(struct hinic_sq *sq,
+				       struct sk_buff **skb,
+				       unsigned int *wqe_size, u16 *cons_idx)
+{
+	struct hinic_hw_wqe *hw_wqe;
+	struct hinic_sq_wqe *sq_wqe;
+	struct hinic_sq_ctrl *ctrl;
+	struct hinic_wq *wq;
+	int buf_sect_len;
+	u32 ctrl_info;
+
+	/* read the ctrl section for getting wqe size */
+	hw_wqe = hinic_read_wqe(sq->wq, sizeof(struct hinic_sq_ctrl),
+				cons_idx);
+	if (IS_ERR(hw_wqe))
+		return NULL;
+
+	sq_wqe = &hw_wqe->sq_wqe;
+	ctrl = &sq_wqe->ctrl;
+	ctrl_info = be32_to_cpu(ctrl->ctrl_info);
+	buf_sect_len = HINIC_SQ_CTRL_GET(ctrl_info, BUFDESC_SECT_LEN);
+
+	*wqe_size = sizeof(*ctrl) + sizeof(sq_wqe->task);
+	*wqe_size += SECT_SIZE_FROM_8BYTES(buf_sect_len);
+
+	wq = sq->wq;
+	*wqe_size = ALIGN(*wqe_size, wq->wqebb_size);
+	*skb = sq->saved_skb[*cons_idx];
+
+	/* using the real wqe size to read wqe again */
+	hw_wqe = hinic_read_wqe(wq, *wqe_size, cons_idx);
+
+	return &hw_wqe->sq_wqe;
+}
+
+/**
+ * hinic_sq_put_wqe - release the ci for new wqes
+ * @sq: send queue
+ * @wqe_size: the size of the wqe
+ **/
+void hinic_sq_put_wqe(struct hinic_sq *sq, unsigned int wqe_size)
+{
+	hinic_put_wqe(sq->wq, wqe_size);
+}
+
+/**
+ * hinic_sq_get_sges - get sges from the wqe
+ * @sq_wqe: wqe to get the sges from its buffer addresses
+ * @sges: returned sges
+ * @nr_sges: number sges to return
+ **/
+void hinic_sq_get_sges(struct hinic_sq_wqe *sq_wqe, struct hinic_sge *sges,
+		       int nr_sges)
+{
+	int i;
+
+	for (i = 0; i < nr_sges && i < HINIC_MAX_SQ_BUFDESCS; i++) {
+		sges[i] = sq_wqe->buf_descs[i].sge;
+		hinic_be32_to_cpu(&sges[i], sizeof(sges[i]));
+	}
+}
+
 /**
  * hinic_rq_get_wqe - get wqe ptr in the current pi and update the pi
  * @rq: rq to get wqe from
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
index 1331bd4..0565edc 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
@@ -16,6 +16,7 @@
 #ifndef HINIC_HW_QP_H
 #define HINIC_HW_QP_H
 
+#include <linux/kernel.h>
 #include <linux/types.h>
 #include <linux/sizes.h>
 #include <linux/pci.h>
@@ -27,6 +28,22 @@
 #include "hinic_hw_wq.h"
 #include "hinic_hw_qp_ctxt.h"
 
+#define HINIC_SQ_DB_INFO_PI_HI_SHIFT            0
+#define HINIC_SQ_DB_INFO_QID_SHIFT              8
+#define HINIC_SQ_DB_INFO_PATH_SHIFT             23
+#define HINIC_SQ_DB_INFO_COS_SHIFT              24
+#define HINIC_SQ_DB_INFO_TYPE_SHIFT             27
+
+#define HINIC_SQ_DB_INFO_PI_HI_MASK             0xFF
+#define HINIC_SQ_DB_INFO_QID_MASK               0x3FF
+#define HINIC_SQ_DB_INFO_PATH_MASK              0x1
+#define HINIC_SQ_DB_INFO_COS_MASK               0x7
+#define HINIC_SQ_DB_INFO_TYPE_MASK              0x1F
+
+#define HINIC_SQ_DB_INFO_SET(val, member)       \
+		(((u32)(val) & HINIC_SQ_DB_INFO_##member##_MASK) \
+		 << HINIC_SQ_DB_INFO_##member##_SHIFT)
+
 #define HINIC_SQ_WQEBB_SIZE                     64
 #define HINIC_RQ_WQEBB_SIZE                     32
 
@@ -38,6 +55,12 @@
 
 #define HINIC_RX_BUF_SZ                         2048
 
+#define HINIC_MIN_TX_WQE_SIZE(wq)               \
+		ALIGN(HINIC_SQ_WQE_SIZE(1), (wq)->wqebb_size)
+
+#define HINIC_MIN_TX_NUM_WQEBBS(sq)             \
+		(HINIC_MIN_TX_WQE_SIZE((sq)->wq) / (sq)->wq->wqebb_size)
+
 struct hinic_sq {
 	struct hinic_hwif       *hwif;
 
@@ -101,8 +124,33 @@ int hinic_init_rq(struct hinic_rq *rq, struct hinic_hwif *hwif,
 
 void hinic_clean_rq(struct hinic_rq *rq);
 
+int hinic_get_sq_free_wqebbs(struct hinic_sq *sq);
+
 int hinic_get_rq_free_wqebbs(struct hinic_rq *rq);
 
+void hinic_sq_prepare_wqe(struct hinic_sq *sq, u16 prod_idx,
+			  struct hinic_sq_wqe *wqe, struct hinic_sge *sges,
+			  int nr_sges);
+
+void hinic_sq_write_db(struct hinic_sq *sq, u16 prod_idx, unsigned int wqe_size,
+		       unsigned int cos);
+
+struct hinic_sq_wqe *hinic_sq_get_wqe(struct hinic_sq *sq,
+				      unsigned int wqe_size, u16 *prod_idx);
+
+void hinic_sq_write_wqe(struct hinic_sq *sq, u16 prod_idx,
+			struct hinic_sq_wqe *wqe, struct sk_buff *skb,
+			unsigned int wqe_size);
+
+struct hinic_sq_wqe *hinic_sq_read_wqe(struct hinic_sq *sq,
+				       struct sk_buff **skb,
+				       unsigned int *wqe_size, u16 *cons_idx);
+
+void hinic_sq_put_wqe(struct hinic_sq *sq, unsigned int wqe_size);
+
+void hinic_sq_get_sges(struct hinic_sq_wqe *wqe, struct hinic_sge *sges,
+		       int nr_sges);
+
 struct hinic_rq_wqe *hinic_rq_get_wqe(struct hinic_rq *rq,
 				      unsigned int wqe_size, u16 *prod_idx);
 
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_main.c b/drivers/net/ethernet/huawei/hinic/hinic_main.c
index 053e5f3..4a5f23f 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_main.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_main.c
@@ -43,6 +43,10 @@
 MODULE_DESCRIPTION("Huawei Intelligent NIC driver");
 MODULE_LICENSE("GPL");
 
+static unsigned int tx_weight = 64;
+module_param(tx_weight, uint, 0644);
+MODULE_PARM_DESC(tx_weight, "Number Tx packets for NAPI budget (default=64)");
+
 static unsigned int rx_weight = 64;
 module_param(rx_weight, uint, 0644);
 MODULE_PARM_DESC(rx_weight, "Number Rx packets for NAPI budget (default=64)");
@@ -569,9 +573,11 @@ static void hinic_set_rx_mode(struct net_device *netdev)
 	queue_work(nic_dev->workq, &rx_mode_work->work);
 }
 
-netdev_tx_t hinic_xmit_frame(struct sk_buff *skb, struct net_device *netdev)
+static void hinic_tx_timeout(struct net_device *netdev)
 {
-	return NETDEV_TX_BUSY;
+	struct hinic_dev *nic_dev = netdev_priv(netdev);
+
+	netif_err(nic_dev, drv, netdev, "Tx timeout\n");
 }
 
 static const struct net_device_ops hinic_netdev_ops = {
@@ -584,6 +590,7 @@ netdev_tx_t hinic_xmit_frame(struct sk_buff *skb, struct net_device *netdev)
 	.ndo_vlan_rx_kill_vid = hinic_vlan_rx_kill_vid,
 	.ndo_set_rx_mode = hinic_set_rx_mode,
 	.ndo_start_xmit = hinic_xmit_frame,
+	.ndo_tx_timeout = hinic_tx_timeout,
 	/* more operations should be filled */
 };
 
@@ -690,6 +697,7 @@ static int nic_dev_init(struct pci_dev *pdev)
 	nic_dev->flags = 0;
 	nic_dev->txqs = NULL;
 	nic_dev->rxqs = NULL;
+	nic_dev->tx_weight = tx_weight;
 	nic_dev->rx_weight = rx_weight;
 
 	sema_init(&nic_dev->mgmt_lock, 1);
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_tx.c b/drivers/net/ethernet/huawei/hinic/hinic_tx.c
index 9c27fb2..d0c539f 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_tx.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_tx.c
@@ -13,12 +13,42 @@
  *
  */
 
+#include <linux/kernel.h>
 #include <linux/netdevice.h>
 #include <linux/u64_stats_sync.h>
+#include <linux/errno.h>
+#include <linux/types.h>
+#include <linux/pci.h>
+#include <linux/device.h>
+#include <linux/dma-mapping.h>
+#include <linux/slab.h>
+#include <linux/interrupt.h>
+#include <linux/skbuff.h>
+#include <linux/smp.h>
+#include <asm/byteorder.h>
 
+#include "hinic_common.h"
+#include "hinic_hw_if.h"
+#include "hinic_hw_wqe.h"
+#include "hinic_hw_wq.h"
 #include "hinic_hw_qp.h"
+#include "hinic_hw_dev.h"
+#include "hinic_dev.h"
 #include "hinic_tx.h"
 
+#define TX_IRQ_NO_PENDING               0
+#define TX_IRQ_NO_COALESC               0
+#define TX_IRQ_NO_LLI_TIMER             0
+#define TX_IRQ_NO_CREDIT                0
+#define TX_IRQ_NO_RESEND_TIMER          0
+
+#define CI_UPDATE_NO_PENDING            0
+#define CI_UPDATE_NO_COALESC            0
+
+#define HW_CONS_IDX(sq)         be16_to_cpu(*(u16 *)((sq)->hw_ci_addr))
+
+#define MIN_SKB_LEN             64
+
 /**
  * hinic_txq_clean_stats - Clean the statistics of specific queue
  * @txq: Logical Tx Queue
@@ -49,6 +79,321 @@ static void txq_stats_init(struct hinic_txq *txq)
 }
 
 /**
+ * tx_map_skb - dma mapping for skb and return sges
+ * @nic_dev: nic device
+ * @skb: the skb
+ * @sges: returned sges
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int tx_map_skb(struct hinic_dev *nic_dev, struct sk_buff *skb,
+		      struct hinic_sge *sges)
+{
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	struct skb_frag_struct *frag;
+	dma_addr_t dma_addr;
+	int i, j;
+
+	dma_addr = dma_map_single(&pdev->dev, skb->data, skb_headlen(skb),
+				  DMA_TO_DEVICE);
+	if (dma_mapping_error(&pdev->dev, dma_addr)) {
+		dev_err(&pdev->dev, "Failed to map Tx skb data\n");
+		return -EFAULT;
+	}
+
+	hinic_set_sge(&sges[0], dma_addr, skb_headlen(skb));
+
+	for (i = 0 ; i < skb_shinfo(skb)->nr_frags; i++) {
+		frag = &skb_shinfo(skb)->frags[i];
+
+		dma_addr = skb_frag_dma_map(&pdev->dev, frag, 0,
+					    skb_frag_size(frag),
+					    DMA_TO_DEVICE);
+		if (dma_mapping_error(&pdev->dev, dma_addr)) {
+			dev_err(&pdev->dev, "Failed to map Tx skb frag\n");
+			goto err_tx_map;
+		}
+
+		hinic_set_sge(&sges[i + 1], dma_addr, skb_frag_size(frag));
+	}
+
+	return 0;
+
+err_tx_map:
+	for (j = 0; j < i; j++)
+		dma_unmap_page(&pdev->dev, hinic_sge_to_dma(&sges[j + 1]),
+			       sges[j + 1].len, DMA_TO_DEVICE);
+
+	dma_unmap_single(&pdev->dev, hinic_sge_to_dma(&sges[0]), sges[0].len,
+			 DMA_TO_DEVICE);
+	return -EFAULT;
+}
+
+/**
+ * tx_unmap_skb - unmap the dma address of the skb
+ * @nic_dev: nic device
+ * @skb: the skb
+ * @sges: the sges that are connected to the skb
+ **/
+static void tx_unmap_skb(struct hinic_dev *nic_dev, struct sk_buff *skb,
+			 struct hinic_sge *sges)
+{
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	int i;
+
+	for (i = 0; i < skb_shinfo(skb)->nr_frags ; i++)
+		dma_unmap_page(&pdev->dev, hinic_sge_to_dma(&sges[i + 1]),
+			       sges[i + 1].len, DMA_TO_DEVICE);
+
+	dma_unmap_single(&pdev->dev, hinic_sge_to_dma(&sges[0]), sges[0].len,
+			 DMA_TO_DEVICE);
+}
+
+netdev_tx_t hinic_xmit_frame(struct sk_buff *skb, struct net_device *netdev)
+{
+	struct hinic_dev *nic_dev = netdev_priv(netdev);
+	struct netdev_queue *netdev_txq;
+	int nr_sges, err = NETDEV_TX_OK;
+	struct hinic_sq_wqe *sq_wqe;
+	unsigned int wqe_size;
+	struct hinic_txq *txq;
+	struct hinic_qp *qp;
+	u16 prod_idx;
+
+	txq = &nic_dev->txqs[skb->queue_mapping];
+	qp = container_of(txq->sq, struct hinic_qp, sq);
+
+	if (skb->len < MIN_SKB_LEN) {
+		if (skb_pad(skb, MIN_SKB_LEN - skb->len)) {
+			netdev_err(netdev, "Failed to pad skb\n");
+			goto skb_error;
+		}
+
+		skb->len = MIN_SKB_LEN;
+	}
+
+	nr_sges = skb_shinfo(skb)->nr_frags + 1;
+	if (nr_sges > txq->max_sges) {
+		netdev_err(netdev, "Too many Tx sges\n");
+		goto skb_error;
+	}
+
+	err = tx_map_skb(nic_dev, skb, txq->sges);
+	if (err)
+		goto skb_error;
+
+	wqe_size = HINIC_SQ_WQE_SIZE(nr_sges);
+
+	sq_wqe = hinic_sq_get_wqe(txq->sq, wqe_size, &prod_idx);
+	if (!sq_wqe) {
+		tx_unmap_skb(nic_dev, skb, txq->sges);
+
+		netif_stop_subqueue(netdev, qp->q_id);
+
+		u64_stats_update_begin(&txq->txq_stats.syncp);
+		txq->txq_stats.tx_busy++;
+		u64_stats_update_end(&txq->txq_stats.syncp);
+		err = NETDEV_TX_BUSY;
+		goto flush_skbs;
+	}
+
+	hinic_sq_prepare_wqe(txq->sq, prod_idx, sq_wqe, txq->sges, nr_sges);
+
+	hinic_sq_write_wqe(txq->sq, prod_idx, sq_wqe, skb, wqe_size);
+
+flush_skbs:
+	netdev_txq = netdev_get_tx_queue(netdev, skb->queue_mapping);
+	if ((!skb->xmit_more) || (netif_xmit_stopped(netdev_txq)))
+		hinic_sq_write_db(txq->sq, prod_idx, wqe_size, 0);
+	return err;
+
+skb_error:
+	dev_kfree_skb_any(skb);
+
+	u64_stats_update_begin(&txq->txq_stats.syncp);
+	txq->txq_stats.tx_dropped++;
+	u64_stats_update_end(&txq->txq_stats.syncp);
+	return err;
+}
+
+/**
+ * tx_free_skb - unmap and free skb
+ * @nic_dev: nic device
+ * @skb: the skb
+ * @sges: the sges that are connected to the skb
+ **/
+static void tx_free_skb(struct hinic_dev *nic_dev, struct sk_buff *skb,
+			struct hinic_sge *sges)
+{
+	tx_unmap_skb(nic_dev, skb, sges);
+
+	dev_kfree_skb_any(skb);
+}
+
+/**
+ * free_all_rx_skbs - free all skbs in tx queue
+ * @txq: tx queue
+ **/
+static void free_all_tx_skbs(struct hinic_txq *txq)
+{
+	struct hinic_dev *nic_dev = netdev_priv(txq->netdev);
+	struct hinic_sq *sq = txq->sq;
+	struct hinic_sq_wqe *sq_wqe;
+	unsigned int wqe_size;
+	struct sk_buff *skb;
+	int nr_sges;
+	u16 ci;
+
+	while ((sq_wqe = hinic_sq_read_wqe(sq, &skb, &wqe_size, &ci))) {
+		nr_sges = skb_shinfo(skb)->nr_frags + 1;
+
+		hinic_sq_get_sges(sq_wqe, txq->free_sges, nr_sges);
+
+		hinic_sq_put_wqe(sq, wqe_size);
+
+		tx_free_skb(nic_dev, skb, txq->free_sges);
+	}
+}
+
+/**
+ * free_tx_poll - free finished tx skbs in tx queue that connected to napi
+ * @napi: napi
+ * @budget: number of tx
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int free_tx_poll(struct napi_struct *napi, int budget)
+{
+	struct hinic_txq *txq = container_of(napi, struct hinic_txq, napi);
+	struct hinic_qp *qp = container_of(txq->sq, struct hinic_qp, sq);
+	struct hinic_dev *nic_dev = netdev_priv(txq->netdev);
+	struct netdev_queue *netdev_txq;
+	struct hinic_sq *sq = txq->sq;
+	struct hinic_wq *wq = sq->wq;
+	struct hinic_sq_wqe *sq_wqe;
+	unsigned int wqe_size;
+	int nr_sges, pkts = 0;
+	struct sk_buff *skb;
+	u64 tx_bytes = 0;
+	u16 hw_ci, sw_ci;
+
+	do {
+		hw_ci = HW_CONS_IDX(sq) & wq->mask;
+
+		sq_wqe = hinic_sq_read_wqe(sq, &skb, &wqe_size, &sw_ci);
+		if ((!sq_wqe) ||
+		    (((hw_ci - sw_ci) & wq->mask) * wq->wqebb_size < wqe_size))
+			break;
+
+		tx_bytes += skb->len;
+		pkts++;
+
+		nr_sges = skb_shinfo(skb)->nr_frags + 1;
+
+		hinic_sq_get_sges(sq_wqe, txq->free_sges, nr_sges);
+
+		hinic_sq_put_wqe(sq, wqe_size);
+
+		tx_free_skb(nic_dev, skb, txq->free_sges);
+	} while (pkts < budget);
+
+	if (__netif_subqueue_stopped(nic_dev->netdev, qp->q_id) &&
+	    hinic_get_sq_free_wqebbs(sq) >= HINIC_MIN_TX_NUM_WQEBBS(sq)) {
+		netdev_txq = netdev_get_tx_queue(txq->netdev, qp->q_id);
+
+		__netif_tx_lock(netdev_txq, smp_processor_id());
+
+		netif_wake_subqueue(nic_dev->netdev, qp->q_id);
+
+		__netif_tx_unlock(netdev_txq);
+
+		u64_stats_update_begin(&txq->txq_stats.syncp);
+		txq->txq_stats.tx_wake++;
+		u64_stats_update_end(&txq->txq_stats.syncp);
+	}
+
+	u64_stats_update_begin(&txq->txq_stats.syncp);
+	txq->txq_stats.bytes += tx_bytes;
+	txq->txq_stats.pkts += pkts;
+	u64_stats_update_end(&txq->txq_stats.syncp);
+
+	if (pkts < budget) {
+		napi_complete(napi);
+		enable_irq(sq->irq);
+		return pkts;
+	}
+
+	return budget;
+}
+
+static void tx_napi_add(struct hinic_txq *txq, int weight)
+{
+	netif_napi_add(txq->netdev, &txq->napi, free_tx_poll, weight);
+	napi_enable(&txq->napi);
+}
+
+static void tx_napi_del(struct hinic_txq *txq)
+{
+	napi_disable(&txq->napi);
+	netif_napi_del(&txq->napi);
+}
+
+static irqreturn_t tx_irq(int irq, void *data)
+{
+	struct hinic_txq *txq = (struct hinic_txq *)data;
+	struct hinic_sq *sq = txq->sq;
+	struct hinic_dev *nic_dev;
+
+	nic_dev = netdev_priv(txq->netdev);
+
+	/* Disable the interrupt until napi will be completed */
+	disable_irq_nosync(sq->irq);
+
+	hinic_hwdev_msix_cnt_set(nic_dev->hwdev, sq->msix_entry);
+
+	napi_schedule(&txq->napi);
+	return IRQ_HANDLED;
+}
+
+static int tx_request_irq(struct hinic_txq *txq)
+{
+	struct hinic_dev *nic_dev = netdev_priv(txq->netdev);
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_sq *sq = txq->sq;
+	int err;
+
+	tx_napi_add(txq, nic_dev->tx_weight);
+
+	hinic_hwdev_msix_set(nic_dev->hwdev, sq->msix_entry,
+			     TX_IRQ_NO_PENDING, TX_IRQ_NO_COALESC,
+			     TX_IRQ_NO_LLI_TIMER, TX_IRQ_NO_CREDIT,
+			     TX_IRQ_NO_RESEND_TIMER);
+
+	err = request_irq(sq->irq, tx_irq, 0, txq->irq_name, txq);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to request Tx irq\n");
+		tx_napi_del(txq);
+		return err;
+	}
+
+	return 0;
+}
+
+static void tx_free_irq(struct hinic_txq *txq)
+{
+	struct hinic_sq *sq = txq->sq;
+
+	free_irq(sq->irq, txq);
+	tx_napi_del(txq);
+}
+
+/**
  * hinic_init_txq - Initialize the Tx Queue
  * @txq: Logical Tx Queue
  * @sq: Hardware Tx Queue to connect the Logical queue with
@@ -59,11 +404,63 @@ static void txq_stats_init(struct hinic_txq *txq)
 int hinic_init_txq(struct hinic_txq *txq, struct hinic_sq *sq,
 		   struct net_device *netdev)
 {
+	struct hinic_qp *qp = container_of(sq, struct hinic_qp, sq);
+	struct hinic_dev *nic_dev = netdev_priv(netdev);
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	int err, irqname_len;
+	size_t sges_size;
+
 	txq->netdev = netdev;
 	txq->sq = sq;
 
 	txq_stats_init(txq);
+
+	txq->max_sges = HINIC_MAX_SQ_BUFDESCS;
+
+	sges_size = txq->max_sges * sizeof(*txq->sges);
+	txq->sges = devm_kzalloc(&netdev->dev, sges_size, GFP_KERNEL);
+	if (!txq->sges)
+		return -ENOMEM;
+
+	sges_size = txq->max_sges * sizeof(*txq->free_sges);
+	txq->free_sges = devm_kzalloc(&netdev->dev, sges_size, GFP_KERNEL);
+	if (!txq->free_sges) {
+		err = -ENOMEM;
+		goto err_alloc_free_sges;
+	}
+
+	irqname_len = snprintf(NULL, 0, "hinic_txq%d", qp->q_id) + 1;
+	txq->irq_name = devm_kzalloc(&netdev->dev, irqname_len, GFP_KERNEL);
+	if (!txq->irq_name) {
+		err = -ENOMEM;
+		goto err_alloc_irqname;
+	}
+
+	sprintf(txq->irq_name, "hinic_txq%d", qp->q_id);
+
+	err = hinic_hwdev_hw_ci_addr_set(hwdev, sq, CI_UPDATE_NO_PENDING,
+					 CI_UPDATE_NO_COALESC);
+	if (err)
+		goto err_hw_ci;
+
+	err = tx_request_irq(txq);
+	if (err) {
+		netdev_err(netdev, "Failed to request Tx irq\n");
+		goto err_req_tx_irq;
+	}
+
 	return 0;
+
+err_req_tx_irq:
+err_hw_ci:
+	devm_kfree(&netdev->dev, txq->irq_name);
+
+err_alloc_irqname:
+	devm_kfree(&netdev->dev, txq->free_sges);
+
+err_alloc_free_sges:
+	devm_kfree(&netdev->dev, txq->sges);
+	return err;
 }
 
 /**
@@ -72,4 +469,13 @@ int hinic_init_txq(struct hinic_txq *txq, struct hinic_sq *sq,
  **/
 void hinic_clean_txq(struct hinic_txq *txq)
 {
+	struct net_device *netdev = txq->netdev;
+
+	tx_free_irq(txq);
+
+	free_all_tx_skbs(txq);
+
+	devm_kfree(&netdev->dev, txq->irq_name);
+	devm_kfree(&netdev->dev, txq->free_sges);
+	devm_kfree(&netdev->dev, txq->sges);
 }
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_tx.h b/drivers/net/ethernet/huawei/hinic/hinic_tx.h
index bbdb4b6..7123c7f 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_tx.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_tx.h
@@ -18,8 +18,10 @@
 
 #include <linux/types.h>
 #include <linux/netdevice.h>
+#include <linux/skbuff.h>
 #include <linux/u64_stats_sync.h>
 
+#include "hinic_common.h"
 #include "hinic_hw_qp.h"
 
 struct hinic_txq_stats {
@@ -37,10 +39,19 @@ struct hinic_txq {
 	struct hinic_sq         *sq;
 
 	struct hinic_txq_stats  txq_stats;
+
+	int                     max_sges;
+	struct hinic_sge        *sges;
+	struct hinic_sge        *free_sges;
+
+	char                    *irq_name;
+	struct napi_struct      napi;
 };
 
 void hinic_txq_clean_stats(struct hinic_txq *txq);
 
+netdev_tx_t hinic_xmit_frame(struct sk_buff *skb, struct net_device *netdev);
+
 int hinic_init_txq(struct hinic_txq *txq, struct hinic_sq *sq,
 		   struct net_device *netdev);
 
-- 
1.9.1

^ permalink raw reply related

* [PATCH V7 net-next 18/22] net-next/hinic: Add Rx handler
From: Aviad Krawczyk @ 2017-08-17 19:04 UTC (permalink / raw)
  To: davem
  Cc: linux-kernel, netdev, bc.y, victor.gissin, aviad.krawczyk,
	zhaochen6, tony.qu
In-Reply-To: <cover.1502996306.git.aviad.krawczyk@huawei.com>

Set the io resources in the nic and handle rx events by qp operations.

Signed-off-by: Aviad Krawczyk <aviad.krawczyk@huawei.com>
Signed-off-by: Zhao Chen <zhaochen6@huawei.com>
---
 drivers/net/ethernet/huawei/hinic/hinic_dev.h     |   1 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h  |   1 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c  | 361 +++++++++++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h  |  77 ++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_if.c   |  36 ++
 drivers/net/ethernet/huawei/hinic/hinic_hw_if.h   |  35 ++
 drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.h |  13 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c   | 210 +++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h   |  29 ++
 drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c   |  12 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h   |   2 +
 drivers/net/ethernet/huawei/hinic/hinic_main.c    |  27 ++
 drivers/net/ethernet/huawei/hinic/hinic_port.c    |  32 ++
 drivers/net/ethernet/huawei/hinic/hinic_port.h    |  19 +
 drivers/net/ethernet/huawei/hinic/hinic_rx.c      | 419 ++++++++++++++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_rx.h      |   7 +
 16 files changed, 1281 insertions(+)

diff --git a/drivers/net/ethernet/huawei/hinic/hinic_dev.h b/drivers/net/ethernet/huawei/hinic/hinic_dev.h
index 5b8231d..3d0f6cf 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_dev.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_dev.h
@@ -43,6 +43,7 @@ struct hinic_dev {
 	struct hinic_hwdev              *hwdev;
 
 	u32                             msg_enable;
+	unsigned int                    rx_weight;
 
 	unsigned int                    flags;
 
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h
index 10b8c7b..f39b184 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h
@@ -20,6 +20,7 @@
 #define HINIC_CSR_FUNC_ATTR0_ADDR                       0x0
 #define HINIC_CSR_FUNC_ATTR1_ADDR                       0x4
 
+#define HINIC_CSR_FUNC_ATTR4_ADDR                       0x10
 #define HINIC_CSR_FUNC_ATTR5_ADDR                       0x14
 
 #define HINIC_DMA_ATTR_BASE                             0xC80
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
index cb4c472..03149e8 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
@@ -20,6 +20,9 @@
 #include <linux/errno.h>
 #include <linux/slab.h>
 #include <linux/bitops.h>
+#include <linux/delay.h>
+#include <linux/jiffies.h>
+#include <linux/log2.h>
 #include <linux/err.h>
 
 #include "hinic_hw_if.h"
@@ -30,6 +33,10 @@
 #include "hinic_hw_io.h"
 #include "hinic_hw_dev.h"
 
+#define IO_STATUS_TIMEOUT               100
+#define OUTBOUND_STATE_TIMEOUT          100
+#define DB_STATE_TIMEOUT                100
+
 #define MAX_IRQS(max_qps, num_aeqs, num_ceqs)   \
 		 (2 * (max_qps) + (num_aeqs) + (num_ceqs))
 
@@ -37,6 +44,15 @@ enum intr_type {
 	INTR_MSIX_TYPE,
 };
 
+enum io_status {
+	IO_STOPPED = 0,
+	IO_RUNNING = 1,
+};
+
+enum hw_ioctxt_set_cmdq_depth {
+	HW_IOCTXT_SET_CMDQ_DEPTH_DEFAULT,
+};
+
 /* HW struct */
 struct hinic_dev_cap {
 	u8      status;
@@ -51,6 +67,31 @@ struct hinic_dev_cap {
 	u8      rsvd3[208];
 };
 
+struct rx_buf_sz {
+	int     idx;
+	size_t  sz;
+};
+
+static struct rx_buf_sz rx_buf_sz_table[] = {
+	{0, 32},
+	{1, 64},
+	{2, 96},
+	{3, 128},
+	{4, 192},
+	{5, 256},
+	{6, 384},
+	{7, 512},
+	{8, 768},
+	{9, 1024},
+	{10, 1536},
+	{11, 2048},
+	{12, 3072},
+	{13, 4096},
+	{14, 8192},
+	{15, 16384},
+	{-1, -1},
+};
+
 /**
  * get_capability - convert device capabilities to NIC capabilities
  * @hwdev: the HW device to set and convert device capabilities for
@@ -236,6 +277,252 @@ int hinic_port_msg_cmd(struct hinic_hwdev *hwdev, enum hinic_port_cmd cmd,
 }
 
 /**
+ * init_fw_ctxt- Init Firmware tables before network mgmt and io operations
+ * @hwdev: the NIC HW device
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int init_fw_ctxt(struct hinic_hwdev *hwdev)
+{
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_cmd_fw_ctxt fw_ctxt;
+	struct hinic_pfhwdev *pfhwdev;
+	u16 out_size;
+	int err;
+
+	if (!HINIC_IS_PF(hwif) && !HINIC_IS_PPF(hwif)) {
+		dev_err(&pdev->dev, "Unsupported PCI Function type\n");
+		return -EINVAL;
+	}
+
+	fw_ctxt.func_idx = HINIC_HWIF_FUNC_IDX(hwif);
+	fw_ctxt.rx_buf_sz = HINIC_RX_BUF_SZ;
+
+	pfhwdev = container_of(hwdev, struct hinic_pfhwdev, hwdev);
+
+	err = hinic_port_msg_cmd(hwdev, HINIC_PORT_CMD_FWCTXT_INIT,
+				 &fw_ctxt, sizeof(fw_ctxt),
+				 &fw_ctxt, &out_size);
+	if (err || (out_size != sizeof(fw_ctxt)) || fw_ctxt.status) {
+		dev_err(&pdev->dev, "Failed to init FW ctxt, ret = %d\n",
+			fw_ctxt.status);
+		return -EFAULT;
+	}
+
+	return 0;
+}
+
+/**
+ * set_hw_ioctxt - set the shape of the IO queues in FW
+ * @hwdev: the NIC HW device
+ * @rq_depth: rq depth
+ * @sq_depth: sq depth
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int set_hw_ioctxt(struct hinic_hwdev *hwdev, unsigned int rq_depth,
+			 unsigned int sq_depth)
+{
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct hinic_cmd_hw_ioctxt hw_ioctxt;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_pfhwdev *pfhwdev;
+	int i;
+
+	if (!HINIC_IS_PF(hwif) && !HINIC_IS_PPF(hwif)) {
+		dev_err(&pdev->dev, "Unsupported PCI Function type\n");
+		return -EINVAL;
+	}
+
+	hw_ioctxt.func_idx = HINIC_HWIF_FUNC_IDX(hwif);
+
+	hw_ioctxt.set_cmdq_depth = HW_IOCTXT_SET_CMDQ_DEPTH_DEFAULT;
+	hw_ioctxt.cmdq_depth = 0;
+
+	hw_ioctxt.rq_depth  = ilog2(rq_depth);
+
+	for (i = 0; ; i++) {
+		if ((rx_buf_sz_table[i].sz == HINIC_RX_BUF_SZ) ||
+		    (rx_buf_sz_table[i].sz == -1)) {
+			hw_ioctxt.rx_buf_sz_idx = rx_buf_sz_table[i].idx;
+			break;
+		}
+	}
+
+	if (hw_ioctxt.rx_buf_sz_idx == -1)
+		return -EINVAL;
+
+	hw_ioctxt.sq_depth  = ilog2(sq_depth);
+
+	pfhwdev = container_of(hwdev, struct hinic_pfhwdev, hwdev);
+
+	return hinic_msg_to_mgmt(&pfhwdev->pf_to_mgmt, HINIC_MOD_COMM,
+				 HINIC_COMM_CMD_HWCTXT_SET,
+				 &hw_ioctxt, sizeof(hw_ioctxt), NULL,
+				 NULL, HINIC_MGMT_MSG_SYNC);
+}
+
+static int wait_for_outbound_state(struct hinic_hwdev *hwdev)
+{
+	enum hinic_outbound_state outbound_state;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	unsigned long end;
+
+	end = jiffies + msecs_to_jiffies(OUTBOUND_STATE_TIMEOUT);
+	do {
+		outbound_state = hinic_outbound_state_get(hwif);
+
+		if (outbound_state == HINIC_OUTBOUND_ENABLE)
+			return 0;
+
+		msleep(20);
+	} while (time_before(jiffies, end));
+
+	dev_err(&pdev->dev, "Wait for OUTBOUND - Timeout\n");
+	return -EFAULT;
+}
+
+static int wait_for_db_state(struct hinic_hwdev *hwdev)
+{
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	enum hinic_db_state db_state;
+	unsigned long end;
+
+	end = jiffies + msecs_to_jiffies(DB_STATE_TIMEOUT);
+	do {
+		db_state = hinic_db_state_get(hwif);
+
+		if (db_state == HINIC_DB_ENABLE)
+			return 0;
+
+		msleep(20);
+	} while (time_before(jiffies, end));
+
+	dev_err(&pdev->dev, "Wait for DB - Timeout\n");
+	return -EFAULT;
+}
+
+static int wait_for_io_stopped(struct hinic_hwdev *hwdev)
+{
+	struct hinic_cmd_io_status cmd_io_status;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_pfhwdev *pfhwdev;
+	unsigned long end;
+	u16 out_size;
+	int err;
+
+	if (!HINIC_IS_PF(hwif) && !HINIC_IS_PPF(hwif)) {
+		dev_err(&pdev->dev, "Unsupported PCI Function type\n");
+		return -EINVAL;
+	}
+
+	pfhwdev = container_of(hwdev, struct hinic_pfhwdev, hwdev);
+
+	cmd_io_status.func_idx = HINIC_HWIF_FUNC_IDX(hwif);
+
+	end = jiffies + msecs_to_jiffies(IO_STATUS_TIMEOUT);
+	do {
+		err = hinic_msg_to_mgmt(&pfhwdev->pf_to_mgmt, HINIC_MOD_COMM,
+					HINIC_COMM_CMD_IO_STATUS_GET,
+					&cmd_io_status, sizeof(cmd_io_status),
+					&cmd_io_status, &out_size,
+					HINIC_MGMT_MSG_SYNC);
+		if ((err) || (out_size != sizeof(cmd_io_status))) {
+			dev_err(&pdev->dev, "Failed to get IO status, ret = %d\n",
+				err);
+			return err;
+		}
+
+		if (cmd_io_status.status == IO_STOPPED) {
+			dev_info(&pdev->dev, "IO stopped\n");
+			return 0;
+		}
+
+		msleep(20);
+	} while (time_before(jiffies, end));
+
+	dev_err(&pdev->dev, "Wait for IO stopped - Timeout\n");
+	return -ETIMEDOUT;
+}
+
+/**
+ * clear_io_resource - set the IO resources as not active in the NIC
+ * @hwdev: the NIC HW device
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int clear_io_resources(struct hinic_hwdev *hwdev)
+{
+	struct hinic_cmd_clear_io_res cmd_clear_io_res;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_pfhwdev *pfhwdev;
+	int err;
+
+	if (!HINIC_IS_PF(hwif) && !HINIC_IS_PPF(hwif)) {
+		dev_err(&pdev->dev, "Unsupported PCI Function type\n");
+		return -EINVAL;
+	}
+
+	err = wait_for_io_stopped(hwdev);
+	if (err) {
+		dev_err(&pdev->dev, "IO has not stopped yet\n");
+		return err;
+	}
+
+	cmd_clear_io_res.func_idx = HINIC_HWIF_FUNC_IDX(hwif);
+
+	pfhwdev = container_of(hwdev, struct hinic_pfhwdev, hwdev);
+
+	err = hinic_msg_to_mgmt(&pfhwdev->pf_to_mgmt, HINIC_MOD_COMM,
+				HINIC_COMM_CMD_IO_RES_CLEAR, &cmd_clear_io_res,
+				sizeof(cmd_clear_io_res), NULL, NULL,
+				HINIC_MGMT_MSG_SYNC);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to clear IO resources\n");
+		return err;
+	}
+
+	return 0;
+}
+
+/**
+ * set_resources_state - set the state of the resources in the NIC
+ * @hwdev: the NIC HW device
+ * @state: the state to set
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int set_resources_state(struct hinic_hwdev *hwdev,
+			       enum hinic_res_state state)
+{
+	struct hinic_cmd_set_res_state res_state;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_pfhwdev *pfhwdev;
+
+	if (!HINIC_IS_PF(hwif) && !HINIC_IS_PPF(hwif)) {
+		dev_err(&pdev->dev, "Unsupported PCI Function type\n");
+		return -EINVAL;
+	}
+
+	res_state.func_idx = HINIC_HWIF_FUNC_IDX(hwif);
+	res_state.state = state;
+
+	pfhwdev = container_of(hwdev, struct hinic_pfhwdev, hwdev);
+
+	return hinic_msg_to_mgmt(&pfhwdev->pf_to_mgmt,
+				 HINIC_MOD_COMM,
+				 HINIC_COMM_CMD_RES_STATE_SET,
+				 &res_state, sizeof(res_state), NULL,
+				 NULL, HINIC_MGMT_MSG_SYNC);
+}
+
+/**
  * get_base_qpn - get the first qp number
  * @hwdev: the NIC HW device
  * @base_qpn: returned qp number
@@ -312,8 +599,23 @@ int hinic_hwdev_ifup(struct hinic_hwdev *hwdev)
 		goto err_create_qps;
 	}
 
+	err = wait_for_db_state(hwdev);
+	if (err) {
+		dev_warn(&pdev->dev, "db - disabled, try again\n");
+		hinic_db_state_set(hwif, HINIC_DB_ENABLE);
+	}
+
+	err = set_hw_ioctxt(hwdev, HINIC_SQ_DEPTH, HINIC_RQ_DEPTH);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to set HW IO ctxt\n");
+		goto err_hw_ioctxt;
+	}
+
 	return 0;
 
+err_hw_ioctxt:
+	hinic_io_destroy_qps(func_to_io, num_qps);
+
 err_create_qps:
 	hinic_io_free(func_to_io);
 	return err;
@@ -329,6 +631,8 @@ void hinic_hwdev_ifdown(struct hinic_hwdev *hwdev)
 	struct hinic_func_to_io *func_to_io = &hwdev->func_to_io;
 	struct hinic_cap *nic_cap = &hwdev->nic_cap;
 
+	clear_io_resources(hwdev);
+
 	hinic_io_destroy_qps(func_to_io, nic_cap->num_qps);
 	hinic_io_free(func_to_io);
 }
@@ -528,6 +832,12 @@ struct hinic_hwdev *hinic_init_hwdev(struct pci_dev *pdev)
 		goto err_init_msix;
 	}
 
+	err = wait_for_outbound_state(hwdev);
+	if (err) {
+		dev_warn(&pdev->dev, "outbound - disabled, try again\n");
+		hinic_outbound_state_set(hwif, HINIC_OUTBOUND_ENABLE);
+	}
+
 	num_aeqs = HINIC_HWIF_NUM_AEQS(hwif);
 
 	err = hinic_aeqs_init(&hwdev->aeqs, hwif, num_aeqs,
@@ -550,8 +860,22 @@ struct hinic_hwdev *hinic_init_hwdev(struct pci_dev *pdev)
 		goto err_dev_cap;
 	}
 
+	err = init_fw_ctxt(hwdev);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to init function table\n");
+		goto err_init_fw_ctxt;
+	}
+
+	err = set_resources_state(hwdev, HINIC_RES_ACTIVE);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to set resources state\n");
+		goto err_resources_state;
+	}
+
 	return hwdev;
 
+err_resources_state:
+err_init_fw_ctxt:
 err_dev_cap:
 	free_pfhwdev(pfhwdev);
 
@@ -578,6 +902,8 @@ void hinic_free_hwdev(struct hinic_hwdev *hwdev)
 						     struct hinic_pfhwdev,
 						     hwdev);
 
+	set_resources_state(hwdev, HINIC_RES_CLEAN);
+
 	free_pfhwdev(pfhwdev);
 
 	hinic_aeqs_free(&hwdev->aeqs);
@@ -635,3 +961,38 @@ struct hinic_rq *hinic_hwdev_get_rq(struct hinic_hwdev *hwdev, int i)
 
 	return &qp->rq;
 }
+
+/**
+ * hinic_hwdev_msix_cnt_set - clear message attribute counters for msix entry
+ * @hwdev: the NIC HW device
+ * @msix_index: msix_index
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_hwdev_msix_cnt_set(struct hinic_hwdev *hwdev, u16 msix_index)
+{
+	return hinic_msix_attr_cnt_clear(hwdev->hwif, msix_index);
+}
+
+/**
+ * hinic_hwdev_msix_set - set message attribute for msix entry
+ * @hwdev: the NIC HW device
+ * @msix_index: msix_index
+ * @pending_limit: the maximum pending interrupt events (unit 8)
+ * @coalesc_timer: coalesc period for interrupt (unit 8 us)
+ * @lli_timer: replenishing period for low latency credit (unit 8 us)
+ * @lli_credit_limit: maximum credits for low latency msix messages (unit 8)
+ * @resend_timer: maximum wait for resending msix (unit coalesc period)
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_hwdev_msix_set(struct hinic_hwdev *hwdev, u16 msix_index,
+			 u8 pending_limit, u8 coalesc_timer,
+			 u8 lli_timer_cfg, u8 lli_credit_limit,
+			 u8 resend_timer)
+{
+	return hinic_msix_attr_set(hwdev->hwif, msix_index,
+				   pending_limit, coalesc_timer,
+				   lli_timer_cfg, lli_credit_limit,
+				   resend_timer);
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h
index 81c2c6e..e7277d1 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h
@@ -74,6 +74,76 @@ enum hinic_cb_state {
 	HINIC_CB_RUNNING = BIT(1),
 };
 
+enum hinic_res_state {
+	HINIC_RES_CLEAN         = 0,
+	HINIC_RES_ACTIVE        = 1,
+};
+
+struct hinic_cmd_fw_ctxt {
+	u8      status;
+	u8      version;
+	u8      rsvd0[6];
+
+	u16     func_idx;
+	u16     rx_buf_sz;
+
+	u32     rsvd1;
+};
+
+struct hinic_cmd_hw_ioctxt {
+	u8      status;
+	u8      version;
+	u8      rsvd0[6];
+
+	u16     func_idx;
+
+	u16     rsvd1;
+
+	u8      set_cmdq_depth;
+	u8      cmdq_depth;
+
+	u8      rsvd2;
+	u8      rsvd3;
+	u8      rsvd4;
+	u8      rsvd5;
+
+	u16     rq_depth;
+	u16     rx_buf_sz_idx;
+	u16     sq_depth;
+};
+
+struct hinic_cmd_io_status {
+	u8      status;
+	u8      version;
+	u8      rsvd0[6];
+
+	u16     func_idx;
+	u8      rsvd1;
+	u8      rsvd2;
+	u32     io_status;
+};
+
+struct hinic_cmd_clear_io_res {
+	u8      status;
+	u8      version;
+	u8      rsvd0[6];
+
+	u16     func_idx;
+	u8      rsvd1;
+	u8      rsvd2;
+};
+
+struct hinic_cmd_set_res_state {
+	u8      status;
+	u8      version;
+	u8      rsvd0[6];
+
+	u16     func_idx;
+	u8      state;
+	u8      rsvd1;
+	u32     rsvd2;
+};
+
 struct hinic_cmd_base_qpn {
 	u8      status;
 	u8      version;
@@ -137,4 +207,11 @@ int hinic_port_msg_cmd(struct hinic_hwdev *hwdev, enum hinic_port_cmd cmd,
 
 struct hinic_rq *hinic_hwdev_get_rq(struct hinic_hwdev *hwdev, int i);
 
+int hinic_hwdev_msix_cnt_set(struct hinic_hwdev *hwdev, u16 msix_index);
+
+int hinic_hwdev_msix_set(struct hinic_hwdev *hwdev, u16 msix_index,
+			 u8 pending_limit, u8 coalesc_timer,
+			 u8 lli_timer_cfg, u8 lli_credit_limit,
+			 u8 resend_timer);
+
 #endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_if.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_if.c
index d4e6ec4..effca4e 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_if.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_if.c
@@ -133,6 +133,42 @@ void hinic_set_pf_action(struct hinic_hwif *hwif, enum hinic_pf_action action)
 	hinic_hwif_write_reg(hwif, HINIC_CSR_FUNC_ATTR5_ADDR, attr5);
 }
 
+enum hinic_outbound_state hinic_outbound_state_get(struct hinic_hwif *hwif)
+{
+	u32 attr4 = hinic_hwif_read_reg(hwif, HINIC_CSR_FUNC_ATTR4_ADDR);
+
+	return HINIC_FA4_GET(attr4, OUTBOUND_STATE);
+}
+
+void hinic_outbound_state_set(struct hinic_hwif *hwif,
+			      enum hinic_outbound_state outbound_state)
+{
+	u32 attr4 = hinic_hwif_read_reg(hwif, HINIC_CSR_FUNC_ATTR4_ADDR);
+
+	attr4 = HINIC_FA4_CLEAR(attr4, OUTBOUND_STATE);
+	attr4 |= HINIC_FA4_SET(outbound_state, OUTBOUND_STATE);
+
+	hinic_hwif_write_reg(hwif, HINIC_CSR_FUNC_ATTR4_ADDR, attr4);
+}
+
+enum hinic_db_state hinic_db_state_get(struct hinic_hwif *hwif)
+{
+	u32 attr4 = hinic_hwif_read_reg(hwif, HINIC_CSR_FUNC_ATTR4_ADDR);
+
+	return HINIC_FA4_GET(attr4, DB_STATE);
+}
+
+void hinic_db_state_set(struct hinic_hwif *hwif,
+			enum hinic_db_state db_state)
+{
+	u32 attr4 = hinic_hwif_read_reg(hwif, HINIC_CSR_FUNC_ATTR4_ADDR);
+
+	attr4 = HINIC_FA4_CLEAR(attr4, DB_STATE);
+	attr4 |= HINIC_FA4_SET(db_state, DB_STATE);
+
+	hinic_hwif_write_reg(hwif, HINIC_CSR_FUNC_ATTR4_ADDR, attr4);
+}
+
 /**
  * hwif_ready - test if the HW is ready for use
  * @hwif: the HW interface of a pci function device
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_if.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_if.h
index 8f59195..5b4760c 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_if.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_if.h
@@ -73,6 +73,21 @@
 #define HINIC_FA1_GET(val, member)                              \
 	(((val) >> HINIC_FA1_##member##_SHIFT) & HINIC_FA1_##member##_MASK)
 
+#define HINIC_FA4_OUTBOUND_STATE_SHIFT                          0
+#define HINIC_FA4_DB_STATE_SHIFT                                1
+
+#define HINIC_FA4_OUTBOUND_STATE_MASK                           0x1
+#define HINIC_FA4_DB_STATE_MASK                                 0x1
+
+#define HINIC_FA4_GET(val, member)                              \
+	(((val) >> HINIC_FA4_##member##_SHIFT) & HINIC_FA4_##member##_MASK)
+
+#define HINIC_FA4_SET(val, member)                              \
+	((((u32)val) & HINIC_FA4_##member##_MASK) << HINIC_FA4_##member##_SHIFT)
+
+#define HINIC_FA4_CLEAR(val, member)                            \
+	((val) & (~(HINIC_FA4_##member##_MASK << HINIC_FA4_##member##_SHIFT)))
+
 #define HINIC_FA5_PF_ACTION_SHIFT                               0
 #define HINIC_FA5_PF_ACTION_MASK                                0xFFFF
 
@@ -182,6 +197,16 @@ enum hinic_pf_action {
 	HINIC_PF_MGMT_ACTIVE = 0x11,
 };
 
+enum hinic_outbound_state {
+	HINIC_OUTBOUND_ENABLE  = 0,
+	HINIC_OUTBOUND_DISABLE = 1,
+};
+
+enum hinic_db_state {
+	HINIC_DB_ENABLE  = 0,
+	HINIC_DB_DISABLE = 1,
+};
+
 struct hinic_func_attr {
 	u16                     func_idx;
 	u8                      pf_idx;
@@ -230,6 +255,16 @@ int hinic_msix_attr_get(struct hinic_hwif *hwif, u16 msix_index,
 
 void hinic_set_pf_action(struct hinic_hwif *hwif, enum hinic_pf_action action);
 
+enum hinic_outbound_state hinic_outbound_state_get(struct hinic_hwif *hwif);
+
+void hinic_outbound_state_set(struct hinic_hwif *hwif,
+			      enum hinic_outbound_state outbound_state);
+
+enum hinic_db_state hinic_db_state_get(struct hinic_hwif *hwif);
+
+void hinic_db_state_set(struct hinic_hwif *hwif,
+			enum hinic_db_state db_state);
+
 int hinic_init_hwif(struct hinic_hwif *hwif, struct pci_dev *pdev);
 
 void hinic_free_hwif(struct hinic_hwif *hwif);
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.h
index 90116c2..320711e 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.h
@@ -69,8 +69,21 @@ enum hinic_cfg_cmd {
 };
 
 enum hinic_comm_cmd {
+	HINIC_COMM_CMD_IO_STATUS_GET    = 0x3,
+
 	HINIC_COMM_CMD_CMDQ_CTXT_SET    = 0x10,
 	HINIC_COMM_CMD_CMDQ_CTXT_GET    = 0x11,
+
+	HINIC_COMM_CMD_HWCTXT_SET       = 0x12,
+	HINIC_COMM_CMD_HWCTXT_GET       = 0x13,
+
+	HINIC_COMM_CMD_SQ_HI_CI_SET     = 0x14,
+
+	HINIC_COMM_CMD_RES_STATE_SET    = 0x24,
+
+	HINIC_COMM_CMD_IO_RES_CLEAR     = 0x29,
+
+	HINIC_COMM_CMD_MAX              = 0x32,
 };
 
 enum hinic_mgmt_cb_state {
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c
index cfed040..6e540d2 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c
@@ -22,10 +22,13 @@
 #include <linux/errno.h>
 #include <linux/sizes.h>
 #include <linux/atomic.h>
+#include <linux/skbuff.h>
+#include <asm/barrier.h>
 #include <asm/byteorder.h>
 
 #include "hinic_common.h"
 #include "hinic_hw_if.h"
+#include "hinic_hw_wqe.h"
 #include "hinic_hw_wq.h"
 #include "hinic_hw_qp_ctxt.h"
 #include "hinic_hw_qp.h"
@@ -51,6 +54,13 @@
 		 + (q_id) * Q_CTXT_SIZE)
 
 #define SIZE_16BYTES(size)      (ALIGN(size, 16) >> 4)
+#define SIZE_8BYTES(size)       (ALIGN(size, 8) >> 3)
+
+#define RQ_MASKED_IDX(rq, idx)  ((idx) & (rq)->wq->mask)
+
+enum rq_completion_fmt {
+	RQ_COMPLETE_SGE = 1
+};
 
 void hinic_qp_prepare_header(struct hinic_qp_ctxt_header *qp_ctxt_hdr,
 			     enum hinic_qp_ctxt_type ctxt_type,
@@ -423,3 +433,203 @@ void hinic_clean_rq(struct hinic_rq *rq)
 	free_rq_cqe(rq);
 	free_rq_skb_arr(rq);
 }
+
+/**
+ * hinic_get_rq_free_wqebbs - return number of free wqebbs for use
+ * @rq: recv queue
+ *
+ * Return number of free wqebbs
+ **/
+int hinic_get_rq_free_wqebbs(struct hinic_rq *rq)
+{
+	struct hinic_wq *wq = rq->wq;
+
+	return atomic_read(&wq->delta) - 1;
+}
+
+/**
+ * hinic_rq_get_wqe - get wqe ptr in the current pi and update the pi
+ * @rq: rq to get wqe from
+ * @wqe_size: wqe size
+ * @prod_idx: returned pi
+ *
+ * Return wqe pointer
+ **/
+struct hinic_rq_wqe *hinic_rq_get_wqe(struct hinic_rq *rq,
+				      unsigned int wqe_size, u16 *prod_idx)
+{
+	struct hinic_hw_wqe *hw_wqe = hinic_get_wqe(rq->wq, wqe_size,
+						    prod_idx);
+
+	if (IS_ERR(hw_wqe))
+		return NULL;
+
+	return &hw_wqe->rq_wqe;
+}
+
+/**
+ * hinic_rq_write_wqe - write the wqe to the rq
+ * @rq: recv queue
+ * @prod_idx: pi of the wqe
+ * @rq_wqe: the wqe to write
+ * @skb: skb to save
+ **/
+void hinic_rq_write_wqe(struct hinic_rq *rq, u16 prod_idx,
+			struct hinic_rq_wqe *rq_wqe, struct sk_buff *skb)
+{
+	struct hinic_hw_wqe *hw_wqe = (struct hinic_hw_wqe *)rq_wqe;
+
+	rq->saved_skb[prod_idx] = skb;
+
+	/* The data in the HW should be in Big Endian Format */
+	hinic_cpu_to_be32(rq_wqe, sizeof(*rq_wqe));
+
+	hinic_write_wqe(rq->wq, hw_wqe, sizeof(*rq_wqe));
+}
+
+/**
+ * hinic_rq_read_wqe - read wqe ptr in the current ci and update the ci
+ * @rq: recv queue
+ * @wqe_size: the size of the wqe
+ * @skb: return saved skb
+ * @cons_idx: consumer index of the wqe
+ *
+ * Return wqe in ci position
+ **/
+struct hinic_rq_wqe *hinic_rq_read_wqe(struct hinic_rq *rq,
+				       unsigned int wqe_size,
+				       struct sk_buff **skb, u16 *cons_idx)
+{
+	struct hinic_hw_wqe *hw_wqe;
+	struct hinic_rq_cqe *cqe;
+	u32 status;
+	int rx_done;
+
+	hw_wqe = hinic_read_wqe(rq->wq, wqe_size, cons_idx);
+
+	if (IS_ERR(hw_wqe))
+		return NULL;
+
+	cqe = rq->cqe[*cons_idx];
+
+	status = be32_to_cpu(cqe->status);
+
+	rx_done = HINIC_RQ_CQE_STATUS_GET(status, RXDONE);
+	if (!rx_done)
+		return NULL;
+
+	*skb = rq->saved_skb[*cons_idx];
+
+	return &hw_wqe->rq_wqe;
+}
+
+/**
+ * hinic_rq_read_next_wqe - increment ci and read the wqe in ci position
+ * @rq: recv queue
+ * @wqe_size: the size of the wqe
+ * @skb: return saved skb
+ * @cons_idx: consumer index in the wq
+ *
+ * Return wqe in incremented ci position
+ **/
+struct hinic_rq_wqe *hinic_rq_read_next_wqe(struct hinic_rq *rq,
+					    unsigned int wqe_size,
+					    struct sk_buff **skb,
+					    u16 *cons_idx)
+{
+	struct hinic_wq *wq = rq->wq;
+	struct hinic_hw_wqe *hw_wqe;
+	unsigned int num_wqebbs;
+
+	wqe_size = ALIGN(wqe_size, wq->wqebb_size);
+	num_wqebbs = wqe_size / wq->wqebb_size;
+
+	*cons_idx = RQ_MASKED_IDX(rq, *cons_idx + num_wqebbs);
+
+	*skb = rq->saved_skb[*cons_idx];
+
+	hw_wqe = hinic_read_wqe_direct(wq, *cons_idx);
+
+	return &hw_wqe->rq_wqe;
+}
+
+/**
+ * hinic_put_wqe - release the ci for new wqes
+ * @rq: recv queue
+ * @cons_idx: consumer index of the wqe
+ * @wqe_size: the size of the wqe
+ **/
+void hinic_rq_put_wqe(struct hinic_rq *rq, u16 cons_idx,
+		      unsigned int wqe_size)
+{
+	struct hinic_rq_cqe *cqe = rq->cqe[cons_idx];
+	u32 status = be32_to_cpu(cqe->status);
+
+	status = HINIC_RQ_CQE_STATUS_CLEAR(status, RXDONE);
+
+	/* Rx WQE size is 1 WQEBB, no wq shadow*/
+	cqe->status = cpu_to_be32(status);
+
+	wmb();          /* clear done flag */
+
+	hinic_put_wqe(rq->wq, wqe_size);
+}
+
+/**
+ * hinic_rq_get_sge - get sge from the wqe
+ * @rq: recv queue
+ * @rq_wqe: wqe to get the sge from its buf address
+ * @cons_idx: consumer index
+ * @sge: returned sge
+ **/
+void hinic_rq_get_sge(struct hinic_rq *rq, struct hinic_rq_wqe *rq_wqe,
+		      u16 cons_idx, struct hinic_sge *sge)
+{
+	struct hinic_rq_cqe *cqe = rq->cqe[cons_idx];
+	u32 len = be32_to_cpu(cqe->len);
+
+	sge->hi_addr = be32_to_cpu(rq_wqe->buf_desc.hi_addr);
+	sge->lo_addr = be32_to_cpu(rq_wqe->buf_desc.lo_addr);
+	sge->len = HINIC_RQ_CQE_SGE_GET(len, LEN);
+}
+
+/**
+ * hinic_rq_prepare_wqe - prepare wqe before insert to the queue
+ * @rq: recv queue
+ * @prod_idx: pi value
+ * @rq_wqe: the wqe
+ * @sge: sge for use by the wqe for recv buf address
+ **/
+void hinic_rq_prepare_wqe(struct hinic_rq *rq, u16 prod_idx,
+			  struct hinic_rq_wqe *rq_wqe, struct hinic_sge *sge)
+{
+	struct hinic_rq_cqe_sect *cqe_sect = &rq_wqe->cqe_sect;
+	struct hinic_rq_bufdesc *buf_desc = &rq_wqe->buf_desc;
+	struct hinic_rq_ctrl *ctrl = &rq_wqe->ctrl;
+	dma_addr_t cqe_dma = rq->cqe_dma[prod_idx];
+
+	ctrl->ctrl_info =
+		HINIC_RQ_CTRL_SET(SIZE_8BYTES(sizeof(*ctrl)), LEN) |
+		HINIC_RQ_CTRL_SET(SIZE_8BYTES(sizeof(*cqe_sect)),
+				  COMPLETE_LEN)                    |
+		HINIC_RQ_CTRL_SET(SIZE_8BYTES(sizeof(*buf_desc)),
+				  BUFDESC_SECT_LEN)                |
+		HINIC_RQ_CTRL_SET(RQ_COMPLETE_SGE, COMPLETE_FORMAT);
+
+	hinic_set_sge(&cqe_sect->sge, cqe_dma, sizeof(struct hinic_rq_cqe));
+
+	buf_desc->hi_addr = sge->hi_addr;
+	buf_desc->lo_addr = sge->lo_addr;
+}
+
+/**
+ * hinic_rq_update - update pi of the rq
+ * @rq: recv queue
+ * @prod_idx: pi value
+ **/
+void hinic_rq_update(struct hinic_rq *rq, u16 prod_idx)
+{
+	struct hinic_wq *wq = rq->wq;
+
+	*rq->pi_virt_addr = cpu_to_be16((prod_idx + 1) & wq->mask);
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
index eb10bd9..1331bd4 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
@@ -21,6 +21,7 @@
 #include <linux/pci.h>
 #include <linux/skbuff.h>
 
+#include "hinic_common.h"
 #include "hinic_hw_if.h"
 #include "hinic_hw_wqe.h"
 #include "hinic_hw_wq.h"
@@ -100,4 +101,32 @@ int hinic_init_rq(struct hinic_rq *rq, struct hinic_hwif *hwif,
 
 void hinic_clean_rq(struct hinic_rq *rq);
 
+int hinic_get_rq_free_wqebbs(struct hinic_rq *rq);
+
+struct hinic_rq_wqe *hinic_rq_get_wqe(struct hinic_rq *rq,
+				      unsigned int wqe_size, u16 *prod_idx);
+
+void hinic_rq_write_wqe(struct hinic_rq *rq, u16 prod_idx,
+			struct hinic_rq_wqe *wqe, struct sk_buff *skb);
+
+struct hinic_rq_wqe *hinic_rq_read_wqe(struct hinic_rq *rq,
+				       unsigned int wqe_size,
+				       struct sk_buff **skb, u16 *cons_idx);
+
+struct hinic_rq_wqe *hinic_rq_read_next_wqe(struct hinic_rq *rq,
+					    unsigned int wqe_size,
+					    struct sk_buff **skb,
+					    u16 *cons_idx);
+
+void hinic_rq_put_wqe(struct hinic_rq *rq, u16 cons_idx,
+		      unsigned int wqe_size);
+
+void hinic_rq_get_sge(struct hinic_rq *rq, struct hinic_rq_wqe *wqe,
+		      u16 cons_idx, struct hinic_sge *sge);
+
+void hinic_rq_prepare_wqe(struct hinic_rq *rq, u16 prod_idx,
+			  struct hinic_rq_wqe *wqe, struct hinic_sge *sge);
+
+void hinic_rq_update(struct hinic_rq *rq, u16 prod_idx);
+
 #endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c
index a22e201..1cb3745 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c
@@ -828,6 +828,18 @@ struct hinic_hw_wqe *hinic_read_wqe(struct hinic_wq *wq, unsigned int wqe_size,
 }
 
 /**
+ * hinic_read_wqe_direct - read wqe directly from ci position
+ * @wq: wq
+ * @cons_idx: ci position
+ *
+ * Return wqe
+ **/
+struct hinic_hw_wqe *hinic_read_wqe_direct(struct hinic_wq *wq, u16 cons_idx)
+{
+	return WQ_PAGE_ADDR(wq, cons_idx) + WQE_PAGE_OFF(wq, cons_idx);
+}
+
+/**
  * wqe_shadow - check if a wqe is shadow
  * @wq: wq of the wqe
  * @wqe: the wqe for shadow checking
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h
index f01477a..9c030a0 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h
@@ -109,6 +109,8 @@ struct hinic_hw_wqe *hinic_get_wqe(struct hinic_wq *wq, unsigned int wqe_size,
 struct hinic_hw_wqe *hinic_read_wqe(struct hinic_wq *wq, unsigned int wqe_size,
 				    u16 *cons_idx);
 
+struct hinic_hw_wqe *hinic_read_wqe_direct(struct hinic_wq *wq, u16 cons_idx);
+
 void hinic_write_wqe(struct hinic_wq *wq, struct hinic_hw_wqe *wqe,
 		     unsigned int wqe_size);
 
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_main.c b/drivers/net/ethernet/huawei/hinic/hinic_main.c
index 914ce52..053e5f3 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_main.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_main.c
@@ -15,6 +15,7 @@
 
 #include <linux/kernel.h>
 #include <linux/module.h>
+#include <linux/moduleparam.h>
 #include <linux/pci.h>
 #include <linux/device.h>
 #include <linux/errno.h>
@@ -42,6 +43,10 @@
 MODULE_DESCRIPTION("Huawei Intelligent NIC driver");
 MODULE_LICENSE("GPL");
 
+static unsigned int rx_weight = 64;
+module_param(rx_weight, uint, 0644);
+MODULE_PARM_DESC(rx_weight, "Number Rx packets for NAPI budget (default=64)");
+
 #define PCI_DEVICE_ID_HI1822_PF         0x1822
 
 #define HINIC_WQ_NAME                   "hinic_dev"
@@ -220,6 +225,13 @@ static int hinic_open(struct net_device *netdev)
 		goto err_port_state;
 	}
 
+	err = hinic_port_set_func_state(nic_dev, HINIC_FUNC_PORT_ENABLE);
+	if (err) {
+		netif_err(nic_dev, drv, netdev,
+			  "Failed to set func port state\n");
+		goto err_func_port_state;
+	}
+
 	/* Wait up to 3 sec between port enable to link state */
 	msleep(3000);
 
@@ -250,6 +262,12 @@ static int hinic_open(struct net_device *netdev)
 
 err_port_link:
 	up(&nic_dev->mgmt_lock);
+	ret = hinic_port_set_func_state(nic_dev, HINIC_FUNC_PORT_DISABLE);
+	if (ret)
+		netif_warn(nic_dev, drv, netdev,
+			   "Failed to revert func port state\n");
+
+err_func_port_state:
 	ret = hinic_port_set_state(nic_dev, HINIC_PORT_DISABLE);
 	if (ret)
 		netif_warn(nic_dev, drv, netdev,
@@ -283,6 +301,14 @@ static int hinic_close(struct net_device *netdev)
 
 	up(&nic_dev->mgmt_lock);
 
+	err = hinic_port_set_func_state(nic_dev, HINIC_FUNC_PORT_DISABLE);
+	if (err) {
+		netif_err(nic_dev, drv, netdev,
+			  "Failed to set func port state\n");
+		nic_dev->flags |= (flags & HINIC_INTF_UP);
+		return err;
+	}
+
 	err = hinic_port_set_state(nic_dev, HINIC_PORT_DISABLE);
 	if (err) {
 		netif_err(nic_dev, drv, netdev, "Failed to set port state\n");
@@ -664,6 +690,7 @@ static int nic_dev_init(struct pci_dev *pdev)
 	nic_dev->flags = 0;
 	nic_dev->txqs = NULL;
 	nic_dev->rxqs = NULL;
+	nic_dev->rx_weight = rx_weight;
 
 	sema_init(&nic_dev->mgmt_lock, 1);
 
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_port.c b/drivers/net/ethernet/huawei/hinic/hinic_port.c
index 0dafede..528ec6f 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_port.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_port.c
@@ -314,3 +314,35 @@ int hinic_port_set_state(struct hinic_dev *nic_dev, enum hinic_port_state state)
 
 	return 0;
 }
+
+/**
+ * hinic_port_set_func_state- set func device state
+ * @nic_dev: nic device
+ * @state: the state to set
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_port_set_func_state(struct hinic_dev *nic_dev,
+			      enum hinic_func_port_state state)
+{
+	struct hinic_port_func_state_cmd func_state;
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	u16 out_size;
+	int err;
+
+	func_state.func_idx = HINIC_HWIF_FUNC_IDX(hwif);
+	func_state.state = state;
+
+	err = hinic_port_msg_cmd(hwdev, HINIC_PORT_CMD_SET_FUNC_STATE,
+				 &func_state, sizeof(func_state),
+				 &func_state, &out_size);
+	if (err || (out_size != sizeof(func_state)) || func_state.status) {
+		dev_err(&pdev->dev, "Failed to set port func state, ret = %d\n",
+			func_state.status);
+		return -EFAULT;
+	}
+
+	return 0;
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_port.h b/drivers/net/ethernet/huawei/hinic/hinic_port.h
index 3a8da8e..17f9d7fc 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_port.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_port.h
@@ -40,6 +40,11 @@ enum hinic_port_state {
 	HINIC_PORT_ENABLE       = 3,
 };
 
+enum hinic_func_port_state {
+	HINIC_FUNC_PORT_DISABLE = 0,
+	HINIC_FUNC_PORT_ENABLE  = 2,
+};
+
 struct hinic_port_mac_cmd {
 	u8              status;
 	u8              version;
@@ -109,6 +114,17 @@ struct hinic_port_link_status {
 	u8      rsvd2;
 };
 
+struct hinic_port_func_state_cmd {
+	u8      status;
+	u8      version;
+	u8      rsvd0[6];
+
+	u16     func_idx;
+	u16     rsvd1;
+	u8      state;
+	u8      rsvd2[3];
+};
+
 int hinic_port_add_mac(struct hinic_dev *nic_dev, const u8 *addr,
 		       u16 vlan_id);
 
@@ -131,4 +147,7 @@ int hinic_port_link_state(struct hinic_dev *nic_dev,
 int hinic_port_set_state(struct hinic_dev *nic_dev,
 			 enum hinic_port_state state);
 
+int hinic_port_set_func_state(struct hinic_dev *nic_dev,
+			      enum hinic_func_port_state state);
+
 #endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_rx.c b/drivers/net/ethernet/huawei/hinic/hinic_rx.c
index 173fe8b..6367fa1 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_rx.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_rx.c
@@ -13,11 +13,35 @@
  *
  */
 
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/errno.h>
+#include <linux/pci.h>
+#include <linux/device.h>
 #include <linux/netdevice.h>
+#include <linux/etherdevice.h>
 #include <linux/u64_stats_sync.h>
+#include <linux/slab.h>
+#include <linux/interrupt.h>
+#include <linux/skbuff.h>
+#include <linux/dma-mapping.h>
+#include <linux/prefetch.h>
+#include <asm/barrier.h>
 
+#include "hinic_common.h"
+#include "hinic_hw_if.h"
+#include "hinic_hw_wqe.h"
+#include "hinic_hw_wq.h"
 #include "hinic_hw_qp.h"
+#include "hinic_hw_dev.h"
 #include "hinic_rx.h"
+#include "hinic_dev.h"
+
+#define RX_IRQ_NO_PENDING               0
+#define RX_IRQ_NO_COALESC               0
+#define RX_IRQ_NO_LLI_TIMER             0
+#define RX_IRQ_NO_CREDIT                0
+#define RX_IRQ_NO_RESEND_TIMER          0
 
 /**
  * hinic_rxq_clean_stats - Clean the statistics of specific queue
@@ -46,6 +70,362 @@ static void rxq_stats_init(struct hinic_rxq *rxq)
 }
 
 /**
+ * rx_alloc_skb - allocate skb and map it to dma address
+ * @rxq: rx queue
+ * @dma_addr: returned dma address for the skb
+ *
+ * Return skb
+ **/
+static struct sk_buff *rx_alloc_skb(struct hinic_rxq *rxq,
+				    dma_addr_t *dma_addr)
+{
+	struct hinic_dev *nic_dev = netdev_priv(rxq->netdev);
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_rq *rq = rxq->rq;
+	struct sk_buff *skb;
+	dma_addr_t addr;
+	int err;
+
+	skb = netdev_alloc_skb_ip_align(rxq->netdev, rq->buf_sz);
+	if (!skb) {
+		netdev_err(rxq->netdev, "Failed to allocate Rx SKB\n");
+		return NULL;
+	}
+
+	addr = dma_map_single(&pdev->dev, skb->data, rq->buf_sz,
+			      DMA_FROM_DEVICE);
+	err = dma_mapping_error(&pdev->dev, addr);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to map Rx DMA, err = %d\n", err);
+		goto err_rx_map;
+	}
+
+	*dma_addr = addr;
+	return skb;
+
+err_rx_map:
+	dev_kfree_skb_any(skb);
+	return NULL;
+}
+
+/**
+ * rx_unmap_skb - unmap the dma address of the skb
+ * @rxq: rx queue
+ * @dma_addr: dma address of the skb
+ **/
+static void rx_unmap_skb(struct hinic_rxq *rxq, dma_addr_t dma_addr)
+{
+	struct hinic_dev *nic_dev = netdev_priv(rxq->netdev);
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_rq *rq = rxq->rq;
+
+	dma_unmap_single(&pdev->dev, dma_addr, rq->buf_sz, DMA_FROM_DEVICE);
+}
+
+/**
+ * rx_free_skb - unmap and free skb
+ * @rxq: rx queue
+ * @skb: skb to free
+ * @dma_addr: dma address of the skb
+ **/
+static void rx_free_skb(struct hinic_rxq *rxq, struct sk_buff *skb,
+			dma_addr_t dma_addr)
+{
+	rx_unmap_skb(rxq, dma_addr);
+	dev_kfree_skb_any(skb);
+}
+
+/**
+ * rx_alloc_pkts - allocate pkts in rx queue
+ * @rxq: rx queue
+ *
+ * Return number of skbs allocated
+ **/
+static int rx_alloc_pkts(struct hinic_rxq *rxq)
+{
+	struct hinic_dev *nic_dev = netdev_priv(rxq->netdev);
+	struct hinic_rq_wqe *rq_wqe;
+	unsigned int free_wqebbs;
+	struct hinic_sge sge;
+	dma_addr_t dma_addr;
+	struct sk_buff *skb;
+	int i, alloc_more;
+	u16 prod_idx;
+
+	free_wqebbs = hinic_get_rq_free_wqebbs(rxq->rq);
+	alloc_more = 0;
+
+	/* Limit the allocation chunks */
+	if (free_wqebbs > nic_dev->rx_weight)
+		free_wqebbs = nic_dev->rx_weight;
+
+	for (i = 0; i < free_wqebbs; i++) {
+		skb = rx_alloc_skb(rxq, &dma_addr);
+		if (!skb) {
+			netdev_err(rxq->netdev, "Failed to alloc Rx skb\n");
+			alloc_more = 1;
+			goto skb_out;
+		}
+
+		hinic_set_sge(&sge, dma_addr, skb->len);
+
+		rq_wqe = hinic_rq_get_wqe(rxq->rq, HINIC_RQ_WQE_SIZE,
+					  &prod_idx);
+		if (!rq_wqe) {
+			rx_free_skb(rxq, skb, dma_addr);
+			alloc_more = 1;
+			goto skb_out;
+		}
+
+		hinic_rq_prepare_wqe(rxq->rq, prod_idx, rq_wqe, &sge);
+
+		hinic_rq_write_wqe(rxq->rq, prod_idx, rq_wqe, skb);
+	}
+
+skb_out:
+	if (i) {
+		wmb();  /* write all the wqes before update PI */
+
+		hinic_rq_update(rxq->rq, prod_idx);
+	}
+
+	if (alloc_more)
+		tasklet_schedule(&rxq->rx_task);
+
+	return i;
+}
+
+/**
+ * free_all_rx_skbs - free all skbs in rx queue
+ * @rxq: rx queue
+ **/
+static void free_all_rx_skbs(struct hinic_rxq *rxq)
+{
+	struct hinic_rq *rq = rxq->rq;
+	struct hinic_hw_wqe *hw_wqe;
+	struct hinic_sge sge;
+	u16 ci;
+
+	while ((hw_wqe = hinic_read_wqe(rq->wq, HINIC_RQ_WQE_SIZE, &ci))) {
+		if (IS_ERR(hw_wqe))
+			break;
+
+		hinic_rq_get_sge(rq, &hw_wqe->rq_wqe, ci, &sge);
+
+		hinic_put_wqe(rq->wq, HINIC_RQ_WQE_SIZE);
+
+		rx_free_skb(rxq, rq->saved_skb[ci], hinic_sge_to_dma(&sge));
+	}
+}
+
+/**
+ * rx_alloc_task - tasklet for queue allocation
+ * @data: rx queue
+ **/
+static void rx_alloc_task(unsigned long data)
+{
+	struct hinic_rxq *rxq = (struct hinic_rxq *)data;
+
+	(void)rx_alloc_pkts(rxq);
+}
+
+/**
+ * rx_recv_jumbo_pkt - Rx handler for jumbo pkt
+ * @rxq: rx queue
+ * @head_skb: the first skb in the list
+ * @left_pkt_len: left size of the pkt exclude head skb
+ * @ci: consumer index
+ *
+ * Return number of wqes that used for the left of the pkt
+ **/
+static int rx_recv_jumbo_pkt(struct hinic_rxq *rxq, struct sk_buff *head_skb,
+			     unsigned int left_pkt_len, u16 ci)
+{
+	struct sk_buff *skb, *curr_skb = head_skb;
+	struct hinic_rq_wqe *rq_wqe;
+	unsigned int curr_len;
+	struct hinic_sge sge;
+	int num_wqes = 0;
+
+	while (left_pkt_len > 0) {
+		rq_wqe = hinic_rq_read_next_wqe(rxq->rq, HINIC_RQ_WQE_SIZE,
+						&skb, &ci);
+
+		num_wqes++;
+
+		hinic_rq_get_sge(rxq->rq, rq_wqe, ci, &sge);
+
+		rx_unmap_skb(rxq, hinic_sge_to_dma(&sge));
+
+		prefetch(skb->data);
+
+		curr_len = (left_pkt_len > HINIC_RX_BUF_SZ) ? HINIC_RX_BUF_SZ :
+			    left_pkt_len;
+
+		left_pkt_len -= curr_len;
+
+		__skb_put(skb, curr_len);
+
+		if (curr_skb == head_skb)
+			skb_shinfo(head_skb)->frag_list = skb;
+		else
+			curr_skb->next = skb;
+
+		head_skb->len += skb->len;
+		head_skb->data_len += skb->len;
+		head_skb->truesize += skb->truesize;
+
+		curr_skb = skb;
+	}
+
+	return num_wqes;
+}
+
+/**
+ * rxq_recv - Rx handler
+ * @rxq: rx queue
+ * @budget: maximum pkts to process
+ *
+ * Return number of pkts received
+ **/
+static int rxq_recv(struct hinic_rxq *rxq, int budget)
+{
+	struct hinic_qp *qp = container_of(rxq->rq, struct hinic_qp, rq);
+	u64 pkt_len = 0, rx_bytes = 0;
+	struct hinic_rq_wqe *rq_wqe;
+	int num_wqes, pkts = 0;
+	struct hinic_sge sge;
+	struct sk_buff *skb;
+	u16 ci;
+
+	while (pkts < budget) {
+		num_wqes = 0;
+
+		rq_wqe = hinic_rq_read_wqe(rxq->rq, HINIC_RQ_WQE_SIZE, &skb,
+					   &ci);
+		if (!rq_wqe)
+			break;
+
+		hinic_rq_get_sge(rxq->rq, rq_wqe, ci, &sge);
+
+		rx_unmap_skb(rxq, hinic_sge_to_dma(&sge));
+
+		prefetch(skb->data);
+
+		pkt_len = sge.len;
+
+		if (pkt_len <= HINIC_RX_BUF_SZ) {
+			__skb_put(skb, pkt_len);
+		} else {
+			__skb_put(skb, HINIC_RX_BUF_SZ);
+			num_wqes = rx_recv_jumbo_pkt(rxq, skb, pkt_len -
+						     HINIC_RX_BUF_SZ, ci);
+		}
+
+		hinic_rq_put_wqe(rxq->rq, ci,
+				 (num_wqes + 1) * HINIC_RQ_WQE_SIZE);
+
+		skb_record_rx_queue(skb, qp->q_id);
+		skb->protocol = eth_type_trans(skb, rxq->netdev);
+
+		napi_gro_receive(&rxq->napi, skb);
+
+		pkts++;
+		rx_bytes += pkt_len;
+	}
+
+	if (pkts)
+		tasklet_schedule(&rxq->rx_task); /* hinic_rx_alloc_pkts */
+
+	u64_stats_update_begin(&rxq->rxq_stats.syncp);
+	rxq->rxq_stats.pkts += pkts;
+	rxq->rxq_stats.bytes += rx_bytes;
+	u64_stats_update_end(&rxq->rxq_stats.syncp);
+
+	return pkts;
+}
+
+static int rx_poll(struct napi_struct *napi, int budget)
+{
+	struct hinic_rxq *rxq = container_of(napi, struct hinic_rxq, napi);
+	struct hinic_rq *rq = rxq->rq;
+	int pkts;
+
+	pkts = rxq_recv(rxq, budget);
+	if (pkts >= budget)
+		return budget;
+
+	napi_complete(napi);
+	enable_irq(rq->irq);
+	return pkts;
+}
+
+static void rx_add_napi(struct hinic_rxq *rxq)
+{
+	struct hinic_dev *nic_dev = netdev_priv(rxq->netdev);
+
+	netif_napi_add(rxq->netdev, &rxq->napi, rx_poll, nic_dev->rx_weight);
+	napi_enable(&rxq->napi);
+}
+
+static void rx_del_napi(struct hinic_rxq *rxq)
+{
+	napi_disable(&rxq->napi);
+	netif_napi_del(&rxq->napi);
+}
+
+static irqreturn_t rx_irq(int irq, void *data)
+{
+	struct hinic_rxq *rxq = (struct hinic_rxq *)data;
+	struct hinic_rq *rq = rxq->rq;
+	struct hinic_dev *nic_dev;
+
+	/* Disable the interrupt until napi will be completed */
+	disable_irq_nosync(rq->irq);
+
+	nic_dev = netdev_priv(rxq->netdev);
+	hinic_hwdev_msix_cnt_set(nic_dev->hwdev, rq->msix_entry);
+
+	napi_schedule(&rxq->napi);
+	return IRQ_HANDLED;
+}
+
+static int rx_request_irq(struct hinic_rxq *rxq)
+{
+	struct hinic_dev *nic_dev = netdev_priv(rxq->netdev);
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	struct hinic_rq *rq = rxq->rq;
+	int err;
+
+	rx_add_napi(rxq);
+
+	hinic_hwdev_msix_set(hwdev, rq->msix_entry,
+			     RX_IRQ_NO_PENDING, RX_IRQ_NO_COALESC,
+			     RX_IRQ_NO_LLI_TIMER, RX_IRQ_NO_CREDIT,
+			     RX_IRQ_NO_RESEND_TIMER);
+
+	err = request_irq(rq->irq, rx_irq, 0, rxq->irq_name, rxq);
+	if (err) {
+		rx_del_napi(rxq);
+		return err;
+	}
+
+	return 0;
+}
+
+static void rx_free_irq(struct hinic_rxq *rxq)
+{
+	struct hinic_rq *rq = rxq->rq;
+
+	free_irq(rq->irq, rxq);
+	rx_del_napi(rxq);
+}
+
+/**
  * hinic_init_rxq - Initialize the Rx Queue
  * @rxq: Logical Rx Queue
  * @rq: Hardware Rx Queue to connect the Logical queue with
@@ -56,11 +436,43 @@ static void rxq_stats_init(struct hinic_rxq *rxq)
 int hinic_init_rxq(struct hinic_rxq *rxq, struct hinic_rq *rq,
 		   struct net_device *netdev)
 {
+	struct hinic_qp *qp = container_of(rq, struct hinic_qp, rq);
+	int err, pkts, irqname_len;
+
 	rxq->netdev = netdev;
 	rxq->rq = rq;
 
 	rxq_stats_init(rxq);
+
+	irqname_len = snprintf(NULL, 0, "hinic_rxq%d", qp->q_id) + 1;
+	rxq->irq_name = devm_kzalloc(&netdev->dev, irqname_len, GFP_KERNEL);
+	if (!rxq->irq_name)
+		return -ENOMEM;
+
+	sprintf(rxq->irq_name, "hinic_rxq%d", qp->q_id);
+
+	tasklet_init(&rxq->rx_task, rx_alloc_task, (unsigned long)rxq);
+
+	pkts = rx_alloc_pkts(rxq);
+	if (!pkts) {
+		err = -ENOMEM;
+		goto err_rx_pkts;
+	}
+
+	err = rx_request_irq(rxq);
+	if (err) {
+		netdev_err(netdev, "Failed to request Rx irq\n");
+		goto err_req_rx_irq;
+	}
+
 	return 0;
+
+err_req_rx_irq:
+err_rx_pkts:
+	tasklet_kill(&rxq->rx_task);
+	free_all_rx_skbs(rxq);
+	devm_kfree(&netdev->dev, rxq->irq_name);
+	return err;
 }
 
 /**
@@ -69,4 +481,11 @@ int hinic_init_rxq(struct hinic_rxq *rxq, struct hinic_rq *rq,
  **/
 void hinic_clean_rxq(struct hinic_rxq *rxq)
 {
+	struct net_device *netdev = rxq->netdev;
+
+	rx_free_irq(rxq);
+
+	tasklet_kill(&rxq->rx_task);
+	free_all_rx_skbs(rxq);
+	devm_kfree(&netdev->dev, rxq->irq_name);
 }
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_rx.h b/drivers/net/ethernet/huawei/hinic/hinic_rx.h
index fbd0246..538c886 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_rx.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_rx.h
@@ -19,6 +19,7 @@
 #include <linux/types.h>
 #include <linux/netdevice.h>
 #include <linux/u64_stats_sync.h>
+#include <linux/interrupt.h>
 
 #include "hinic_hw_qp.h"
 
@@ -34,6 +35,12 @@ struct hinic_rxq {
 	struct hinic_rq         *rq;
 
 	struct hinic_rxq_stats  rxq_stats;
+
+	char                    *irq_name;
+
+	struct tasklet_struct   rx_task;
+
+	struct napi_struct      napi;
 };
 
 void hinic_rxq_clean_stats(struct hinic_rxq *rxq);
-- 
1.9.1

^ permalink raw reply related

* [PATCH V7 net-next 17/22] net-next/hinic: Add cmdq completion handler
From: Aviad Krawczyk @ 2017-08-17 19:04 UTC (permalink / raw)
  To: davem
  Cc: linux-kernel, netdev, bc.y, victor.gissin, aviad.krawczyk,
	zhaochen6, tony.qu
In-Reply-To: <cover.1502996306.git.aviad.krawczyk@huawei.com>

Add cmdq completion handler for getting a notification about the
completion of cmdq commands.

Signed-off-by: Aviad Krawczyk <aviad.krawczyk@huawei.com>
Signed-off-by: Zhao Chen <zhaochen6@huawei.com>
---
 drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c | 297 +++++++++++++++++++++-
 drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h |  12 +
 2 files changed, 308 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c
index d8c0807..8d72762 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c
@@ -40,12 +40,31 @@
 #include "hinic_hw_io.h"
 #include "hinic_hw_dev.h"
 
+#define CMDQ_CEQE_TYPE_SHIFT                    0
+
+#define CMDQ_CEQE_TYPE_MASK                     0x7
+
+#define CMDQ_CEQE_GET(val, member)              \
+			(((val) >> CMDQ_CEQE_##member##_SHIFT) \
+			 & CMDQ_CEQE_##member##_MASK)
+
+#define CMDQ_WQE_ERRCODE_VAL_SHIFT              20
+
+#define CMDQ_WQE_ERRCODE_VAL_MASK               0xF
+
+#define CMDQ_WQE_ERRCODE_GET(val, member)       \
+			(((val) >> CMDQ_WQE_ERRCODE_##member##_SHIFT) \
+			 & CMDQ_WQE_ERRCODE_##member##_MASK)
+
 #define CMDQ_DB_PI_OFF(pi)              (((u16)LOWER_8_BITS(pi)) << 3)
 
 #define CMDQ_DB_ADDR(db_base, pi)       ((db_base) + CMDQ_DB_PI_OFF(pi))
 
 #define CMDQ_WQE_HEADER(wqe)            ((struct hinic_cmdq_header *)(wqe))
 
+#define CMDQ_WQE_COMPLETED(ctrl_info)   \
+			HINIC_CMDQ_CTRL_GET(ctrl_info, HW_BUSY_BIT)
+
 #define FIRST_DATA_TO_WRITE_LAST        sizeof(u64)
 
 #define CMDQ_DB_OFF                     SZ_2K
@@ -145,6 +164,22 @@ void hinic_free_cmdq_buf(struct hinic_cmdqs *cmdqs,
 	pci_pool_free(cmdqs->cmdq_buf_pool, cmdq_buf->buf, cmdq_buf->dma_addr);
 }
 
+static unsigned int cmdq_wqe_size_from_bdlen(enum bufdesc_len len)
+{
+	unsigned int wqe_size = 0;
+
+	switch (len) {
+	case BUFDESC_LCMD_LEN:
+		wqe_size = WQE_LCMD_SIZE;
+		break;
+	case BUFDESC_SCMD_LEN:
+		wqe_size = WQE_SCMD_SIZE;
+		break;
+	}
+
+	return wqe_size;
+}
+
 static void cmdq_set_sge_completion(struct hinic_cmdq_completion *completion,
 				    struct hinic_cmdq_buf *buf_out)
 {
@@ -211,6 +246,15 @@ static void cmdq_set_lcmd_bufdesc(struct hinic_cmdq_wqe_lcmd *wqe_lcmd,
 	hinic_set_sge(&wqe_lcmd->buf_desc.sge, buf_in->dma_addr, buf_in->size);
 }
 
+static void cmdq_set_direct_wqe_data(struct hinic_cmdq_direct_wqe *wqe,
+				     void *buf_in, u32 in_size)
+{
+	struct hinic_cmdq_wqe_scmd *wqe_scmd = &wqe->wqe_scmd;
+
+	wqe_scmd->buf_desc.buf_len = in_size;
+	memcpy(wqe_scmd->buf_desc.data, buf_in, in_size);
+}
+
 static void cmdq_set_lcmd_wqe(struct hinic_cmdq_wqe *wqe,
 			      enum cmdq_cmd_type cmd_type,
 			      struct hinic_cmdq_buf *buf_in,
@@ -239,6 +283,36 @@ static void cmdq_set_lcmd_wqe(struct hinic_cmdq_wqe *wqe,
 	cmdq_set_lcmd_bufdesc(wqe_lcmd, buf_in);
 }
 
+static void cmdq_set_direct_wqe(struct hinic_cmdq_wqe *wqe,
+				enum cmdq_cmd_type cmd_type,
+				void *buf_in, u16 in_size,
+				struct hinic_cmdq_buf *buf_out, int wrapped,
+				enum hinic_cmd_ack_type ack_type,
+				enum hinic_mod_type mod, u8 cmd, u16 prod_idx)
+{
+	struct hinic_cmdq_direct_wqe *direct_wqe = &wqe->direct_wqe;
+	enum completion_format complete_format;
+	struct hinic_cmdq_wqe_scmd *wqe_scmd;
+
+	wqe_scmd = &direct_wqe->wqe_scmd;
+
+	switch (cmd_type) {
+	case CMDQ_CMD_SYNC_SGE_RESP:
+		complete_format = COMPLETE_SGE;
+		cmdq_set_sge_completion(&wqe_scmd->completion, buf_out);
+		break;
+	case CMDQ_CMD_SYNC_DIRECT_RESP:
+		complete_format = COMPLETE_DIRECT;
+		wqe_scmd->completion.direct_resp = 0;
+		break;
+	}
+
+	cmdq_prepare_wqe_ctrl(wqe, wrapped, ack_type, mod, cmd, prod_idx,
+			      complete_format, DATA_DIRECT, BUFDESC_SCMD_LEN);
+
+	cmdq_set_direct_wqe_data(direct_wqe, buf_in, in_size);
+}
+
 static void cmdq_wqe_fill(void *dst, void *src)
 {
 	memcpy(dst + FIRST_DATA_TO_WRITE_LAST, src + FIRST_DATA_TO_WRITE_LAST,
@@ -352,6 +426,52 @@ static int cmdq_sync_cmd_direct_resp(struct hinic_cmdq *cmdq,
 	return 0;
 }
 
+static int cmdq_set_arm_bit(struct hinic_cmdq *cmdq, void *buf_in,
+			    u16 in_size)
+{
+	struct hinic_cmdq_wqe *curr_cmdq_wqe, cmdq_wqe;
+	u16 curr_prod_idx, next_prod_idx;
+	struct hinic_wq *wq = cmdq->wq;
+	struct hinic_hw_wqe *hw_wqe;
+	int wrapped, num_wqebbs;
+
+	/* Keep doorbell index correct */
+	spin_lock(&cmdq->cmdq_lock);
+
+	/* WQE_SIZE = WQEBB_SIZE, we will get the wq element and not shadow*/
+	hw_wqe = hinic_get_wqe(wq, WQE_SCMD_SIZE, &curr_prod_idx);
+	if (IS_ERR(hw_wqe)) {
+		spin_unlock(&cmdq->cmdq_lock);
+		return -EBUSY;
+	}
+
+	curr_cmdq_wqe = &hw_wqe->cmdq_wqe;
+
+	wrapped = cmdq->wrapped;
+
+	num_wqebbs = ALIGN(WQE_SCMD_SIZE, wq->wqebb_size) / wq->wqebb_size;
+	next_prod_idx = curr_prod_idx + num_wqebbs;
+	if (next_prod_idx >= wq->q_depth) {
+		cmdq->wrapped = !cmdq->wrapped;
+		next_prod_idx -= wq->q_depth;
+	}
+
+	cmdq_set_direct_wqe(&cmdq_wqe, CMDQ_CMD_SYNC_DIRECT_RESP, buf_in,
+			    in_size, NULL, wrapped, HINIC_CMD_ACK_TYPE_CMDQ,
+			    HINIC_MOD_COMM, CMDQ_SET_ARM_CMD, curr_prod_idx);
+
+	/* The data that is written to HW should be in Big Endian Format */
+	hinic_cpu_to_be32(&cmdq_wqe, WQE_SCMD_SIZE);
+
+	/* cmdq wqe is not shadow, therefore wqe will be written to wq */
+	cmdq_wqe_fill(curr_cmdq_wqe, &cmdq_wqe);
+
+	cmdq_set_db(cmdq, HINIC_CMDQ_SYNC, next_prod_idx);
+
+	spin_unlock(&cmdq->cmdq_lock);
+	return 0;
+}
+
 static int cmdq_params_valid(struct hinic_cmdq_buf *buf_in)
 {
 	if (buf_in->size > HINIC_CMDQ_MAX_DATA_SIZE)
@@ -389,13 +509,188 @@ int hinic_cmdq_direct_resp(struct hinic_cmdqs *cmdqs,
 }
 
 /**
+ * hinic_set_arm_bit - set arm bit for enable interrupt again
+ * @cmdqs: the cmdqs
+ * @q_type: type of queue to set the arm bit for
+ * @q_id: the queue number
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_set_arm_bit(struct hinic_cmdqs *cmdqs,
+		      enum hinic_set_arm_qtype q_type, u32 q_id)
+{
+	struct hinic_cmdq *cmdq = &cmdqs->cmdq[HINIC_CMDQ_SYNC];
+	struct hinic_hwif *hwif = cmdqs->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_cmdq_arm_bit arm_bit;
+	int err;
+
+	arm_bit.q_type = q_type;
+	arm_bit.q_id   = q_id;
+
+	err = cmdq_set_arm_bit(cmdq, &arm_bit, sizeof(arm_bit));
+	if (err) {
+		dev_err(&pdev->dev, "Failed to set arm for qid %d\n", q_id);
+		return err;
+	}
+
+	return 0;
+}
+
+static void clear_wqe_complete_bit(struct hinic_cmdq *cmdq,
+				   struct hinic_cmdq_wqe *wqe)
+{
+	u32 header_info = be32_to_cpu(CMDQ_WQE_HEADER(wqe)->header_info);
+	unsigned int bufdesc_len, wqe_size;
+	struct hinic_ctrl *ctrl;
+
+	bufdesc_len = HINIC_CMDQ_WQE_HEADER_GET(header_info, BUFDESC_LEN);
+	wqe_size = cmdq_wqe_size_from_bdlen(bufdesc_len);
+	if (wqe_size == WQE_LCMD_SIZE) {
+		struct hinic_cmdq_wqe_lcmd *wqe_lcmd = &wqe->wqe_lcmd;
+
+		ctrl = &wqe_lcmd->ctrl;
+	} else {
+		struct hinic_cmdq_direct_wqe *direct_wqe = &wqe->direct_wqe;
+		struct hinic_cmdq_wqe_scmd *wqe_scmd;
+
+		wqe_scmd = &direct_wqe->wqe_scmd;
+		ctrl = &wqe_scmd->ctrl;
+	}
+
+	/* clear HW busy bit */
+	ctrl->ctrl_info = 0;
+
+	wmb();  /* verify wqe is clear */
+}
+
+/**
+ * cmdq_arm_ceq_handler - cmdq completion event handler for arm command
+ * @cmdq: the cmdq of the arm command
+ * @wqe: the wqe of the arm command
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int cmdq_arm_ceq_handler(struct hinic_cmdq *cmdq,
+				struct hinic_cmdq_wqe *wqe)
+{
+	struct hinic_cmdq_direct_wqe *direct_wqe = &wqe->direct_wqe;
+	struct hinic_cmdq_wqe_scmd *wqe_scmd;
+	struct hinic_ctrl *ctrl;
+	u32 ctrl_info;
+
+	wqe_scmd = &direct_wqe->wqe_scmd;
+	ctrl = &wqe_scmd->ctrl;
+	ctrl_info = be32_to_cpu(ctrl->ctrl_info);
+
+	/* HW should toggle the HW BUSY BIT */
+	if (!CMDQ_WQE_COMPLETED(ctrl_info))
+		return -EBUSY;
+
+	clear_wqe_complete_bit(cmdq, wqe);
+
+	hinic_put_wqe(cmdq->wq, WQE_SCMD_SIZE);
+	return 0;
+}
+
+static void cmdq_update_errcode(struct hinic_cmdq *cmdq, u16 prod_idx,
+				int errcode)
+{
+	if (cmdq->errcode[prod_idx])
+		*cmdq->errcode[prod_idx] = errcode;
+}
+
+/**
+ * cmdq_arm_ceq_handler - cmdq completion event handler for sync command
+ * @cmdq: the cmdq of the command
+ * @cons_idx: the consumer index to update the error code for
+ * @errcode: the error code
+ **/
+static void cmdq_sync_cmd_handler(struct hinic_cmdq *cmdq, u16 cons_idx,
+				  int errcode)
+{
+	u16 prod_idx = cons_idx;
+
+	spin_lock(&cmdq->cmdq_lock);
+	cmdq_update_errcode(cmdq, prod_idx, errcode);
+
+	wmb();  /* write all before update for the command request */
+
+	if (cmdq->done[prod_idx])
+		complete(cmdq->done[prod_idx]);
+	spin_unlock(&cmdq->cmdq_lock);
+}
+
+static int cmdq_cmd_ceq_handler(struct hinic_cmdq *cmdq, u16 ci,
+				struct hinic_cmdq_wqe *cmdq_wqe)
+{
+	struct hinic_cmdq_wqe_lcmd *wqe_lcmd = &cmdq_wqe->wqe_lcmd;
+	struct hinic_status *status = &wqe_lcmd->status;
+	struct hinic_ctrl *ctrl = &wqe_lcmd->ctrl;
+	int errcode;
+
+	if (!CMDQ_WQE_COMPLETED(be32_to_cpu(ctrl->ctrl_info)))
+		return -EBUSY;
+
+	errcode = CMDQ_WQE_ERRCODE_GET(be32_to_cpu(status->status_info), VAL);
+
+	cmdq_sync_cmd_handler(cmdq, ci, errcode);
+
+	clear_wqe_complete_bit(cmdq, cmdq_wqe);
+	hinic_put_wqe(cmdq->wq, WQE_LCMD_SIZE);
+	return 0;
+}
+
+/**
  * cmdq_ceq_handler - cmdq completion event handler
  * @handle: private data for the handler(cmdqs)
  * @ceqe_data: ceq element data
  **/
 static void cmdq_ceq_handler(void *handle, u32 ceqe_data)
 {
-	/* should be implemented */
+	enum hinic_cmdq_type cmdq_type = CMDQ_CEQE_GET(ceqe_data, TYPE);
+	struct hinic_cmdqs *cmdqs = (struct hinic_cmdqs *)handle;
+	struct hinic_cmdq *cmdq = &cmdqs->cmdq[cmdq_type];
+	struct hinic_cmdq_header *header;
+	struct hinic_hw_wqe *hw_wqe;
+	int err, set_arm = 0;
+	u32 saved_data;
+	u16 ci;
+
+	/* Read the smallest wqe size for getting wqe size */
+	while ((hw_wqe = hinic_read_wqe(cmdq->wq, WQE_SCMD_SIZE, &ci))) {
+		if (IS_ERR(hw_wqe))
+			break;
+
+		header = CMDQ_WQE_HEADER(&hw_wqe->cmdq_wqe);
+		saved_data = be32_to_cpu(header->saved_data);
+
+		if (HINIC_SAVED_DATA_GET(saved_data, ARM)) {
+			/* arm_bit was set until here */
+			set_arm = 0;
+
+			if (cmdq_arm_ceq_handler(cmdq, &hw_wqe->cmdq_wqe))
+				break;
+		} else {
+			set_arm = 1;
+
+			hw_wqe = hinic_read_wqe(cmdq->wq, WQE_LCMD_SIZE, &ci);
+			if (IS_ERR(hw_wqe))
+				break;
+
+			if (cmdq_cmd_ceq_handler(cmdq, ci, &hw_wqe->cmdq_wqe))
+				break;
+		}
+	}
+
+	if (set_arm) {
+		struct hinic_hwif *hwif = cmdqs->hwif;
+		struct pci_dev *pdev = hwif->pdev;
+
+		err = hinic_set_arm_bit(cmdqs, HINIC_SET_ARM_CMDQ, cmdq_type);
+		if (err)
+			dev_err(&pdev->dev, "Failed to set arm for CMDQ\n");
+	}
 }
 
 /**
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h
index e11a4f0..b355834 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h
@@ -100,6 +100,10 @@ enum hinic_cmdq_type {
 	HINIC_MAX_CMDQ_TYPES,
 };
 
+enum hinic_set_arm_qtype {
+	HINIC_SET_ARM_CMDQ,
+};
+
 enum hinic_cmd_ack_type {
 	HINIC_CMD_ACK_TYPE_CMDQ,
 };
@@ -110,6 +114,11 @@ struct hinic_cmdq_buf {
 	size_t          size;
 };
 
+struct hinic_cmdq_arm_bit {
+	u32     q_type;
+	u32     q_id;
+};
+
 struct hinic_cmdq_ctxt_info {
 	u64     curr_wqe_page_pfn;
 	u64     wq_block_pfn;
@@ -167,6 +176,9 @@ int hinic_cmdq_direct_resp(struct hinic_cmdqs *cmdqs,
 			   enum hinic_mod_type mod, u8 cmd,
 			   struct hinic_cmdq_buf *buf_in, u64 *out_param);
 
+int hinic_set_arm_bit(struct hinic_cmdqs *cmdqs,
+		      enum hinic_set_arm_qtype q_type, u32 q_id);
+
 int hinic_init_cmdqs(struct hinic_cmdqs *cmdqs, struct hinic_hwif *hwif,
 		     void __iomem **db_area);
 
-- 
1.9.1

^ permalink raw reply related

* [PATCH V7 net-next 16/22] net-next/hinic: Add cmdq commands
From: Aviad Krawczyk @ 2017-08-17 19:04 UTC (permalink / raw)
  To: davem
  Cc: linux-kernel, netdev, bc.y, victor.gissin, aviad.krawczyk,
	zhaochen6, tony.qu
In-Reply-To: <cover.1502996306.git.aviad.krawczyk@huawei.com>

Add cmdq commands for setting queue pair contexts in the nic.

Signed-off-by: Aviad Krawczyk <aviad.krawczyk@huawei.com>
Signed-off-by: Zhao Chen <zhaochen6@huawei.com>
---
 drivers/net/ethernet/huawei/hinic/hinic_common.c  |  25 ++
 drivers/net/ethernet/huawei/hinic/hinic_common.h  |   9 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c | 282 +++++++++++++++++++++-
 drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h |  38 +++
 drivers/net/ethernet/huawei/hinic/hinic_hw_io.h   |  10 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c   | 194 +++++++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h   |  12 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_wqe.h  | 115 +++++++++
 8 files changed, 683 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/huawei/hinic/hinic_common.c b/drivers/net/ethernet/huawei/hinic/hinic_common.c
index 1915ad6..02c74fd 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_common.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_common.c
@@ -13,6 +13,7 @@
  *
  */
 
+#include <linux/kernel.h>
 #include <linux/types.h>
 #include <asm/byteorder.h>
 
@@ -53,3 +54,27 @@ void hinic_be32_to_cpu(void *data, int len)
 		mem++;
 	}
 }
+
+/**
+ * hinic_set_sge - set dma area in scatter gather entry
+ * @sge: scatter gather entry
+ * @addr: dma address
+ * @len: length of relevant data in the dma address
+ **/
+void hinic_set_sge(struct hinic_sge *sge, dma_addr_t addr, int len)
+{
+	sge->hi_addr = upper_32_bits(addr);
+	sge->lo_addr = lower_32_bits(addr);
+	sge->len  = len;
+}
+
+/**
+ * hinic_sge_to_dma - get dma address from scatter gather entry
+ * @sge: scatter gather entry
+ *
+ * Return dma address of sg entry
+ **/
+dma_addr_t hinic_sge_to_dma(struct hinic_sge *sge)
+{
+	return (dma_addr_t)((((u64)sge->hi_addr) << 32) | sge->lo_addr);
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_common.h b/drivers/net/ethernet/huawei/hinic/hinic_common.h
index 0f2f4ff..2c06b76 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_common.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_common.h
@@ -16,6 +16,11 @@
 #ifndef HINIC_COMMON_H
 #define HINIC_COMMON_H
 
+#include <linux/types.h>
+
+#define UPPER_8_BITS(data)      (((data) >> 8) & 0xFF)
+#define LOWER_8_BITS(data)      ((data) & 0xFF)
+
 struct hinic_sge {
 	u32             hi_addr;
 	u32             lo_addr;
@@ -26,4 +31,8 @@ struct hinic_sge {
 
 void hinic_be32_to_cpu(void *data, int len);
 
+void hinic_set_sge(struct hinic_sge *sge, dma_addr_t addr, int len);
+
+dma_addr_t hinic_sge_to_dma(struct hinic_sge *sge);
+
 #endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c
index ec24b95..d8c0807 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c
@@ -24,19 +24,34 @@
 #include <linux/sizes.h>
 #include <linux/atomic.h>
 #include <linux/log2.h>
+#include <linux/io.h>
+#include <linux/completion.h>
+#include <linux/err.h>
 #include <asm/byteorder.h>
+#include <asm/barrier.h>
 
+#include "hinic_common.h"
 #include "hinic_hw_if.h"
 #include "hinic_hw_eqs.h"
 #include "hinic_hw_mgmt.h"
+#include "hinic_hw_wqe.h"
 #include "hinic_hw_wq.h"
 #include "hinic_hw_cmdq.h"
 #include "hinic_hw_io.h"
 #include "hinic_hw_dev.h"
 
+#define CMDQ_DB_PI_OFF(pi)              (((u16)LOWER_8_BITS(pi)) << 3)
+
+#define CMDQ_DB_ADDR(db_base, pi)       ((db_base) + CMDQ_DB_PI_OFF(pi))
+
+#define CMDQ_WQE_HEADER(wqe)            ((struct hinic_cmdq_header *)(wqe))
+
+#define FIRST_DATA_TO_WRITE_LAST        sizeof(u64)
+
 #define CMDQ_DB_OFF                     SZ_2K
 
 #define CMDQ_WQEBB_SIZE                 64
+#define CMDQ_WQE_SIZE                   64
 #define CMDQ_DEPTH                      SZ_4K
 
 #define CMDQ_WQ_PAGE_SIZE               SZ_4K
@@ -44,6 +59,10 @@
 #define WQE_LCMD_SIZE                   64
 #define WQE_SCMD_SIZE                   64
 
+#define COMPLETE_LEN                    3
+
+#define CMDQ_TIMEOUT                    1000
+
 #define CMDQ_PFN(addr, page_size)       ((addr) >> (ilog2(page_size)))
 
 #define cmdq_to_cmdqs(cmdq)     container_of((cmdq) - (cmdq)->cmdq_type, \
@@ -58,6 +77,40 @@ enum cmdq_wqe_type {
 	WQE_SCMD_TYPE = 1,
 };
 
+enum completion_format {
+	COMPLETE_DIRECT = 0,
+	COMPLETE_SGE    = 1,
+};
+
+enum data_format {
+	DATA_SGE        = 0,
+	DATA_DIRECT     = 1,
+};
+
+enum bufdesc_len {
+	BUFDESC_LCMD_LEN = 2,   /* 16 bytes - 2(8 byte unit) */
+	BUFDESC_SCMD_LEN = 3,   /* 24 bytes - 3(8 byte unit) */
+};
+
+enum ctrl_sect_len {
+	CTRL_SECT_LEN        = 1, /* 4 bytes (ctrl) - 1(8 byte unit) */
+	CTRL_DIRECT_SECT_LEN = 2, /* 12 bytes (ctrl + rsvd) - 2(8 byte unit) */
+};
+
+enum cmdq_scmd_type {
+	CMDQ_SET_ARM_CMD = 2,
+};
+
+enum cmdq_cmd_type {
+	CMDQ_CMD_SYNC_DIRECT_RESP = 0,
+	CMDQ_CMD_SYNC_SGE_RESP    = 1,
+};
+
+enum completion_request {
+	NO_CEQ  = 0,
+	CEQ_SET = 1,
+};
+
 /**
  * hinic_alloc_cmdq_buf - alloc buffer for sending command
  * @cmdqs: the cmdqs
@@ -92,6 +145,221 @@ void hinic_free_cmdq_buf(struct hinic_cmdqs *cmdqs,
 	pci_pool_free(cmdqs->cmdq_buf_pool, cmdq_buf->buf, cmdq_buf->dma_addr);
 }
 
+static void cmdq_set_sge_completion(struct hinic_cmdq_completion *completion,
+				    struct hinic_cmdq_buf *buf_out)
+{
+	struct hinic_sge_resp *sge_resp = &completion->sge_resp;
+
+	hinic_set_sge(&sge_resp->sge, buf_out->dma_addr,
+		      buf_out->size);
+}
+
+static void cmdq_prepare_wqe_ctrl(struct hinic_cmdq_wqe *wqe, int wrapped,
+				  enum hinic_cmd_ack_type ack_type,
+				  enum hinic_mod_type mod, u8 cmd, u16 prod_idx,
+				  enum completion_format complete_format,
+				  enum data_format data_format,
+				  enum bufdesc_len buf_len)
+{
+	struct hinic_cmdq_wqe_lcmd *wqe_lcmd;
+	struct hinic_cmdq_wqe_scmd *wqe_scmd;
+	enum ctrl_sect_len ctrl_len;
+	struct hinic_ctrl *ctrl;
+	u32 saved_data;
+
+	if (data_format == DATA_SGE) {
+		wqe_lcmd = &wqe->wqe_lcmd;
+
+		wqe_lcmd->status.status_info = 0;
+		ctrl = &wqe_lcmd->ctrl;
+		ctrl_len = CTRL_SECT_LEN;
+	} else {
+		wqe_scmd = &wqe->direct_wqe.wqe_scmd;
+
+		wqe_scmd->status.status_info = 0;
+		ctrl = &wqe_scmd->ctrl;
+		ctrl_len = CTRL_DIRECT_SECT_LEN;
+	}
+
+	ctrl->ctrl_info = HINIC_CMDQ_CTRL_SET(prod_idx, PI)             |
+			  HINIC_CMDQ_CTRL_SET(cmd, CMD)                 |
+			  HINIC_CMDQ_CTRL_SET(mod, MOD)                 |
+			  HINIC_CMDQ_CTRL_SET(ack_type, ACK_TYPE);
+
+	CMDQ_WQE_HEADER(wqe)->header_info =
+		HINIC_CMDQ_WQE_HEADER_SET(buf_len, BUFDESC_LEN)            |
+		HINIC_CMDQ_WQE_HEADER_SET(complete_format, COMPLETE_FMT)   |
+		HINIC_CMDQ_WQE_HEADER_SET(data_format, DATA_FMT)           |
+		HINIC_CMDQ_WQE_HEADER_SET(CEQ_SET, COMPLETE_REQ)           |
+		HINIC_CMDQ_WQE_HEADER_SET(COMPLETE_LEN, COMPLETE_SECT_LEN) |
+		HINIC_CMDQ_WQE_HEADER_SET(ctrl_len, CTRL_LEN)              |
+		HINIC_CMDQ_WQE_HEADER_SET(wrapped, TOGGLED_WRAPPED);
+
+	saved_data = CMDQ_WQE_HEADER(wqe)->saved_data;
+	saved_data = HINIC_SAVED_DATA_CLEAR(saved_data, ARM);
+
+	if ((cmd == CMDQ_SET_ARM_CMD) && (mod == HINIC_MOD_COMM))
+		CMDQ_WQE_HEADER(wqe)->saved_data |=
+						HINIC_SAVED_DATA_SET(1, ARM);
+	else
+		CMDQ_WQE_HEADER(wqe)->saved_data = saved_data;
+}
+
+static void cmdq_set_lcmd_bufdesc(struct hinic_cmdq_wqe_lcmd *wqe_lcmd,
+				  struct hinic_cmdq_buf *buf_in)
+{
+	hinic_set_sge(&wqe_lcmd->buf_desc.sge, buf_in->dma_addr, buf_in->size);
+}
+
+static void cmdq_set_lcmd_wqe(struct hinic_cmdq_wqe *wqe,
+			      enum cmdq_cmd_type cmd_type,
+			      struct hinic_cmdq_buf *buf_in,
+			      struct hinic_cmdq_buf *buf_out, int wrapped,
+			      enum hinic_cmd_ack_type ack_type,
+			      enum hinic_mod_type mod, u8 cmd, u16 prod_idx)
+{
+	struct hinic_cmdq_wqe_lcmd *wqe_lcmd = &wqe->wqe_lcmd;
+	enum completion_format complete_format;
+
+	switch (cmd_type) {
+	case CMDQ_CMD_SYNC_SGE_RESP:
+		complete_format = COMPLETE_SGE;
+		cmdq_set_sge_completion(&wqe_lcmd->completion, buf_out);
+		break;
+	case CMDQ_CMD_SYNC_DIRECT_RESP:
+		complete_format = COMPLETE_DIRECT;
+		wqe_lcmd->completion.direct_resp = 0;
+		break;
+	}
+
+	cmdq_prepare_wqe_ctrl(wqe, wrapped, ack_type, mod, cmd,
+			      prod_idx, complete_format, DATA_SGE,
+			      BUFDESC_LCMD_LEN);
+
+	cmdq_set_lcmd_bufdesc(wqe_lcmd, buf_in);
+}
+
+static void cmdq_wqe_fill(void *dst, void *src)
+{
+	memcpy(dst + FIRST_DATA_TO_WRITE_LAST, src + FIRST_DATA_TO_WRITE_LAST,
+	       CMDQ_WQE_SIZE - FIRST_DATA_TO_WRITE_LAST);
+
+	wmb();          /* The first 8 bytes should be written last */
+
+	*(u64 *)dst = *(u64 *)src;
+}
+
+static void cmdq_fill_db(u32 *db_info,
+			 enum hinic_cmdq_type cmdq_type, u16 prod_idx)
+{
+	*db_info = HINIC_CMDQ_DB_INFO_SET(UPPER_8_BITS(prod_idx), HI_PROD_IDX) |
+		   HINIC_CMDQ_DB_INFO_SET(HINIC_CTRL_PATH, PATH)               |
+		   HINIC_CMDQ_DB_INFO_SET(cmdq_type, CMDQ_TYPE)                |
+		   HINIC_CMDQ_DB_INFO_SET(HINIC_DB_CMDQ_TYPE, DB_TYPE);
+}
+
+static void cmdq_set_db(struct hinic_cmdq *cmdq,
+			enum hinic_cmdq_type cmdq_type, u16 prod_idx)
+{
+	u32 db_info;
+
+	cmdq_fill_db(&db_info, cmdq_type, prod_idx);
+
+	/* The data that is written to HW should be in Big Endian Format */
+	db_info = cpu_to_be32(db_info);
+
+	wmb();  /* write all before the doorbell */
+
+	writel(db_info, CMDQ_DB_ADDR(cmdq->db_base, prod_idx));
+}
+
+static int cmdq_sync_cmd_direct_resp(struct hinic_cmdq *cmdq,
+				     enum hinic_mod_type mod, u8 cmd,
+				     struct hinic_cmdq_buf *buf_in,
+				     u64 *resp)
+{
+	struct hinic_cmdq_wqe *curr_cmdq_wqe, cmdq_wqe;
+	u16 curr_prod_idx, next_prod_idx;
+	int errcode, wrapped, num_wqebbs;
+	struct hinic_wq *wq = cmdq->wq;
+	struct hinic_hw_wqe *hw_wqe;
+	struct completion done;
+
+	/* Keep doorbell index correct. bh - for tasklet(ceq). */
+	spin_lock_bh(&cmdq->cmdq_lock);
+
+	/* WQE_SIZE = WQEBB_SIZE, we will get the wq element and not shadow*/
+	hw_wqe = hinic_get_wqe(wq, WQE_LCMD_SIZE, &curr_prod_idx);
+	if (IS_ERR(hw_wqe)) {
+		spin_unlock_bh(&cmdq->cmdq_lock);
+		return -EBUSY;
+	}
+
+	curr_cmdq_wqe = &hw_wqe->cmdq_wqe;
+
+	wrapped = cmdq->wrapped;
+
+	num_wqebbs = ALIGN(WQE_LCMD_SIZE, wq->wqebb_size) / wq->wqebb_size;
+	next_prod_idx = curr_prod_idx + num_wqebbs;
+	if (next_prod_idx >= wq->q_depth) {
+		cmdq->wrapped = !cmdq->wrapped;
+		next_prod_idx -= wq->q_depth;
+	}
+
+	cmdq->errcode[curr_prod_idx] = &errcode;
+
+	init_completion(&done);
+	cmdq->done[curr_prod_idx] = &done;
+
+	cmdq_set_lcmd_wqe(&cmdq_wqe, CMDQ_CMD_SYNC_DIRECT_RESP, buf_in, NULL,
+			  wrapped, HINIC_CMD_ACK_TYPE_CMDQ, mod, cmd,
+			  curr_prod_idx);
+
+	/* The data that is written to HW should be in Big Endian Format */
+	hinic_cpu_to_be32(&cmdq_wqe, WQE_LCMD_SIZE);
+
+	/* CMDQ WQE is not shadow, therefore wqe will be written to wq */
+	cmdq_wqe_fill(curr_cmdq_wqe, &cmdq_wqe);
+
+	cmdq_set_db(cmdq, HINIC_CMDQ_SYNC, next_prod_idx);
+
+	spin_unlock_bh(&cmdq->cmdq_lock);
+
+	if (!wait_for_completion_timeout(&done, CMDQ_TIMEOUT)) {
+		spin_lock_bh(&cmdq->cmdq_lock);
+
+		if (cmdq->errcode[curr_prod_idx] == &errcode)
+			cmdq->errcode[curr_prod_idx] = NULL;
+
+		if (cmdq->done[curr_prod_idx] == &done)
+			cmdq->done[curr_prod_idx] = NULL;
+
+		spin_unlock_bh(&cmdq->cmdq_lock);
+
+		return -ETIMEDOUT;
+	}
+
+	smp_rmb();      /* read error code after completion */
+
+	if (resp) {
+		struct hinic_cmdq_wqe_lcmd *wqe_lcmd = &curr_cmdq_wqe->wqe_lcmd;
+		*resp = cpu_to_be64(wqe_lcmd->completion.direct_resp);
+	}
+
+	if (errcode != 0)
+		return -EFAULT;
+
+	return 0;
+}
+
+static int cmdq_params_valid(struct hinic_cmdq_buf *buf_in)
+{
+	if (buf_in->size > HINIC_CMDQ_MAX_DATA_SIZE)
+		return -EINVAL;
+
+	return 0;
+}
+
 /**
  * hinic_cmdq_direct_resp - send command with direct data as resp
  * @cmdqs: the cmdqs
@@ -106,8 +374,18 @@ int hinic_cmdq_direct_resp(struct hinic_cmdqs *cmdqs,
 			   enum hinic_mod_type mod, u8 cmd,
 			   struct hinic_cmdq_buf *buf_in, u64 *resp)
 {
-	/* should be implemented */
-	return -EINVAL;
+	struct hinic_hwif *hwif = cmdqs->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	int err;
+
+	err = cmdq_params_valid(buf_in);
+	if (err) {
+		dev_err(&pdev->dev, "Invalid CMDQ parameters\n");
+		return err;
+	}
+
+	return cmdq_sync_cmd_direct_resp(&cmdqs->cmdq[HINIC_CMDQ_SYNC],
+					 mod, cmd, buf_in, resp);
 }
 
 /**
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h
index 5ec59f1..e11a4f0 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h
@@ -58,14 +58,52 @@
 			((val) & (~((u64)HINIC_CMDQ_CTXT_##member##_MASK \
 			 << HINIC_CMDQ_CTXT_##member##_SHIFT)))
 
+#define HINIC_SAVED_DATA_ARM_SHIFT                      31
+
+#define HINIC_SAVED_DATA_ARM_MASK                       0x1
+
+#define HINIC_SAVED_DATA_SET(val, member)               \
+			(((u32)(val) & HINIC_SAVED_DATA_##member##_MASK) \
+			 << HINIC_SAVED_DATA_##member##_SHIFT)
+
+#define HINIC_SAVED_DATA_GET(val, member)               \
+			(((val) >> HINIC_SAVED_DATA_##member##_SHIFT) \
+			 & HINIC_SAVED_DATA_##member##_MASK)
+
+#define HINIC_SAVED_DATA_CLEAR(val, member)             \
+			((val) & (~(HINIC_SAVED_DATA_##member##_MASK \
+			 << HINIC_SAVED_DATA_##member##_SHIFT)))
+
+#define HINIC_CMDQ_DB_INFO_HI_PROD_IDX_SHIFT            0
+#define HINIC_CMDQ_DB_INFO_PATH_SHIFT                   23
+#define HINIC_CMDQ_DB_INFO_CMDQ_TYPE_SHIFT              24
+#define HINIC_CMDQ_DB_INFO_DB_TYPE_SHIFT                27
+
+#define HINIC_CMDQ_DB_INFO_HI_PROD_IDX_MASK             0xFF
+#define HINIC_CMDQ_DB_INFO_PATH_MASK                    0x1
+#define HINIC_CMDQ_DB_INFO_CMDQ_TYPE_MASK               0x7
+#define HINIC_CMDQ_DB_INFO_DB_TYPE_MASK                 0x1F
+
+#define HINIC_CMDQ_DB_INFO_SET(val, member)             \
+			(((u32)(val) & HINIC_CMDQ_DB_INFO_##member##_MASK) \
+			 << HINIC_CMDQ_DB_INFO_##member##_SHIFT)
+
 #define HINIC_CMDQ_BUF_SIZE             2048
 
+#define HINIC_CMDQ_BUF_HW_RSVD          8
+#define HINIC_CMDQ_MAX_DATA_SIZE        (HINIC_CMDQ_BUF_SIZE - \
+					 HINIC_CMDQ_BUF_HW_RSVD)
+
 enum hinic_cmdq_type {
 	HINIC_CMDQ_SYNC,
 
 	HINIC_MAX_CMDQ_TYPES,
 };
 
+enum hinic_cmd_ack_type {
+	HINIC_CMD_ACK_TYPE_CMDQ,
+};
+
 struct hinic_cmdq_buf {
 	void            *buf;
 	dma_addr_t      dma_addr;
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
index cfc21ba..adb6417 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
@@ -32,6 +32,16 @@
 
 #define HINIC_DB_MAX_AREAS      (HINIC_DB_SIZE / HINIC_DB_PAGE_SIZE)
 
+enum hinic_db_type {
+	HINIC_DB_CMDQ_TYPE,
+	HINIC_DB_SQ_TYPE,
+};
+
+enum hinic_io_path {
+	HINIC_CTRL_PATH,
+	HINIC_DATA_PATH,
+};
+
 struct hinic_free_db_area {
 	int             db_idx[HINIC_DB_MAX_AREAS];
 
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c
index 5a3689a..a22e201 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c
@@ -23,9 +23,11 @@
 #include <linux/semaphore.h>
 #include <linux/errno.h>
 #include <linux/vmalloc.h>
+#include <linux/err.h>
 #include <asm/byteorder.h>
 
 #include "hinic_hw_if.h"
+#include "hinic_hw_wqe.h"
 #include "hinic_hw_wq.h"
 #include "hinic_hw_cmdq.h"
 
@@ -72,6 +74,25 @@
 			((void *)((cmdq_pages)->shadow_page_vaddr) \
 				+ (wq)->block_idx * CMDQ_BLOCK_SIZE)
 
+#define WQE_PAGE_OFF(wq, idx)   (((idx) & ((wq)->num_wqebbs_per_page - 1)) * \
+					(wq)->wqebb_size)
+
+#define WQE_PAGE_NUM(wq, idx)   (((idx) / ((wq)->num_wqebbs_per_page)) \
+					& ((wq)->num_q_pages - 1))
+
+#define WQ_PAGE_ADDR(wq, idx)           \
+			((wq)->shadow_block_vaddr[WQE_PAGE_NUM(wq, idx)])
+
+#define MASKED_WQE_IDX(wq, idx)         ((idx) & (wq)->mask)
+
+#define WQE_IN_RANGE(wqe, start, end)   \
+		(((unsigned long)(wqe) >= (unsigned long)(start)) && \
+		 ((unsigned long)(wqe) < (unsigned long)(end)))
+
+#define WQE_SHADOW_PAGE(wq, wqe)        \
+		(((unsigned long)(wqe) - (unsigned long)(wq)->shadow_wqe) \
+			/ (wq)->max_wqe_size)
+
 /**
  * queue_alloc_page - allocate page for Queue
  * @hwif: HW interface for allocating DMA
@@ -671,3 +692,176 @@ void hinic_wqs_cmdq_free(struct hinic_cmdq_pages *cmdq_pages,
 
 	cmdq_free_page(cmdq_pages);
 }
+
+static void copy_wqe_to_shadow(struct hinic_wq *wq, void *shadow_addr,
+			       int num_wqebbs, u16 idx)
+{
+	void *wqebb_addr;
+	int i;
+
+	for (i = 0; i < num_wqebbs; i++, idx++) {
+		idx = MASKED_WQE_IDX(wq, idx);
+		wqebb_addr = WQ_PAGE_ADDR(wq, idx) +
+			     WQE_PAGE_OFF(wq, idx);
+
+		memcpy(shadow_addr, wqebb_addr, wq->wqebb_size);
+
+		shadow_addr += wq->wqebb_size;
+	}
+}
+
+static void copy_wqe_from_shadow(struct hinic_wq *wq, void *shadow_addr,
+				 int num_wqebbs, u16 idx)
+{
+	void *wqebb_addr;
+	int i;
+
+	for (i = 0; i < num_wqebbs; i++, idx++) {
+		idx = MASKED_WQE_IDX(wq, idx);
+		wqebb_addr = WQ_PAGE_ADDR(wq, idx) +
+			     WQE_PAGE_OFF(wq, idx);
+
+		memcpy(wqebb_addr, shadow_addr, wq->wqebb_size);
+		shadow_addr += wq->wqebb_size;
+	}
+}
+
+/**
+ * hinic_get_wqe - get wqe ptr in the current pi and update the pi
+ * @wq: wq to get wqe from
+ * @wqe_size: wqe size
+ * @prod_idx: returned pi
+ *
+ * Return wqe pointer
+ **/
+struct hinic_hw_wqe *hinic_get_wqe(struct hinic_wq *wq, unsigned int wqe_size,
+				   u16 *prod_idx)
+{
+	int curr_pg, end_pg, num_wqebbs;
+	u16 curr_prod_idx, end_prod_idx;
+
+	*prod_idx = MASKED_WQE_IDX(wq, atomic_read(&wq->prod_idx));
+
+	num_wqebbs = ALIGN(wqe_size, wq->wqebb_size) / wq->wqebb_size;
+
+	if (atomic_sub_return(num_wqebbs, &wq->delta) <= 0) {
+		atomic_add(num_wqebbs, &wq->delta);
+		return ERR_PTR(-EBUSY);
+	}
+
+	end_prod_idx = atomic_add_return(num_wqebbs, &wq->prod_idx);
+
+	end_prod_idx = MASKED_WQE_IDX(wq, end_prod_idx);
+	curr_prod_idx = end_prod_idx - num_wqebbs;
+	curr_prod_idx = MASKED_WQE_IDX(wq, curr_prod_idx);
+
+	/* end prod index points to the next wqebb, therefore minus 1 */
+	end_prod_idx = MASKED_WQE_IDX(wq, end_prod_idx - 1);
+
+	curr_pg = WQE_PAGE_NUM(wq, curr_prod_idx);
+	end_pg = WQE_PAGE_NUM(wq, end_prod_idx);
+
+	*prod_idx = curr_prod_idx;
+
+	if (curr_pg != end_pg) {
+		void *shadow_addr = &wq->shadow_wqe[curr_pg * wq->max_wqe_size];
+
+		copy_wqe_to_shadow(wq, shadow_addr, num_wqebbs, *prod_idx);
+
+		wq->shadow_idx[curr_pg] = *prod_idx;
+		return shadow_addr;
+	}
+
+	return WQ_PAGE_ADDR(wq, *prod_idx) + WQE_PAGE_OFF(wq, *prod_idx);
+}
+
+/**
+ * hinic_put_wqe - return the wqe place to use for a new wqe
+ * @wq: wq to return wqe
+ * @wqe_size: wqe size
+ **/
+void hinic_put_wqe(struct hinic_wq *wq, unsigned int wqe_size)
+{
+	int num_wqebbs = ALIGN(wqe_size, wq->wqebb_size) / wq->wqebb_size;
+
+	atomic_add(num_wqebbs, &wq->cons_idx);
+
+	atomic_add(num_wqebbs, &wq->delta);
+}
+
+/**
+ * hinic_read_wqe - read wqe ptr in the current ci
+ * @wq: wq to get read from
+ * @wqe_size: wqe size
+ * @cons_idx: returned ci
+ *
+ * Return wqe pointer
+ **/
+struct hinic_hw_wqe *hinic_read_wqe(struct hinic_wq *wq, unsigned int wqe_size,
+				    u16 *cons_idx)
+{
+	int num_wqebbs = ALIGN(wqe_size, wq->wqebb_size) / wq->wqebb_size;
+	u16 curr_cons_idx, end_cons_idx;
+	int curr_pg, end_pg;
+
+	if ((atomic_read(&wq->delta) + num_wqebbs) > wq->q_depth)
+		return ERR_PTR(-EBUSY);
+
+	curr_cons_idx = atomic_read(&wq->cons_idx);
+
+	curr_cons_idx = MASKED_WQE_IDX(wq, curr_cons_idx);
+	end_cons_idx = MASKED_WQE_IDX(wq, curr_cons_idx + num_wqebbs - 1);
+
+	curr_pg = WQE_PAGE_NUM(wq, curr_cons_idx);
+	end_pg = WQE_PAGE_NUM(wq, end_cons_idx);
+
+	*cons_idx = curr_cons_idx;
+
+	if (curr_pg != end_pg) {
+		void *shadow_addr = &wq->shadow_wqe[curr_pg * wq->max_wqe_size];
+
+		copy_wqe_to_shadow(wq, shadow_addr, num_wqebbs, *cons_idx);
+		return shadow_addr;
+	}
+
+	return WQ_PAGE_ADDR(wq, *cons_idx) + WQE_PAGE_OFF(wq, *cons_idx);
+}
+
+/**
+ * wqe_shadow - check if a wqe is shadow
+ * @wq: wq of the wqe
+ * @wqe: the wqe for shadow checking
+ *
+ * Return true - shadow, false - Not shadow
+ **/
+static inline bool wqe_shadow(struct hinic_wq *wq, struct hinic_hw_wqe *wqe)
+{
+	size_t wqe_shadow_size = wq->num_q_pages * wq->max_wqe_size;
+
+	return WQE_IN_RANGE(wqe, wq->shadow_wqe,
+			    &wq->shadow_wqe[wqe_shadow_size]);
+}
+
+/**
+ * hinic_write_wqe - write the wqe to the wq
+ * @wq: wq to write wqe to
+ * @wqe: wqe to write
+ * @wqe_size: wqe size
+ **/
+void hinic_write_wqe(struct hinic_wq *wq, struct hinic_hw_wqe *wqe,
+		     unsigned int wqe_size)
+{
+	int curr_pg, num_wqebbs;
+	void *shadow_addr;
+	u16 prod_idx;
+
+	if (wqe_shadow(wq, wqe)) {
+		curr_pg = WQE_SHADOW_PAGE(wq, wqe);
+
+		prod_idx = wq->shadow_idx[curr_pg];
+		num_wqebbs = ALIGN(wqe_size, wq->wqebb_size) / wq->wqebb_size;
+		shadow_addr = &wq->shadow_wqe[curr_pg * wq->max_wqe_size];
+
+		copy_wqe_from_shadow(wq, shadow_addr, num_wqebbs, prod_idx);
+	}
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h
index a3c4469..f01477a 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h
@@ -21,6 +21,7 @@
 #include <linux/atomic.h>
 
 #include "hinic_hw_if.h"
+#include "hinic_hw_wqe.h"
 
 struct hinic_free_block {
 	int     page_idx;
@@ -100,4 +101,15 @@ int hinic_wq_allocate(struct hinic_wqs *wqs, struct hinic_wq *wq,
 
 void hinic_wq_free(struct hinic_wqs *wqs, struct hinic_wq *wq);
 
+struct hinic_hw_wqe *hinic_get_wqe(struct hinic_wq *wq, unsigned int wqe_size,
+				   u16 *prod_idx);
+
+void hinic_put_wqe(struct hinic_wq *wq, unsigned int wqe_size);
+
+struct hinic_hw_wqe *hinic_read_wqe(struct hinic_wq *wq, unsigned int wqe_size,
+				    u16 *cons_idx);
+
+void hinic_write_wqe(struct hinic_wq *wq, struct hinic_hw_wqe *wqe,
+		     unsigned int wqe_size);
+
 #endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_wqe.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_wqe.h
index d727c4d..bc73485 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_wqe.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_wqe.h
@@ -18,6 +18,50 @@
 
 #include "hinic_common.h"
 
+#define HINIC_CMDQ_CTRL_PI_SHIFT                        0
+#define HINIC_CMDQ_CTRL_CMD_SHIFT                       16
+#define HINIC_CMDQ_CTRL_MOD_SHIFT                       24
+#define HINIC_CMDQ_CTRL_ACK_TYPE_SHIFT                  29
+#define HINIC_CMDQ_CTRL_HW_BUSY_BIT_SHIFT               31
+
+#define HINIC_CMDQ_CTRL_PI_MASK                         0xFFFF
+#define HINIC_CMDQ_CTRL_CMD_MASK                        0xFF
+#define HINIC_CMDQ_CTRL_MOD_MASK                        0x1F
+#define HINIC_CMDQ_CTRL_ACK_TYPE_MASK                   0x3
+#define HINIC_CMDQ_CTRL_HW_BUSY_BIT_MASK                0x1
+
+#define HINIC_CMDQ_CTRL_SET(val, member)                        \
+			(((u32)(val) & HINIC_CMDQ_CTRL_##member##_MASK) \
+			 << HINIC_CMDQ_CTRL_##member##_SHIFT)
+
+#define HINIC_CMDQ_CTRL_GET(val, member)                        \
+			(((val) >> HINIC_CMDQ_CTRL_##member##_SHIFT) \
+			 & HINIC_CMDQ_CTRL_##member##_MASK)
+
+#define HINIC_CMDQ_WQE_HEADER_BUFDESC_LEN_SHIFT         0
+#define HINIC_CMDQ_WQE_HEADER_COMPLETE_FMT_SHIFT        15
+#define HINIC_CMDQ_WQE_HEADER_DATA_FMT_SHIFT            22
+#define HINIC_CMDQ_WQE_HEADER_COMPLETE_REQ_SHIFT        23
+#define HINIC_CMDQ_WQE_HEADER_COMPLETE_SECT_LEN_SHIFT   27
+#define HINIC_CMDQ_WQE_HEADER_CTRL_LEN_SHIFT            29
+#define HINIC_CMDQ_WQE_HEADER_TOGGLED_WRAPPED_SHIFT     31
+
+#define HINIC_CMDQ_WQE_HEADER_BUFDESC_LEN_MASK          0xFF
+#define HINIC_CMDQ_WQE_HEADER_COMPLETE_FMT_MASK         0x1
+#define HINIC_CMDQ_WQE_HEADER_DATA_FMT_MASK             0x1
+#define HINIC_CMDQ_WQE_HEADER_COMPLETE_REQ_MASK         0x1
+#define HINIC_CMDQ_WQE_HEADER_COMPLETE_SECT_LEN_MASK    0x3
+#define HINIC_CMDQ_WQE_HEADER_CTRL_LEN_MASK             0x3
+#define HINIC_CMDQ_WQE_HEADER_TOGGLED_WRAPPED_MASK      0x1
+
+#define HINIC_CMDQ_WQE_HEADER_SET(val, member)                  \
+			(((u32)(val) & HINIC_CMDQ_WQE_HEADER_##member##_MASK) \
+			 << HINIC_CMDQ_WQE_HEADER_##member##_SHIFT)
+
+#define HINIC_CMDQ_WQE_HEADER_GET(val, member)                  \
+			(((val) >> HINIC_CMDQ_WQE_HEADER_##member##_SHIFT) \
+			 & HINIC_CMDQ_WQE_HEADER_##member##_MASK)
+
 #define HINIC_SQ_CTRL_BUFDESC_SECT_LEN_SHIFT    0
 #define HINIC_SQ_CTRL_TASKSECT_LEN_SHIFT        16
 #define HINIC_SQ_CTRL_DATA_FORMAT_SHIFT         22
@@ -143,6 +187,8 @@
 		 sizeof(struct hinic_sq_task) + \
 		 (nr_sges) * sizeof(struct hinic_sq_bufdesc))
 
+#define HINIC_SCMD_DATA_LEN             16
+
 #define HINIC_MAX_SQ_BUFDESCS           17
 
 #define HINIC_SQ_WQE_MAX_SIZE           320
@@ -184,6 +230,74 @@ enum hinc_tunnel_l4type {
 	HINIC_TUNNEL_L4TYPE_UNKNOWN = 0,
 };
 
+struct hinic_cmdq_header {
+	u32     header_info;
+	u32     saved_data;
+};
+
+struct hinic_status {
+	u32 status_info;
+};
+
+struct hinic_ctrl {
+	u32 ctrl_info;
+};
+
+struct hinic_sge_resp {
+	struct hinic_sge        sge;
+	u32                     rsvd;
+};
+
+struct hinic_cmdq_completion {
+	/* HW Format */
+	union {
+		struct hinic_sge_resp   sge_resp;
+		u64                     direct_resp;
+	};
+};
+
+struct hinic_scmd_bufdesc {
+	u32     buf_len;
+	u32     rsvd;
+	u8      data[HINIC_SCMD_DATA_LEN];
+};
+
+struct hinic_lcmd_bufdesc {
+	struct hinic_sge        sge;
+	u32                     rsvd1;
+	u64                     rsvd2;
+	u64                     rsvd3;
+};
+
+struct hinic_cmdq_wqe_scmd {
+	struct hinic_cmdq_header        header;
+	u64                             rsvd;
+	struct hinic_status             status;
+	struct hinic_ctrl               ctrl;
+	struct hinic_cmdq_completion    completion;
+	struct hinic_scmd_bufdesc       buf_desc;
+};
+
+struct hinic_cmdq_wqe_lcmd {
+	struct hinic_cmdq_header        header;
+	struct hinic_status             status;
+	struct hinic_ctrl               ctrl;
+	struct hinic_cmdq_completion    completion;
+	struct hinic_lcmd_bufdesc       buf_desc;
+};
+
+struct hinic_cmdq_direct_wqe {
+	struct hinic_cmdq_wqe_scmd      wqe_scmd;
+};
+
+struct hinic_cmdq_wqe {
+	/* HW Format */
+	union {
+		struct hinic_cmdq_direct_wqe    direct_wqe;
+		struct hinic_cmdq_wqe_lcmd      wqe_lcmd;
+	};
+};
+
 struct hinic_sq_ctrl {
 	u32     ctrl_info;
 	u32     queue_info;
@@ -245,6 +359,7 @@ struct hinic_rq_wqe {
 struct hinic_hw_wqe {
 	/* HW Format */
 	union {
+		struct hinic_cmdq_wqe   cmdq_wqe;
 		struct hinic_sq_wqe     sq_wqe;
 		struct hinic_rq_wqe     rq_wqe;
 	};
-- 
1.9.1

^ permalink raw reply related

* [PATCH V7 net-next 15/22] net-next/hinic: Add ceqs
From: Aviad Krawczyk @ 2017-08-17 19:04 UTC (permalink / raw)
  To: davem
  Cc: linux-kernel, netdev, bc.y, victor.gissin, aviad.krawczyk,
	zhaochen6, tony.qu
In-Reply-To: <cover.1502996306.git.aviad.krawczyk@huawei.com>

Initialize the completion event queues and handle ceq events by calling
the registered handlers. Used for cmdq command completion.

Signed-off-by: Aviad Krawczyk <aviad.krawczyk@huawei.com>
Signed-off-by: Zhao Chen <zhaochen6@huawei.com>
---
 drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c |  16 ++
 drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h  |  29 +++
 drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c  |   7 +-
 drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.c  | 291 +++++++++++++++++++++-
 drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.h  |  75 ++++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_io.c   |  15 +-
 drivers/net/ethernet/huawei/hinic/hinic_hw_io.h   |   3 +
 7 files changed, 428 insertions(+), 8 deletions(-)

diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c
index 0dccbe6..ec24b95 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c
@@ -27,6 +27,7 @@
 #include <asm/byteorder.h>
 
 #include "hinic_hw_if.h"
+#include "hinic_hw_eqs.h"
 #include "hinic_hw_mgmt.h"
 #include "hinic_hw_wq.h"
 #include "hinic_hw_cmdq.h"
@@ -110,6 +111,16 @@ int hinic_cmdq_direct_resp(struct hinic_cmdqs *cmdqs,
 }
 
 /**
+ * cmdq_ceq_handler - cmdq completion event handler
+ * @handle: private data for the handler(cmdqs)
+ * @ceqe_data: ceq element data
+ **/
+static void cmdq_ceq_handler(void *handle, u32 ceqe_data)
+{
+	/* should be implemented */
+}
+
+/**
  * cmdq_init_queue_ctxt - init the queue ctxt of a cmdq
  * @cmdq_ctxt: cmdq ctxt to initialize
  * @cmdq: the cmdq
@@ -320,6 +331,8 @@ int hinic_init_cmdqs(struct hinic_cmdqs *cmdqs, struct hinic_hwif *hwif,
 		goto err_cmdq_ctxt;
 	}
 
+	hinic_ceq_register_cb(&func_to_io->ceqs, HINIC_CEQ_CMDQ, cmdqs,
+			      cmdq_ceq_handler);
 	return 0;
 
 err_cmdq_ctxt:
@@ -340,10 +353,13 @@ int hinic_init_cmdqs(struct hinic_cmdqs *cmdqs, struct hinic_hwif *hwif,
  **/
 void hinic_free_cmdqs(struct hinic_cmdqs *cmdqs)
 {
+	struct hinic_func_to_io *func_to_io = cmdqs_to_func_to_io(cmdqs);
 	struct hinic_hwif *hwif = cmdqs->hwif;
 	struct pci_dev *pdev = hwif->pdev;
 	enum hinic_cmdq_type cmdq_type;
 
+	hinic_ceq_unregister_cb(&func_to_io->ceqs, HINIC_CEQ_CMDQ);
+
 	cmdq_type = HINIC_CMDQ_SYNC;
 	for (; cmdq_type < HINIC_MAX_CMDQ_TYPES; cmdq_type++)
 		free_cmdq(&cmdqs->cmdq[cmdq_type]);
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h
index 1f57301..10b8c7b 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h
@@ -81,27 +81,44 @@
 
 /* EQ registers */
 #define HINIC_AEQ_MTT_OFF_BASE_ADDR                     0x200
+#define HINIC_CEQ_MTT_OFF_BASE_ADDR                     0x400
 
 #define HINIC_EQ_MTT_OFF_STRIDE                         0x40
 
 #define HINIC_CSR_AEQ_MTT_OFF(id)                       \
 	(HINIC_AEQ_MTT_OFF_BASE_ADDR + (id) * HINIC_EQ_MTT_OFF_STRIDE)
 
+#define HINIC_CSR_CEQ_MTT_OFF(id)                       \
+	(HINIC_CEQ_MTT_OFF_BASE_ADDR + (id) * HINIC_EQ_MTT_OFF_STRIDE)
+
 #define HINIC_CSR_EQ_PAGE_OFF_STRIDE                    8
 
 #define HINIC_CSR_AEQ_HI_PHYS_ADDR_REG(q_id, pg_num)    \
 	(HINIC_CSR_AEQ_MTT_OFF(q_id) + \
 	 (pg_num) * HINIC_CSR_EQ_PAGE_OFF_STRIDE)
 
+#define HINIC_CSR_CEQ_HI_PHYS_ADDR_REG(q_id, pg_num)    \
+	(HINIC_CSR_CEQ_MTT_OFF(q_id) +          \
+	 (pg_num) * HINIC_CSR_EQ_PAGE_OFF_STRIDE)
+
 #define HINIC_CSR_AEQ_LO_PHYS_ADDR_REG(q_id, pg_num)    \
 	(HINIC_CSR_AEQ_MTT_OFF(q_id) + \
 	 (pg_num) * HINIC_CSR_EQ_PAGE_OFF_STRIDE + 4)
 
+#define HINIC_CSR_CEQ_LO_PHYS_ADDR_REG(q_id, pg_num)    \
+	(HINIC_CSR_CEQ_MTT_OFF(q_id) +  \
+	 (pg_num) * HINIC_CSR_EQ_PAGE_OFF_STRIDE + 4)
+
 #define HINIC_AEQ_CTRL_0_ADDR_BASE                      0xE00
 #define HINIC_AEQ_CTRL_1_ADDR_BASE                      0xE04
 #define HINIC_AEQ_CONS_IDX_ADDR_BASE                    0xE08
 #define HINIC_AEQ_PROD_IDX_ADDR_BASE                    0xE0C
 
+#define HINIC_CEQ_CTRL_0_ADDR_BASE                      0x1000
+#define HINIC_CEQ_CTRL_1_ADDR_BASE                      0x1004
+#define HINIC_CEQ_CONS_IDX_ADDR_BASE                    0x1008
+#define HINIC_CEQ_PROD_IDX_ADDR_BASE                    0x100C
+
 #define HINIC_EQ_OFF_STRIDE                             0x80
 
 #define HINIC_CSR_AEQ_CTRL_0_ADDR(idx)                  \
@@ -116,4 +133,16 @@
 #define HINIC_CSR_AEQ_PROD_IDX_ADDR(idx)                \
 	(HINIC_AEQ_PROD_IDX_ADDR_BASE + (idx) * HINIC_EQ_OFF_STRIDE)
 
+#define HINIC_CSR_CEQ_CTRL_0_ADDR(idx)                  \
+	(HINIC_CEQ_CTRL_0_ADDR_BASE + (idx) * HINIC_EQ_OFF_STRIDE)
+
+#define HINIC_CSR_CEQ_CTRL_1_ADDR(idx)                  \
+	(HINIC_CEQ_CTRL_1_ADDR_BASE + (idx) * HINIC_EQ_OFF_STRIDE)
+
+#define HINIC_CSR_CEQ_CONS_IDX_ADDR(idx)                \
+	(HINIC_CEQ_CONS_IDX_ADDR_BASE + (idx) * HINIC_EQ_OFF_STRIDE)
+
+#define HINIC_CSR_CEQ_PROD_IDX_ADDR(idx)                \
+	(HINIC_CEQ_PROD_IDX_ADDR_BASE + (idx) * HINIC_EQ_OFF_STRIDE)
+
 #endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
index cf0e165..cb4c472 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
@@ -277,6 +277,7 @@ int hinic_hwdev_ifup(struct hinic_hwdev *hwdev)
 	struct hinic_cap *nic_cap = &hwdev->nic_cap;
 	struct hinic_hwif *hwif = hwdev->hwif;
 	int err, num_aeqs, num_ceqs, num_qps;
+	struct msix_entry *ceq_msix_entries;
 	struct msix_entry *sq_msix_entries;
 	struct msix_entry *rq_msix_entries;
 	struct pci_dev *pdev = hwif->pdev;
@@ -290,7 +291,11 @@ int hinic_hwdev_ifup(struct hinic_hwdev *hwdev)
 
 	num_aeqs = HINIC_HWIF_NUM_AEQS(hwif);
 	num_ceqs = HINIC_HWIF_NUM_CEQS(hwif);
-	err = hinic_io_init(func_to_io, hwif, nic_cap->max_qps, 0, NULL);
+
+	ceq_msix_entries = &hwdev->msix_entries[num_aeqs];
+
+	err = hinic_io_init(func_to_io, hwif, nic_cap->max_qps, num_ceqs,
+			    ceq_msix_entries);
 	if (err) {
 		dev_err(&pdev->dev, "Failed to init IO channel\n");
 		return err;
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.c
index a53d5b3..986ad93 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.c
@@ -37,14 +37,21 @@
 
 #define GET_EQ_NUM_ELEMS_IN_PG(eq, pg_size)     ((pg_size) / (eq)->elem_size)
 
-#define EQ_CONS_IDX_REG_ADDR(eq)        HINIC_CSR_AEQ_CONS_IDX_ADDR((eq)->q_id)
-#define EQ_PROD_IDX_REG_ADDR(eq)        HINIC_CSR_AEQ_PROD_IDX_ADDR((eq)->q_id)
+#define EQ_CONS_IDX_REG_ADDR(eq)        (((eq)->type == HINIC_AEQ) ? \
+			HINIC_CSR_AEQ_CONS_IDX_ADDR((eq)->q_id) : \
+			HINIC_CSR_CEQ_CONS_IDX_ADDR((eq)->q_id))
 
-#define EQ_HI_PHYS_ADDR_REG(eq, pg_num) \
-		HINIC_CSR_AEQ_HI_PHYS_ADDR_REG((eq)->q_id, pg_num)
+#define EQ_PROD_IDX_REG_ADDR(eq)        (((eq)->type == HINIC_AEQ) ? \
+			HINIC_CSR_AEQ_PROD_IDX_ADDR((eq)->q_id) : \
+			HINIC_CSR_CEQ_PROD_IDX_ADDR((eq)->q_id))
 
-#define EQ_LO_PHYS_ADDR_REG(eq, pg_num) \
-		HINIC_CSR_AEQ_LO_PHYS_ADDR_REG((eq)->q_id, pg_num)
+#define EQ_HI_PHYS_ADDR_REG(eq, pg_num) (((eq)->type == HINIC_AEQ) ? \
+			HINIC_CSR_AEQ_HI_PHYS_ADDR_REG((eq)->q_id, pg_num) : \
+			HINIC_CSR_CEQ_HI_PHYS_ADDR_REG((eq)->q_id, pg_num))
+
+#define EQ_LO_PHYS_ADDR_REG(eq, pg_num) (((eq)->type == HINIC_AEQ) ? \
+			HINIC_CSR_AEQ_LO_PHYS_ADDR_REG((eq)->q_id, pg_num) : \
+			HINIC_CSR_CEQ_LO_PHYS_ADDR_REG((eq)->q_id, pg_num))
 
 #define GET_EQ_ELEMENT(eq, idx)         \
 		((eq)->virt_addr[(idx) / (eq)->num_elem_in_pg] + \
@@ -53,8 +60,13 @@
 #define GET_AEQ_ELEM(eq, idx)           ((struct hinic_aeq_elem *) \
 					GET_EQ_ELEMENT(eq, idx))
 
+#define GET_CEQ_ELEM(eq, idx)           ((u32 *) \
+					 GET_EQ_ELEMENT(eq, idx))
+
 #define GET_CURR_AEQ_ELEM(eq)           GET_AEQ_ELEM(eq, (eq)->cons_idx)
 
+#define GET_CURR_CEQ_ELEM(eq)           GET_CEQ_ELEM(eq, (eq)->cons_idx)
+
 #define PAGE_IN_4K(page_size)           ((page_size) >> 12)
 #define EQ_SET_HW_PAGE_SIZE_VAL(eq)     (ilog2(PAGE_IN_4K((eq)->page_size)))
 
@@ -63,13 +75,29 @@
 
 #define EQ_MAX_PAGES                    8
 
+#define CEQE_TYPE_SHIFT                 23
+#define CEQE_TYPE_MASK                  0x7
+
+#define CEQE_TYPE(ceqe)                 (((ceqe) >> CEQE_TYPE_SHIFT) &  \
+					 CEQE_TYPE_MASK)
+
+#define CEQE_DATA_MASK                  0x3FFFFFF
+#define CEQE_DATA(ceqe)                 ((ceqe) & CEQE_DATA_MASK)
+
 #define aeq_to_aeqs(eq)                 \
 		container_of((eq) - (eq)->q_id, struct hinic_aeqs, aeq[0])
 
+#define ceq_to_ceqs(eq)                 \
+		container_of((eq) - (eq)->q_id, struct hinic_ceqs, ceq[0])
+
 #define work_to_aeq_work(work)          \
 		container_of(work, struct hinic_eq_work, work)
 
 #define DMA_ATTR_AEQ_DEFAULT            0
+#define DMA_ATTR_CEQ_DEFAULT            0
+
+/* No coalescence */
+#define THRESH_CEQ_DEFAULT              0
 
 enum eq_int_mode {
 	EQ_INT_MODE_ARMED,
@@ -118,6 +146,42 @@ void hinic_aeq_unregister_hw_cb(struct hinic_aeqs *aeqs,
 	hwe_cb->hwe_handler = NULL;
 }
 
+/**
+ * hinic_ceq_register_cb - register CEQ callback for specific event
+ * @ceqs: pointer to Completion eqs part of the chip
+ * @event: ceq event to register callback for it
+ * @handle: private data will be used by the callback
+ * @handler: callback function
+ **/
+void hinic_ceq_register_cb(struct hinic_ceqs *ceqs,
+			   enum hinic_ceq_type event, void *handle,
+			   void (*handler)(void *handle, u32 ceqe_data))
+{
+	struct hinic_ceq_cb *ceq_cb = &ceqs->ceq_cb[event];
+
+	ceq_cb->handler = handler;
+	ceq_cb->handle = handle;
+	ceq_cb->ceqe_state = HINIC_EQE_ENABLED;
+}
+
+/**
+ * hinic_ceq_unregister_cb - unregister the CEQ callback for specific event
+ * @ceqs: pointer to Completion eqs part of the chip
+ * @event: ceq event to unregister callback for it
+ **/
+void hinic_ceq_unregister_cb(struct hinic_ceqs *ceqs,
+			     enum hinic_ceq_type event)
+{
+	struct hinic_ceq_cb *ceq_cb = &ceqs->ceq_cb[event];
+
+	ceq_cb->ceqe_state &= ~HINIC_EQE_ENABLED;
+
+	while (ceq_cb->ceqe_state & HINIC_EQE_RUNNING)
+		schedule();
+
+	ceq_cb->handler = NULL;
+}
+
 static u8 eq_cons_idx_checksum_set(u32 val)
 {
 	u8 checksum = 0;
@@ -216,6 +280,70 @@ static void aeq_irq_handler(struct hinic_eq *eq)
 }
 
 /**
+ * ceq_event_handler - handler for the ceq events
+ * @ceqs: ceqs part of the chip
+ * @ceqe: ceq element that describes the event
+ **/
+static void ceq_event_handler(struct hinic_ceqs *ceqs, u32 ceqe)
+{
+	struct hinic_hwif *hwif = ceqs->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_ceq_cb *ceq_cb;
+	enum hinic_ceq_type event;
+	unsigned long eqe_state;
+
+	event = CEQE_TYPE(ceqe);
+	if (event >= HINIC_MAX_CEQ_EVENTS) {
+		dev_err(&pdev->dev, "Unknown CEQ event, event = %d\n", event);
+		return;
+	}
+
+	ceq_cb = &ceqs->ceq_cb[event];
+
+	eqe_state = cmpxchg(&ceq_cb->ceqe_state,
+			    HINIC_EQE_ENABLED,
+			    HINIC_EQE_ENABLED | HINIC_EQE_RUNNING);
+
+	if ((eqe_state == HINIC_EQE_ENABLED) && (ceq_cb->handler))
+		ceq_cb->handler(ceq_cb->handle, CEQE_DATA(ceqe));
+	else
+		dev_err(&pdev->dev, "Unhandled CEQ Event %d\n", event);
+
+	ceq_cb->ceqe_state &= ~HINIC_EQE_RUNNING;
+}
+
+/**
+ * ceq_irq_handler - handler for the CEQ event
+ * @eq: the Completion Event Queue that received the event
+ **/
+static void ceq_irq_handler(struct hinic_eq *eq)
+{
+	struct hinic_ceqs *ceqs = ceq_to_ceqs(eq);
+	u32 ceqe;
+	int i;
+
+	for (i = 0; i < eq->q_len; i++) {
+		ceqe = *(GET_CURR_CEQ_ELEM(eq));
+
+		/* Data in HW is in Big endian Format */
+		ceqe = be32_to_cpu(ceqe);
+
+		/* HW toggles the wrapped bit, when it adds eq element event */
+		if (HINIC_EQ_ELEM_DESC_GET(ceqe, WRAPPED) == eq->wrapped)
+			break;
+
+		ceq_event_handler(ceqs, ceqe);
+
+		eq->cons_idx++;
+
+		if (eq->cons_idx == eq->q_len) {
+			eq->cons_idx = 0;
+			eq->wrapped = !eq->wrapped;
+		}
+	}
+}
+
+/**
  * eq_irq_handler - handler for the EQ event
  * @data: the Event Queue that received the event
  **/
@@ -225,6 +353,8 @@ static void eq_irq_handler(void *data)
 
 	if (eq->type == HINIC_AEQ)
 		aeq_irq_handler(eq);
+	else if (eq->type == HINIC_CEQ)
+		ceq_irq_handler(eq);
 
 	eq_update_ci(eq);
 }
@@ -243,6 +373,17 @@ static void eq_irq_work(struct work_struct *work)
 }
 
 /**
+ * ceq_tasklet - the tasklet of the EQ that received the event
+ * @ceq_data: the eq
+ **/
+static void ceq_tasklet(unsigned long ceq_data)
+{
+	struct hinic_eq *ceq = (struct hinic_eq *)ceq_data;
+
+	eq_irq_handler(ceq);
+}
+
+/**
  * aeq_interrupt - aeq interrupt handler
  * @irq: irq number
  * @data: the Async Event Queue that collected the event
@@ -264,6 +405,24 @@ static irqreturn_t aeq_interrupt(int irq, void *data)
 	return IRQ_HANDLED;
 }
 
+/**
+ * ceq_interrupt - ceq interrupt handler
+ * @irq: irq number
+ * @data: the Completion Event Queue that collected the event
+ **/
+static irqreturn_t ceq_interrupt(int irq, void *data)
+{
+	struct hinic_eq *ceq = (struct hinic_eq *)data;
+	struct hinic_hwif *hwif = ceq->hwif;
+
+	/* clear resend timer cnt register */
+	hinic_msix_attr_cnt_clear(hwif, ceq->msix_entry.entry);
+
+	tasklet_schedule(&ceq->ceq_tasklet);
+
+	return IRQ_HANDLED;
+}
+
 void set_ctrl0(struct hinic_eq *eq)
 {
 	struct msix_entry *msix_entry = &eq->msix_entry;
@@ -291,6 +450,28 @@ void set_ctrl0(struct hinic_eq *eq)
 		val |= ctrl0;
 
 		hinic_hwif_write_reg(hwif, addr, val);
+	} else if (type == HINIC_CEQ) {
+		/* RMW Ctrl0 */
+		addr = HINIC_CSR_CEQ_CTRL_0_ADDR(eq->q_id);
+
+		val = hinic_hwif_read_reg(hwif, addr);
+
+		val = HINIC_CEQ_CTRL_0_CLEAR(val, INTR_IDX)     &
+		      HINIC_CEQ_CTRL_0_CLEAR(val, DMA_ATTR)     &
+		      HINIC_CEQ_CTRL_0_CLEAR(val, KICK_THRESH)  &
+		      HINIC_CEQ_CTRL_0_CLEAR(val, PCI_INTF_IDX) &
+		      HINIC_CEQ_CTRL_0_CLEAR(val, INTR_MODE);
+
+		ctrl0 = HINIC_CEQ_CTRL_0_SET(msix_entry->entry, INTR_IDX)     |
+			HINIC_CEQ_CTRL_0_SET(DMA_ATTR_CEQ_DEFAULT, DMA_ATTR)  |
+			HINIC_CEQ_CTRL_0_SET(THRESH_CEQ_DEFAULT, KICK_THRESH) |
+			HINIC_CEQ_CTRL_0_SET(HINIC_HWIF_PCI_INTF(hwif),
+					     PCI_INTF_IDX)                    |
+			HINIC_CEQ_CTRL_0_SET(EQ_INT_MODE_ARMED, INTR_MODE);
+
+		val |= ctrl0;
+
+		hinic_hwif_write_reg(hwif, addr, val);
 	}
 }
 
@@ -321,6 +502,23 @@ void set_ctrl1(struct hinic_eq *eq)
 		val |= ctrl1;
 
 		hinic_hwif_write_reg(hwif, addr, val);
+	} else if (type == HINIC_CEQ) {
+		/* RMW Ctrl1 */
+		addr = HINIC_CSR_CEQ_CTRL_1_ADDR(eq->q_id);
+
+		page_size_val = EQ_SET_HW_PAGE_SIZE_VAL(eq);
+
+		val = hinic_hwif_read_reg(hwif, addr);
+
+		val = HINIC_CEQ_CTRL_1_CLEAR(val, LEN) &
+		      HINIC_CEQ_CTRL_1_CLEAR(val, PAGE_SIZE);
+
+		ctrl1 = HINIC_CEQ_CTRL_1_SET(eq->q_len, LEN) |
+			HINIC_CEQ_CTRL_1_SET(page_size_val, PAGE_SIZE);
+
+		val |= ctrl1;
+
+		hinic_hwif_write_reg(hwif, addr, val);
 	}
 }
 
@@ -353,6 +551,24 @@ static void aeq_elements_init(struct hinic_eq *eq, u32 init_val)
 }
 
 /**
+ * ceq_elements_init - Initialize all the elements in the ceq
+ * @eq: the event queue
+ * @init_val: value to init with it the elements
+ **/
+static void ceq_elements_init(struct hinic_eq *eq, u32 init_val)
+{
+	u32 *ceqe;
+	int i;
+
+	for (i = 0; i < eq->q_len; i++) {
+		ceqe = GET_CEQ_ELEM(eq, i);
+		*(ceqe) = cpu_to_be32(init_val);
+	}
+
+	wmb();  /* Write the initilzation values */
+}
+
+/**
  * alloc_eq_pages - allocate the pages for the queue
  * @eq: the event queue
  *
@@ -403,6 +619,8 @@ static int alloc_eq_pages(struct hinic_eq *eq)
 
 	if (eq->type == HINIC_AEQ)
 		aeq_elements_init(eq, init_val);
+	else if (eq->type == HINIC_CEQ)
+		ceq_elements_init(eq, init_val);
 
 	return 0;
 
@@ -472,6 +690,8 @@ static int init_eq(struct hinic_eq *eq, struct hinic_hwif *hwif,
 
 	if (type == HINIC_AEQ) {
 		eq->elem_size = HINIC_AEQE_SIZE;
+	} else if (type == HINIC_CEQ) {
+		eq->elem_size = HINIC_CEQE_SIZE;
 	} else {
 		dev_err(&pdev->dev, "Invalid EQ type\n");
 		return -EINVAL;
@@ -505,6 +725,9 @@ static int init_eq(struct hinic_eq *eq, struct hinic_hwif *hwif,
 		struct hinic_eq_work *aeq_work = &eq->aeq_work;
 
 		INIT_WORK(&aeq_work->work, eq_irq_work);
+	} else if (type == HINIC_CEQ) {
+		tasklet_init(&eq->ceq_tasklet, ceq_tasklet,
+			     (unsigned long)eq);
 	}
 
 	/* set the attributes of the msix entry */
@@ -518,6 +741,9 @@ static int init_eq(struct hinic_eq *eq, struct hinic_hwif *hwif,
 	if (type == HINIC_AEQ)
 		err = request_irq(entry.vector, aeq_interrupt, 0,
 				  "hinic_aeq", eq);
+	else if (type == HINIC_CEQ)
+		err = request_irq(entry.vector, ceq_interrupt, 0,
+				  "hinic_ceq", eq);
 
 	if (err) {
 		dev_err(&pdev->dev, "Failed to request irq for the EQ\n");
@@ -545,6 +771,8 @@ static void remove_eq(struct hinic_eq *eq)
 		struct hinic_eq_work *aeq_work = &eq->aeq_work;
 
 		cancel_work_sync(&aeq_work->work);
+	} else if (eq->type == HINIC_CEQ) {
+		tasklet_kill(&eq->ceq_tasklet);
 	}
 
 	free_eq_pages(eq);
@@ -607,3 +835,54 @@ void hinic_aeqs_free(struct hinic_aeqs *aeqs)
 
 	destroy_workqueue(aeqs->workq);
 }
+
+/**
+ * hinic_ceqs_init - init all the ceqs
+ * @ceqs: ceqs part of the chip
+ * @hwif: the hardware interface of a pci function device
+ * @num_ceqs: number of CEQs
+ * @q_len: number of EQ elements
+ * @page_size: the page size of the event queue
+ * @msix_entries: msix entries associated with the event queues
+ *
+ * Return 0 - Success, Negative - Failure
+ **/
+int hinic_ceqs_init(struct hinic_ceqs *ceqs, struct hinic_hwif *hwif,
+		    int num_ceqs, u32 q_len, u32 page_size,
+		    struct msix_entry *msix_entries)
+{
+	struct pci_dev *pdev = hwif->pdev;
+	int i, q_id, err;
+
+	ceqs->hwif = hwif;
+	ceqs->num_ceqs = num_ceqs;
+
+	for (q_id = 0; q_id < num_ceqs; q_id++) {
+		err = init_eq(&ceqs->ceq[q_id], hwif, HINIC_CEQ, q_id, q_len,
+			      page_size, msix_entries[q_id]);
+		if (err) {
+			dev_err(&pdev->dev, "Failed to init ceq %d\n", q_id);
+			goto err_init_ceq;
+		}
+	}
+
+	return 0;
+
+err_init_ceq:
+	for (i = 0; i < q_id; i++)
+		remove_eq(&ceqs->ceq[i]);
+
+	return err;
+}
+
+/**
+ * hinic_ceqs_free - free all the ceqs
+ * @ceqs: ceqs part of the chip
+ **/
+void hinic_ceqs_free(struct hinic_ceqs *ceqs)
+{
+	int q_id;
+
+	for (q_id = 0; q_id < ceqs->num_ceqs; q_id++)
+		remove_eq(&ceqs->ceq[q_id]);
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.h
index ca584c0..ecb9c2b 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.h
@@ -21,6 +21,7 @@
 #include <linux/pci.h>
 #include <linux/sizes.h>
 #include <linux/bitops.h>
+#include <linux/interrupt.h>
 
 #include "hinic_hw_if.h"
 
@@ -58,6 +59,40 @@
 			((val) & (~(HINIC_AEQ_CTRL_1_##member##_MASK \
 			 << HINIC_AEQ_CTRL_1_##member##_SHIFT)))
 
+#define HINIC_CEQ_CTRL_0_INTR_IDX_SHIFT         0
+#define HINIC_CEQ_CTRL_0_DMA_ATTR_SHIFT         12
+#define HINIC_CEQ_CTRL_0_KICK_THRESH_SHIFT      20
+#define HINIC_CEQ_CTRL_0_PCI_INTF_IDX_SHIFT     24
+#define HINIC_CEQ_CTRL_0_INTR_MODE_SHIFT        31
+
+#define HINIC_CEQ_CTRL_0_INTR_IDX_MASK          0x3FF
+#define HINIC_CEQ_CTRL_0_DMA_ATTR_MASK          0x3F
+#define HINIC_CEQ_CTRL_0_KICK_THRESH_MASK       0xF
+#define HINIC_CEQ_CTRL_0_PCI_INTF_IDX_MASK      0x3
+#define HINIC_CEQ_CTRL_0_INTR_MODE_MASK         0x1
+
+#define HINIC_CEQ_CTRL_0_SET(val, member)       \
+			(((u32)(val) & HINIC_CEQ_CTRL_0_##member##_MASK) << \
+			 HINIC_CEQ_CTRL_0_##member##_SHIFT)
+
+#define HINIC_CEQ_CTRL_0_CLEAR(val, member)     \
+			((val) & (~(HINIC_CEQ_CTRL_0_##member##_MASK \
+			 << HINIC_CEQ_CTRL_0_##member##_SHIFT)))
+
+#define HINIC_CEQ_CTRL_1_LEN_SHIFT              0
+#define HINIC_CEQ_CTRL_1_PAGE_SIZE_SHIFT        28
+
+#define HINIC_CEQ_CTRL_1_LEN_MASK               0x1FFFFF
+#define HINIC_CEQ_CTRL_1_PAGE_SIZE_MASK         0xF
+
+#define HINIC_CEQ_CTRL_1_SET(val, member)       \
+			(((u32)(val) & HINIC_CEQ_CTRL_1_##member##_MASK) << \
+			 HINIC_CEQ_CTRL_1_##member##_SHIFT)
+
+#define HINIC_CEQ_CTRL_1_CLEAR(val, member)     \
+			((val) & (~(HINIC_CEQ_CTRL_1_##member##_MASK \
+			 << HINIC_CEQ_CTRL_1_##member##_SHIFT)))
+
 #define HINIC_EQ_ELEM_DESC_TYPE_SHIFT           0
 #define HINIC_EQ_ELEM_DESC_SRC_SHIFT            7
 #define HINIC_EQ_ELEM_DESC_SIZE_SHIFT           8
@@ -95,14 +130,17 @@
 			 << HINIC_EQ_CI_##member##_SHIFT)))
 
 #define HINIC_MAX_AEQS                  4
+#define HINIC_MAX_CEQS                  32
 
 #define HINIC_AEQE_SIZE                 64
+#define HINIC_CEQE_SIZE                 4
 
 #define HINIC_AEQE_DESC_SIZE            4
 #define HINIC_AEQE_DATA_SIZE            \
 			(HINIC_AEQE_SIZE - HINIC_AEQE_DESC_SIZE)
 
 #define HINIC_DEFAULT_AEQ_LEN           64
+#define HINIC_DEFAULT_CEQ_LEN           1024
 
 #define HINIC_EQ_PAGE_SIZE              SZ_4K
 
@@ -110,6 +148,7 @@
 
 enum hinic_eq_type {
 	HINIC_AEQ,
+	HINIC_CEQ,
 };
 
 enum hinic_aeq_type {
@@ -118,6 +157,12 @@ enum hinic_aeq_type {
 	HINIC_MAX_AEQ_EVENTS,
 };
 
+enum hinic_ceq_type {
+	HINIC_CEQ_CMDQ = 3,
+
+	HINIC_MAX_CEQ_EVENTS,
+};
+
 enum hinic_eqe_state {
 	HINIC_EQE_ENABLED = BIT(0),
 	HINIC_EQE_RUNNING = BIT(1),
@@ -154,6 +199,8 @@ struct hinic_eq {
 	void                    **virt_addr;
 
 	struct hinic_eq_work    aeq_work;
+
+	struct tasklet_struct   ceq_tasklet;
 };
 
 struct hinic_hw_event_cb {
@@ -173,6 +220,21 @@ struct hinic_aeqs {
 	struct workqueue_struct *workq;
 };
 
+struct hinic_ceq_cb {
+	void    (*handler)(void *handle, u32 ceqe_data);
+	void                    *handle;
+	enum hinic_eqe_state    ceqe_state;
+};
+
+struct hinic_ceqs {
+	struct hinic_hwif       *hwif;
+
+	struct hinic_eq         ceq[HINIC_MAX_CEQS];
+	int                     num_ceqs;
+
+	struct hinic_ceq_cb     ceq_cb[HINIC_MAX_CEQ_EVENTS];
+};
+
 void hinic_aeq_register_hw_cb(struct hinic_aeqs *aeqs,
 			      enum hinic_aeq_type event, void *handle,
 			      void (*hwe_handler)(void *handle, void *data,
@@ -181,10 +243,23 @@ void hinic_aeq_register_hw_cb(struct hinic_aeqs *aeqs,
 void hinic_aeq_unregister_hw_cb(struct hinic_aeqs *aeqs,
 				enum hinic_aeq_type event);
 
+void hinic_ceq_register_cb(struct hinic_ceqs *ceqs,
+			   enum hinic_ceq_type event, void *handle,
+			   void (*ceq_cb)(void *handle, u32 ceqe_data));
+
+void hinic_ceq_unregister_cb(struct hinic_ceqs *ceqs,
+			     enum hinic_ceq_type event);
+
 int hinic_aeqs_init(struct hinic_aeqs *aeqs, struct hinic_hwif *hwif,
 		    int num_aeqs, u32 q_len, u32 page_size,
 		    struct msix_entry *msix_entries);
 
 void hinic_aeqs_free(struct hinic_aeqs *aeqs);
 
+int hinic_ceqs_init(struct hinic_ceqs *ceqs, struct hinic_hwif *hwif,
+		    int num_ceqs, u32 q_len, u32 page_size,
+		    struct msix_entry *msix_entries);
+
+void hinic_ceqs_free(struct hinic_ceqs *ceqs);
+
 #endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c
index bb4b93f..8e58976 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c
@@ -25,6 +25,7 @@
 #include <linux/err.h>
 
 #include "hinic_hw_if.h"
+#include "hinic_hw_eqs.h"
 #include "hinic_hw_wqe.h"
 #include "hinic_hw_wq.h"
 #include "hinic_hw_cmdq.h"
@@ -455,10 +456,18 @@ int hinic_io_init(struct hinic_func_to_io *func_to_io,
 	func_to_io->qps = NULL;
 	func_to_io->max_qps = max_qps;
 
+	err = hinic_ceqs_init(&func_to_io->ceqs, hwif, num_ceqs,
+			      HINIC_DEFAULT_CEQ_LEN, HINIC_EQ_PAGE_SIZE,
+			      ceq_msix_entries);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to init CEQs\n");
+		return err;
+	}
+
 	err = hinic_wqs_alloc(&func_to_io->wqs, 2 * max_qps, hwif);
 	if (err) {
 		dev_err(&pdev->dev, "Failed to allocate WQS for IO\n");
-		return err;
+		goto err_wqs_alloc;
 	}
 
 	func_to_io->db_base = pci_ioremap_bar(pdev, HINIC_PCI_DB_BAR);
@@ -499,6 +508,9 @@ int hinic_io_init(struct hinic_func_to_io *func_to_io,
 
 err_db_ioremap:
 	hinic_wqs_free(&func_to_io->wqs);
+
+err_wqs_alloc:
+	hinic_ceqs_free(&func_to_io->ceqs);
 	return err;
 }
 
@@ -517,4 +529,5 @@ void hinic_io_free(struct hinic_func_to_io *func_to_io)
 
 	iounmap(func_to_io->db_base);
 	hinic_wqs_free(&func_to_io->wqs);
+	hinic_ceqs_free(&func_to_io->ceqs);
 }
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
index 60d77b34..cfc21ba 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
@@ -22,6 +22,7 @@
 #include <linux/sizes.h>
 
 #include "hinic_hw_if.h"
+#include "hinic_hw_eqs.h"
 #include "hinic_hw_wq.h"
 #include "hinic_hw_cmdq.h"
 #include "hinic_hw_qp.h"
@@ -46,6 +47,8 @@ struct hinic_free_db_area {
 struct hinic_func_to_io {
 	struct hinic_hwif       *hwif;
 
+	struct hinic_ceqs       ceqs;
+
 	struct hinic_wqs        wqs;
 
 	struct hinic_wq         *sq_wq;
-- 
1.9.1

^ permalink raw reply related

* [PATCH V7 net-next 14/22] net-next/hinic: Initialize cmdq
From: Aviad Krawczyk @ 2017-08-17 19:04 UTC (permalink / raw)
  To: davem
  Cc: linux-kernel, netdev, bc.y, victor.gissin, aviad.krawczyk,
	zhaochen6, tony.qu
In-Reply-To: <cover.1502996306.git.aviad.krawczyk@huawei.com>

Create the work queues for cmdq and update the nic about the
work queue contexts. cmdq commands are used for updating the nic about
the qp contexts.

Signed-off-by: Aviad Krawczyk <aviad.krawczyk@huawei.com>
Signed-off-by: Zhao Chen <zhaochen6@huawei.com>
---
 drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c | 282 +++++++++++++++++++++-
 drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h |  53 ++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.h  |   2 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.h |   5 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c   | 156 ++++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h   |   8 +
 6 files changed, 500 insertions(+), 6 deletions(-)

diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c
index 2fd3924..0dccbe6 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c
@@ -13,11 +13,49 @@
  *
  */
 
+#include <linux/kernel.h>
 #include <linux/types.h>
 #include <linux/errno.h>
+#include <linux/pci.h>
+#include <linux/device.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+#include <linux/spinlock.h>
+#include <linux/sizes.h>
+#include <linux/atomic.h>
+#include <linux/log2.h>
+#include <asm/byteorder.h>
 
 #include "hinic_hw_if.h"
+#include "hinic_hw_mgmt.h"
+#include "hinic_hw_wq.h"
 #include "hinic_hw_cmdq.h"
+#include "hinic_hw_io.h"
+#include "hinic_hw_dev.h"
+
+#define CMDQ_DB_OFF                     SZ_2K
+
+#define CMDQ_WQEBB_SIZE                 64
+#define CMDQ_DEPTH                      SZ_4K
+
+#define CMDQ_WQ_PAGE_SIZE               SZ_4K
+
+#define WQE_LCMD_SIZE                   64
+#define WQE_SCMD_SIZE                   64
+
+#define CMDQ_PFN(addr, page_size)       ((addr) >> (ilog2(page_size)))
+
+#define cmdq_to_cmdqs(cmdq)     container_of((cmdq) - (cmdq)->cmdq_type, \
+					     struct hinic_cmdqs, cmdq[0])
+
+#define cmdqs_to_func_to_io(cmdqs)      container_of(cmdqs, \
+						     struct hinic_func_to_io, \
+						     cmdqs)
+
+enum cmdq_wqe_type {
+	WQE_LCMD_TYPE = 0,
+	WQE_SCMD_TYPE = 1,
+};
 
 /**
  * hinic_alloc_cmdq_buf - alloc buffer for sending command
@@ -29,8 +67,17 @@
 int hinic_alloc_cmdq_buf(struct hinic_cmdqs *cmdqs,
 			 struct hinic_cmdq_buf *cmdq_buf)
 {
-	/* should be implemented */
-	return -ENOMEM;
+	struct hinic_hwif *hwif = cmdqs->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+
+	cmdq_buf->buf = pci_pool_alloc(cmdqs->cmdq_buf_pool, GFP_KERNEL,
+				       &cmdq_buf->dma_addr);
+	if (!cmdq_buf->buf) {
+		dev_err(&pdev->dev, "Failed to allocate cmd from the pool\n");
+		return -ENOMEM;
+	}
+
+	return 0;
 }
 
 /**
@@ -41,7 +88,7 @@ int hinic_alloc_cmdq_buf(struct hinic_cmdqs *cmdqs,
 void hinic_free_cmdq_buf(struct hinic_cmdqs *cmdqs,
 			 struct hinic_cmdq_buf *cmdq_buf)
 {
-	/* should be implemented */
+	pci_pool_free(cmdqs->cmdq_buf_pool, cmdq_buf->buf, cmdq_buf->dma_addr);
 }
 
 /**
@@ -63,6 +110,169 @@ int hinic_cmdq_direct_resp(struct hinic_cmdqs *cmdqs,
 }
 
 /**
+ * cmdq_init_queue_ctxt - init the queue ctxt of a cmdq
+ * @cmdq_ctxt: cmdq ctxt to initialize
+ * @cmdq: the cmdq
+ * @cmdq_pages: the memory of the queue
+ **/
+static void cmdq_init_queue_ctxt(struct hinic_cmdq_ctxt *cmdq_ctxt,
+				 struct hinic_cmdq *cmdq,
+				 struct hinic_cmdq_pages *cmdq_pages)
+{
+	struct hinic_cmdq_ctxt_info *ctxt_info = &cmdq_ctxt->ctxt_info;
+	u64 wq_first_page_paddr, cmdq_first_block_paddr, pfn;
+	struct hinic_cmdqs *cmdqs = cmdq_to_cmdqs(cmdq);
+	struct hinic_wq *wq = cmdq->wq;
+
+	/* The data in the HW is in Big Endian Format */
+	wq_first_page_paddr = be64_to_cpu(*wq->block_vaddr);
+
+	pfn = CMDQ_PFN(wq_first_page_paddr, wq->wq_page_size);
+
+	ctxt_info->curr_wqe_page_pfn =
+		HINIC_CMDQ_CTXT_PAGE_INFO_SET(pfn, CURR_WQE_PAGE_PFN)   |
+		HINIC_CMDQ_CTXT_PAGE_INFO_SET(HINIC_CEQ_ID_CMDQ, EQ_ID) |
+		HINIC_CMDQ_CTXT_PAGE_INFO_SET(1, CEQ_ARM)               |
+		HINIC_CMDQ_CTXT_PAGE_INFO_SET(1, CEQ_EN)                |
+		HINIC_CMDQ_CTXT_PAGE_INFO_SET(cmdq->wrapped, WRAPPED);
+
+	/* block PFN - Read Modify Write */
+	cmdq_first_block_paddr = cmdq_pages->page_paddr;
+
+	pfn = CMDQ_PFN(cmdq_first_block_paddr, wq->wq_page_size);
+
+	ctxt_info->wq_block_pfn =
+		HINIC_CMDQ_CTXT_BLOCK_INFO_SET(pfn, WQ_BLOCK_PFN) |
+		HINIC_CMDQ_CTXT_BLOCK_INFO_SET(atomic_read(&wq->cons_idx), CI);
+
+	cmdq_ctxt->func_idx = HINIC_HWIF_FUNC_IDX(cmdqs->hwif);
+	cmdq_ctxt->cmdq_type  = cmdq->cmdq_type;
+}
+
+/**
+ * init_cmdq - initialize cmdq
+ * @cmdq: the cmdq
+ * @wq: the wq attaced to the cmdq
+ * @q_type: the cmdq type of the cmdq
+ * @db_area: doorbell area for the cmdq
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int init_cmdq(struct hinic_cmdq *cmdq, struct hinic_wq *wq,
+		     enum hinic_cmdq_type q_type, void __iomem *db_area)
+{
+	int err;
+
+	cmdq->wq = wq;
+	cmdq->cmdq_type = q_type;
+	cmdq->wrapped = 1;
+
+	spin_lock_init(&cmdq->cmdq_lock);
+
+	cmdq->done = vzalloc(wq->q_depth * sizeof(*cmdq->done));
+	if (!cmdq->done)
+		return -ENOMEM;
+
+	cmdq->errcode = vzalloc(wq->q_depth * sizeof(*cmdq->errcode));
+	if (!cmdq->errcode) {
+		err = -ENOMEM;
+		goto err_errcode;
+	}
+
+	cmdq->db_base = db_area + CMDQ_DB_OFF;
+	return 0;
+
+err_errcode:
+	vfree(cmdq->done);
+	return err;
+}
+
+/**
+ * free_cmdq - Free cmdq
+ * @cmdq: the cmdq to free
+ **/
+static void free_cmdq(struct hinic_cmdq *cmdq)
+{
+	vfree(cmdq->errcode);
+	vfree(cmdq->done);
+}
+
+/**
+ * init_cmdqs_ctxt - write the cmdq ctxt to HW after init all cmdq
+ * @hwdev: the NIC HW device
+ * @cmdqs: cmdqs to write the ctxts for
+ * &db_area: db_area for all the cmdqs
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int init_cmdqs_ctxt(struct hinic_hwdev *hwdev,
+			   struct hinic_cmdqs *cmdqs, void __iomem **db_area)
+{
+	struct hinic_hwif *hwif = hwdev->hwif;
+	enum hinic_cmdq_type type, cmdq_type;
+	struct hinic_cmdq_ctxt *cmdq_ctxts;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_pfhwdev *pfhwdev;
+	size_t cmdq_ctxts_size;
+	int err;
+
+	if (!HINIC_IS_PF(hwif) && !HINIC_IS_PPF(hwif)) {
+		dev_err(&pdev->dev, "Unsupported PCI function type\n");
+		return -EINVAL;
+	}
+
+	cmdq_ctxts_size = HINIC_MAX_CMDQ_TYPES * sizeof(*cmdq_ctxts);
+	cmdq_ctxts = devm_kzalloc(&pdev->dev, cmdq_ctxts_size, GFP_KERNEL);
+	if (!cmdq_ctxts)
+		return -ENOMEM;
+
+	pfhwdev = container_of(hwdev, struct hinic_pfhwdev, hwdev);
+
+	cmdq_type = HINIC_CMDQ_SYNC;
+	for (; cmdq_type < HINIC_MAX_CMDQ_TYPES; cmdq_type++) {
+		err = init_cmdq(&cmdqs->cmdq[cmdq_type],
+				&cmdqs->saved_wqs[cmdq_type], cmdq_type,
+				db_area[cmdq_type]);
+		if (err) {
+			dev_err(&pdev->dev, "Failed to initialize cmdq\n");
+			goto err_init_cmdq;
+		}
+
+		cmdq_init_queue_ctxt(&cmdq_ctxts[cmdq_type],
+				     &cmdqs->cmdq[cmdq_type],
+				     &cmdqs->cmdq_pages);
+	}
+
+	/* Write the CMDQ ctxts */
+	cmdq_type = HINIC_CMDQ_SYNC;
+	for (; cmdq_type < HINIC_MAX_CMDQ_TYPES; cmdq_type++) {
+		err = hinic_msg_to_mgmt(&pfhwdev->pf_to_mgmt, HINIC_MOD_COMM,
+					HINIC_COMM_CMD_CMDQ_CTXT_SET,
+					&cmdq_ctxts[cmdq_type],
+					sizeof(cmdq_ctxts[cmdq_type]),
+					NULL, NULL, HINIC_MGMT_MSG_SYNC);
+		if (err) {
+			dev_err(&pdev->dev, "Failed to set CMDQ CTXT type = %d\n",
+				cmdq_type);
+			goto err_write_cmdq_ctxt;
+		}
+	}
+
+	devm_kfree(&pdev->dev, cmdq_ctxts);
+	return 0;
+
+err_write_cmdq_ctxt:
+	cmdq_type = HINIC_MAX_CMDQ_TYPES;
+
+err_init_cmdq:
+	for (type = HINIC_CMDQ_SYNC; type < cmdq_type; type++)
+		free_cmdq(&cmdqs->cmdq[type]);
+
+	devm_kfree(&pdev->dev, cmdq_ctxts);
+	return err;
+}
+
+/**
  * hinic_init_cmdqs - init all cmdqs
  * @cmdqs: cmdqs to init
  * @hwif: HW interface for accessing cmdqs
@@ -73,8 +283,55 @@ int hinic_cmdq_direct_resp(struct hinic_cmdqs *cmdqs,
 int hinic_init_cmdqs(struct hinic_cmdqs *cmdqs, struct hinic_hwif *hwif,
 		     void __iomem **db_area)
 {
-	/* should be implemented */
-	return -EINVAL;
+	struct hinic_func_to_io *func_to_io = cmdqs_to_func_to_io(cmdqs);
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_hwdev *hwdev;
+	size_t saved_wqs_size;
+	u16 max_wqe_size;
+	int err;
+
+	cmdqs->hwif = hwif;
+	cmdqs->cmdq_buf_pool = pci_pool_create("hinic_cmdq", pdev,
+					       HINIC_CMDQ_BUF_SIZE,
+					       HINIC_CMDQ_BUF_SIZE, 0);
+	if (!cmdqs->cmdq_buf_pool)
+		return -ENOMEM;
+
+	saved_wqs_size = HINIC_MAX_CMDQ_TYPES * sizeof(struct hinic_wq);
+	cmdqs->saved_wqs = devm_kzalloc(&pdev->dev, saved_wqs_size, GFP_KERNEL);
+	if (!cmdqs->saved_wqs) {
+		err = -ENOMEM;
+		goto err_saved_wqs;
+	}
+
+	max_wqe_size = WQE_LCMD_SIZE;
+	err = hinic_wqs_cmdq_alloc(&cmdqs->cmdq_pages, cmdqs->saved_wqs, hwif,
+				   HINIC_MAX_CMDQ_TYPES, CMDQ_WQEBB_SIZE,
+				   CMDQ_WQ_PAGE_SIZE, CMDQ_DEPTH, max_wqe_size);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to allocate CMDQ wqs\n");
+		goto err_cmdq_wqs;
+	}
+
+	hwdev = container_of(func_to_io, struct hinic_hwdev, func_to_io);
+	err = init_cmdqs_ctxt(hwdev, cmdqs, db_area);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to write cmdq ctxt\n");
+		goto err_cmdq_ctxt;
+	}
+
+	return 0;
+
+err_cmdq_ctxt:
+	hinic_wqs_cmdq_free(&cmdqs->cmdq_pages, cmdqs->saved_wqs,
+			    HINIC_MAX_CMDQ_TYPES);
+
+err_cmdq_wqs:
+	devm_kfree(&pdev->dev, cmdqs->saved_wqs);
+
+err_saved_wqs:
+	pci_pool_destroy(cmdqs->cmdq_buf_pool);
+	return err;
 }
 
 /**
@@ -83,5 +340,18 @@ int hinic_init_cmdqs(struct hinic_cmdqs *cmdqs, struct hinic_hwif *hwif,
  **/
 void hinic_free_cmdqs(struct hinic_cmdqs *cmdqs)
 {
-	/* should be implemented */
+	struct hinic_hwif *hwif = cmdqs->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	enum hinic_cmdq_type cmdq_type;
+
+	cmdq_type = HINIC_CMDQ_SYNC;
+	for (; cmdq_type < HINIC_MAX_CMDQ_TYPES; cmdq_type++)
+		free_cmdq(&cmdqs->cmdq[cmdq_type]);
+
+	hinic_wqs_cmdq_free(&cmdqs->cmdq_pages, cmdqs->saved_wqs,
+			    HINIC_MAX_CMDQ_TYPES);
+
+	devm_kfree(&pdev->dev, cmdqs->saved_wqs);
+
+	pci_pool_destroy(cmdqs->cmdq_buf_pool);
 }
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h
index c9e97ca..5ec59f1 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h
@@ -24,6 +24,40 @@
 #include "hinic_hw_if.h"
 #include "hinic_hw_wq.h"
 
+#define HINIC_CMDQ_CTXT_CURR_WQE_PAGE_PFN_SHIFT         0
+#define HINIC_CMDQ_CTXT_EQ_ID_SHIFT                     56
+#define HINIC_CMDQ_CTXT_CEQ_ARM_SHIFT                   61
+#define HINIC_CMDQ_CTXT_CEQ_EN_SHIFT                    62
+#define HINIC_CMDQ_CTXT_WRAPPED_SHIFT                   63
+
+#define HINIC_CMDQ_CTXT_CURR_WQE_PAGE_PFN_MASK          0xFFFFFFFFFFFFF
+#define HINIC_CMDQ_CTXT_EQ_ID_MASK                      0x1F
+#define HINIC_CMDQ_CTXT_CEQ_ARM_MASK                    0x1
+#define HINIC_CMDQ_CTXT_CEQ_EN_MASK                     0x1
+#define HINIC_CMDQ_CTXT_WRAPPED_MASK                    0x1
+
+#define HINIC_CMDQ_CTXT_PAGE_INFO_SET(val, member)      \
+			(((u64)(val) & HINIC_CMDQ_CTXT_##member##_MASK) \
+			 << HINIC_CMDQ_CTXT_##member##_SHIFT)
+
+#define HINIC_CMDQ_CTXT_PAGE_INFO_CLEAR(val, member)    \
+			((val) & (~((u64)HINIC_CMDQ_CTXT_##member##_MASK \
+			 << HINIC_CMDQ_CTXT_##member##_SHIFT)))
+
+#define HINIC_CMDQ_CTXT_WQ_BLOCK_PFN_SHIFT              0
+#define HINIC_CMDQ_CTXT_CI_SHIFT                        52
+
+#define HINIC_CMDQ_CTXT_WQ_BLOCK_PFN_MASK               0xFFFFFFFFFFFFF
+#define HINIC_CMDQ_CTXT_CI_MASK                         0xFFF
+
+#define HINIC_CMDQ_CTXT_BLOCK_INFO_SET(val, member)     \
+			(((u64)(val) & HINIC_CMDQ_CTXT_##member##_MASK) \
+			 << HINIC_CMDQ_CTXT_##member##_SHIFT)
+
+#define HINIC_CMDQ_CTXT_BLOCK_INFO_CLEAR(val, member)   \
+			((val) & (~((u64)HINIC_CMDQ_CTXT_##member##_MASK \
+			 << HINIC_CMDQ_CTXT_##member##_SHIFT)))
+
 #define HINIC_CMDQ_BUF_SIZE             2048
 
 enum hinic_cmdq_type {
@@ -38,6 +72,25 @@ struct hinic_cmdq_buf {
 	size_t          size;
 };
 
+struct hinic_cmdq_ctxt_info {
+	u64     curr_wqe_page_pfn;
+	u64     wq_block_pfn;
+};
+
+struct hinic_cmdq_ctxt {
+	u8      status;
+	u8      version;
+	u8      rsvd0[6];
+
+	u16     func_idx;
+	u8      cmdq_type;
+	u8      rsvd1[1];
+
+	u8      rsvd2[4];
+
+	struct hinic_cmdq_ctxt_info ctxt_info;
+};
+
 struct hinic_cmdq {
 	struct hinic_wq         *wq;
 
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.h
index 7f50b2f..ca584c0 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.h
@@ -106,6 +106,8 @@
 
 #define HINIC_EQ_PAGE_SIZE              SZ_4K
 
+#define HINIC_CEQ_ID_CMDQ               0
+
 enum hinic_eq_type {
 	HINIC_AEQ,
 };
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.h
index 8021406..90116c2 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.h
@@ -68,6 +68,11 @@ enum hinic_cfg_cmd {
 	HINIC_CFG_NIC_CAP = 0,
 };
 
+enum hinic_comm_cmd {
+	HINIC_COMM_CMD_CMDQ_CTXT_SET    = 0x10,
+	HINIC_COMM_CMD_CMDQ_CTXT_GET    = 0x11,
+};
+
 enum hinic_mgmt_cb_state {
 	HINIC_MGMT_CB_ENABLED = BIT(0),
 	HINIC_MGMT_CB_RUNNING = BIT(1),
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c
index 88d370e..5a3689a 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c
@@ -27,6 +27,7 @@
 
 #include "hinic_hw_if.h"
 #include "hinic_hw_wq.h"
+#include "hinic_hw_cmdq.h"
 
 #define WQS_BLOCKS_PER_PAGE             4
 
@@ -42,6 +43,11 @@
 #define WQ_PAGE_ADDR_SIZE               sizeof(u64)
 #define WQ_MAX_PAGES                    (WQ_BLOCK_SIZE / WQ_PAGE_ADDR_SIZE)
 
+#define CMDQ_BLOCK_SIZE                 512
+#define CMDQ_PAGE_SIZE                  4096
+
+#define CMDQ_WQ_MAX_PAGES               (CMDQ_BLOCK_SIZE / WQ_PAGE_ADDR_SIZE)
+
 #define WQ_BASE_VADDR(wqs, wq)          \
 			((void *)((wqs)->page_vaddr[(wq)->page_idx]) \
 				+ (wq)->block_idx * WQ_BLOCK_SIZE)
@@ -54,6 +60,18 @@
 			((void *)((wqs)->shadow_page_vaddr[(wq)->page_idx]) \
 				+ (wq)->block_idx * WQ_BLOCK_SIZE)
 
+#define CMDQ_BASE_VADDR(cmdq_pages, wq) \
+			((void *)((cmdq_pages)->page_vaddr) \
+				+ (wq)->block_idx * CMDQ_BLOCK_SIZE)
+
+#define CMDQ_BASE_PADDR(cmdq_pages, wq) \
+			((cmdq_pages)->page_paddr \
+				+ (wq)->block_idx * CMDQ_BLOCK_SIZE)
+
+#define CMDQ_BASE_ADDR(cmdq_pages, wq)  \
+			((void *)((cmdq_pages)->shadow_page_vaddr) \
+				+ (wq)->block_idx * CMDQ_BLOCK_SIZE)
+
 /**
  * queue_alloc_page - allocate page for Queue
  * @hwif: HW interface for allocating DMA
@@ -122,6 +140,37 @@ static void wqs_free_page(struct hinic_wqs *wqs, int page_idx)
 	vfree(wqs->shadow_page_vaddr[page_idx]);
 }
 
+/**
+ * cmdq_allocate_page - allocate page for cmdq
+ * @cmdq_pages: the pages of the cmdq queue struct to hold the page
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int cmdq_allocate_page(struct hinic_cmdq_pages *cmdq_pages)
+{
+	return queue_alloc_page(cmdq_pages->hwif, &cmdq_pages->page_vaddr,
+				&cmdq_pages->page_paddr,
+				&cmdq_pages->shadow_page_vaddr,
+				CMDQ_PAGE_SIZE);
+}
+
+/**
+ * cmdq_free_page - free page from cmdq
+ * @cmdq_pages: the pages of the cmdq queue struct that hold the page
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static void cmdq_free_page(struct hinic_cmdq_pages *cmdq_pages)
+{
+	struct hinic_hwif *hwif = cmdq_pages->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+
+	dma_free_coherent(&pdev->dev, CMDQ_PAGE_SIZE,
+			  cmdq_pages->page_vaddr,
+			  (dma_addr_t)cmdq_pages->page_paddr);
+	vfree(cmdq_pages->shadow_page_vaddr);
+}
+
 static int alloc_page_arrays(struct hinic_wqs *wqs)
 {
 	struct hinic_hwif *hwif = wqs->hwif;
@@ -515,3 +564,110 @@ void hinic_wq_free(struct hinic_wqs *wqs, struct hinic_wq *wq)
 
 	wqs_return_block(wqs, wq->page_idx, wq->block_idx);
 }
+
+/**
+ * hinic_wqs_cmdq_alloc - Allocate wqs for cmdqs
+ * @cmdq_pages: will hold the pages of the cmdq
+ * @wq: returned wqs
+ * @hwif: HW interface
+ * @cmdq_blocks: number of cmdq blocks/wq to allocate
+ * @wqebb_size: Work Queue Block Byte Size
+ * @wq_page_size: the page size in the Work Queue
+ * @q_depth: number of wqebbs in WQ
+ * @max_wqe_size: maximum WQE size that will be used in the WQ
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_wqs_cmdq_alloc(struct hinic_cmdq_pages *cmdq_pages,
+			 struct hinic_wq *wq, struct hinic_hwif *hwif,
+			 int cmdq_blocks, u16 wqebb_size, u16 wq_page_size,
+			 u16 q_depth, u16 max_wqe_size)
+{
+	struct pci_dev *pdev = hwif->pdev;
+	u16 num_wqebbs_per_page;
+	int i, j, err = -ENOMEM;
+
+	if (wqebb_size == 0) {
+		dev_err(&pdev->dev, "wqebb_size must be > 0\n");
+		return -EINVAL;
+	}
+
+	if (wq_page_size == 0) {
+		dev_err(&pdev->dev, "wq_page_size must be > 0\n");
+		return -EINVAL;
+	}
+
+	if (q_depth & (q_depth - 1)) {
+		dev_err(&pdev->dev, "WQ q_depth must be power of 2\n");
+		return -EINVAL;
+	}
+
+	num_wqebbs_per_page = ALIGN(wq_page_size, wqebb_size) / wqebb_size;
+
+	if (num_wqebbs_per_page & (num_wqebbs_per_page - 1)) {
+		dev_err(&pdev->dev, "num wqebbs per page must be power of 2\n");
+		return -EINVAL;
+	}
+
+	cmdq_pages->hwif = hwif;
+
+	err = cmdq_allocate_page(cmdq_pages);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to allocate CMDQ page\n");
+		return err;
+	}
+
+	for (i = 0; i < cmdq_blocks; i++) {
+		wq[i].hwif = hwif;
+		wq[i].page_idx = 0;
+		wq[i].block_idx = i;
+
+		wq[i].wqebb_size = wqebb_size;
+		wq[i].wq_page_size = wq_page_size;
+		wq[i].q_depth = q_depth;
+		wq[i].max_wqe_size = max_wqe_size;
+		wq[i].num_wqebbs_per_page = num_wqebbs_per_page;
+
+		wq[i].block_vaddr = CMDQ_BASE_VADDR(cmdq_pages, &wq[i]);
+		wq[i].shadow_block_vaddr = CMDQ_BASE_ADDR(cmdq_pages, &wq[i]);
+		wq[i].block_paddr = CMDQ_BASE_PADDR(cmdq_pages, &wq[i]);
+
+		err = alloc_wq_pages(&wq[i], cmdq_pages->hwif,
+				     CMDQ_WQ_MAX_PAGES);
+		if (err) {
+			dev_err(&pdev->dev, "Failed to alloc CMDQ blocks\n");
+			goto err_cmdq_block;
+		}
+
+		atomic_set(&wq[i].cons_idx, 0);
+		atomic_set(&wq[i].prod_idx, 0);
+		atomic_set(&wq[i].delta, q_depth);
+		wq[i].mask = q_depth - 1;
+	}
+
+	return 0;
+
+err_cmdq_block:
+	for (j = 0; j < i; j++)
+		free_wq_pages(&wq[j], cmdq_pages->hwif, wq[j].num_q_pages);
+
+	cmdq_free_page(cmdq_pages);
+	return err;
+}
+
+/**
+ * hinic_wqs_cmdq_free - Free wqs from cmdqs
+ * @cmdq_pages: hold the pages of the cmdq
+ * @wq: wqs to free
+ * @cmdq_blocks: number of wqs to free
+ **/
+void hinic_wqs_cmdq_free(struct hinic_cmdq_pages *cmdq_pages,
+			 struct hinic_wq *wq, int cmdq_blocks)
+{
+	int i;
+
+	for (i = 0; i < cmdq_blocks; i++)
+		free_wq_pages(&wq[i], cmdq_pages->hwif, wq[i].num_q_pages);
+
+	cmdq_free_page(cmdq_pages);
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h
index 8ce259a..a3c4469 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h
@@ -81,6 +81,14 @@ struct hinic_cmdq_pages {
 	struct hinic_hwif       *hwif;
 };
 
+int hinic_wqs_cmdq_alloc(struct hinic_cmdq_pages *cmdq_pages,
+			 struct hinic_wq *wq, struct hinic_hwif *hwif,
+			 int cmdq_blocks, u16 wqebb_size, u16 wq_page_size,
+			 u16 q_depth, u16 max_wqe_size);
+
+void hinic_wqs_cmdq_free(struct hinic_cmdq_pages *cmdq_pages,
+			 struct hinic_wq *wq, int cmdq_blocks);
+
 int hinic_wqs_alloc(struct hinic_wqs *wqs, int num_wqs,
 		    struct hinic_hwif *hwif);
 
-- 
1.9.1

^ permalink raw reply related

* [PATCH V7 net-next 13/22] net-next/hinic: Set qp context
From: Aviad Krawczyk @ 2017-08-17 19:04 UTC (permalink / raw)
  To: davem
  Cc: linux-kernel, netdev, bc.y, victor.gissin, aviad.krawczyk,
	zhaochen6, tony.qu
In-Reply-To: <cover.1502996306.git.aviad.krawczyk@huawei.com>

Update the nic about the resources of the queue pairs.

Signed-off-by: Aviad Krawczyk <aviad.krawczyk@huawei.com>
Signed-off-by: Zhao Chen <zhaochen6@huawei.com>
---
 drivers/net/ethernet/huawei/hinic/Makefile         |   5 +-
 drivers/net/ethernet/huawei/hinic/hinic_common.c   |  55 ++++++
 drivers/net/ethernet/huawei/hinic/hinic_common.h   |   4 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c  |  87 +++++++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h  |  84 ++++++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c   |   4 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_io.c    | 151 +++++++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_io.h    |   5 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c    | 159 +++++++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h    |  11 ++
 .../net/ethernet/huawei/hinic/hinic_hw_qp_ctxt.h   | 214 +++++++++++++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h    |   9 +
 12 files changed, 786 insertions(+), 2 deletions(-)
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_common.c
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_hw_qp_ctxt.h

diff --git a/drivers/net/ethernet/huawei/hinic/Makefile b/drivers/net/ethernet/huawei/hinic/Makefile
index 84815f7..289ce88b 100644
--- a/drivers/net/ethernet/huawei/hinic/Makefile
+++ b/drivers/net/ethernet/huawei/hinic/Makefile
@@ -1,5 +1,6 @@
 obj-$(CONFIG_HINIC) += hinic.o
 
 hinic-y := hinic_main.o hinic_tx.o hinic_rx.o hinic_port.o hinic_hw_dev.o \
-	   hinic_hw_io.o hinic_hw_qp.o hinic_hw_wq.o hinic_hw_mgmt.o \
-	   hinic_hw_api_cmd.o hinic_hw_eqs.o hinic_hw_if.o
+	   hinic_hw_io.o hinic_hw_qp.o hinic_hw_cmdq.o hinic_hw_wq.o \
+	   hinic_hw_mgmt.o hinic_hw_api_cmd.o hinic_hw_eqs.o hinic_hw_if.o \
+	   hinic_common.o
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_common.c b/drivers/net/ethernet/huawei/hinic/hinic_common.c
new file mode 100644
index 0000000..1915ad6
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_common.c
@@ -0,0 +1,55 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#include <linux/types.h>
+#include <asm/byteorder.h>
+
+#include "hinic_common.h"
+
+/**
+ * hinic_cpu_to_be32 - convert data to big endian 32 bit format
+ * @data: the data to convert
+ * @len: length of data to convert
+ **/
+void hinic_cpu_to_be32(void *data, int len)
+{
+	u32 *mem = data;
+	int i;
+
+	len = len / sizeof(u32);
+
+	for (i = 0; i < len; i++) {
+		*mem = cpu_to_be32(*mem);
+		mem++;
+	}
+}
+
+/**
+ * hinic_be32_to_cpu - convert data from big endian 32 bit format
+ * @data: the data to convert
+ * @len: length of data to convert
+ **/
+void hinic_be32_to_cpu(void *data, int len)
+{
+	u32 *mem = data;
+	int i;
+
+	len = len / sizeof(u32);
+
+	for (i = 0; i < len; i++) {
+		*mem = be32_to_cpu(*mem);
+		mem++;
+	}
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_common.h b/drivers/net/ethernet/huawei/hinic/hinic_common.h
index 6a83c15..0f2f4ff 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_common.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_common.h
@@ -22,4 +22,8 @@ struct hinic_sge {
 	u32             len;
 };
 
+void hinic_cpu_to_be32(void *data, int len);
+
+void hinic_be32_to_cpu(void *data, int len);
+
 #endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c
new file mode 100644
index 0000000..2fd3924
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.c
@@ -0,0 +1,87 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#include <linux/types.h>
+#include <linux/errno.h>
+
+#include "hinic_hw_if.h"
+#include "hinic_hw_cmdq.h"
+
+/**
+ * hinic_alloc_cmdq_buf - alloc buffer for sending command
+ * @cmdqs: the cmdqs
+ * @cmdq_buf: the buffer returned in this struct
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_alloc_cmdq_buf(struct hinic_cmdqs *cmdqs,
+			 struct hinic_cmdq_buf *cmdq_buf)
+{
+	/* should be implemented */
+	return -ENOMEM;
+}
+
+/**
+ * hinic_free_cmdq_buf - free buffer
+ * @cmdqs: the cmdqs
+ * @cmdq_buf: the buffer to free that is in this struct
+ **/
+void hinic_free_cmdq_buf(struct hinic_cmdqs *cmdqs,
+			 struct hinic_cmdq_buf *cmdq_buf)
+{
+	/* should be implemented */
+}
+
+/**
+ * hinic_cmdq_direct_resp - send command with direct data as resp
+ * @cmdqs: the cmdqs
+ * @mod: module on the card that will handle the command
+ * @cmd: the command
+ * @buf_in: the buffer for the command
+ * @resp: the response to return
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_cmdq_direct_resp(struct hinic_cmdqs *cmdqs,
+			   enum hinic_mod_type mod, u8 cmd,
+			   struct hinic_cmdq_buf *buf_in, u64 *resp)
+{
+	/* should be implemented */
+	return -EINVAL;
+}
+
+/**
+ * hinic_init_cmdqs - init all cmdqs
+ * @cmdqs: cmdqs to init
+ * @hwif: HW interface for accessing cmdqs
+ * @db_area: doorbell areas for all the cmdqs
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_init_cmdqs(struct hinic_cmdqs *cmdqs, struct hinic_hwif *hwif,
+		     void __iomem **db_area)
+{
+	/* should be implemented */
+	return -EINVAL;
+}
+
+/**
+ * hinic_free_cmdqs - free all cmdqs
+ * @cmdqs: cmdqs to free
+ **/
+void hinic_free_cmdqs(struct hinic_cmdqs *cmdqs)
+{
+	/* should be implemented */
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h
new file mode 100644
index 0000000..c9e97ca
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_cmdq.h
@@ -0,0 +1,84 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#ifndef HINIC_CMDQ_H
+#define HINIC_CMDQ_H
+
+#include <linux/types.h>
+#include <linux/spinlock.h>
+#include <linux/completion.h>
+#include <linux/pci.h>
+
+#include "hinic_hw_if.h"
+#include "hinic_hw_wq.h"
+
+#define HINIC_CMDQ_BUF_SIZE             2048
+
+enum hinic_cmdq_type {
+	HINIC_CMDQ_SYNC,
+
+	HINIC_MAX_CMDQ_TYPES,
+};
+
+struct hinic_cmdq_buf {
+	void            *buf;
+	dma_addr_t      dma_addr;
+	size_t          size;
+};
+
+struct hinic_cmdq {
+	struct hinic_wq         *wq;
+
+	enum hinic_cmdq_type    cmdq_type;
+	int                     wrapped;
+
+	/* Lock for keeping the doorbell order */
+	spinlock_t              cmdq_lock;
+
+	struct completion       **done;
+	int                     **errcode;
+
+	/* doorbell area */
+	void __iomem            *db_base;
+};
+
+struct hinic_cmdqs {
+	struct hinic_hwif       *hwif;
+
+	struct pci_pool         *cmdq_buf_pool;
+
+	struct hinic_wq         *saved_wqs;
+
+	struct hinic_cmdq_pages cmdq_pages;
+
+	struct hinic_cmdq       cmdq[HINIC_MAX_CMDQ_TYPES];
+};
+
+int hinic_alloc_cmdq_buf(struct hinic_cmdqs *cmdqs,
+			 struct hinic_cmdq_buf *cmdq_buf);
+
+void hinic_free_cmdq_buf(struct hinic_cmdqs *cmdqs,
+			 struct hinic_cmdq_buf *cmdq_buf);
+
+int hinic_cmdq_direct_resp(struct hinic_cmdqs *cmdqs,
+			   enum hinic_mod_type mod, u8 cmd,
+			   struct hinic_cmdq_buf *buf_in, u64 *out_param);
+
+int hinic_init_cmdqs(struct hinic_cmdqs *cmdqs, struct hinic_hwif *hwif,
+		     void __iomem **db_area);
+
+void hinic_free_cmdqs(struct hinic_cmdqs *cmdqs);
+
+#endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
index d113908..cf0e165 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
@@ -25,6 +25,7 @@
 #include "hinic_hw_if.h"
 #include "hinic_hw_eqs.h"
 #include "hinic_hw_mgmt.h"
+#include "hinic_hw_qp_ctxt.h"
 #include "hinic_hw_qp.h"
 #include "hinic_hw_io.h"
 #include "hinic_hw_dev.h"
@@ -76,6 +77,9 @@ static int get_capability(struct hinic_hwdev *hwdev,
 	/* Each QP has its own (SQ + RQ) interrupts */
 	nic_cap->num_qps = (num_irqs - (num_aeqs + num_ceqs)) / 2;
 
+	if (nic_cap->num_qps > HINIC_Q_CTXT_MAX)
+		nic_cap->num_qps = HINIC_Q_CTXT_MAX;
+
 	/* num_qps must be power of 2 */
 	nic_cap->num_qps = BIT(fls(nic_cap->num_qps) - 1);
 
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c
index ad12cc7..bb4b93f 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c
@@ -27,6 +27,8 @@
 #include "hinic_hw_if.h"
 #include "hinic_hw_wqe.h"
 #include "hinic_hw_wq.h"
+#include "hinic_hw_cmdq.h"
+#include "hinic_hw_qp_ctxt.h"
 #include "hinic_hw_qp.h"
 #include "hinic_hw_io.h"
 
@@ -40,6 +42,10 @@
 #define DB_IDX(db, db_base)             \
 	(((unsigned long)(db) - (unsigned long)(db_base)) / HINIC_DB_PAGE_SIZE)
 
+enum io_cmd {
+	IO_CMD_MODIFY_QUEUE_CTXT = 0,
+};
+
 static void init_db_area_idx(struct hinic_free_db_area *free_db_area)
 {
 	int i;
@@ -100,6 +106,109 @@ static void return_db_area(struct hinic_func_to_io *func_to_io,
 	up(&free_db_area->idx_lock);
 }
 
+static int write_sq_ctxts(struct hinic_func_to_io *func_to_io, u16 base_qpn,
+			  u16 num_sqs)
+{
+	struct hinic_hwif *hwif = func_to_io->hwif;
+	struct hinic_sq_ctxt_block *sq_ctxt_block;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_cmdq_buf cmdq_buf;
+	struct hinic_sq_ctxt *sq_ctxt;
+	struct hinic_qp *qp;
+	u64 out_param;
+	int err, i;
+
+	err = hinic_alloc_cmdq_buf(&func_to_io->cmdqs, &cmdq_buf);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to allocate cmdq buf\n");
+		return err;
+	}
+
+	sq_ctxt_block = cmdq_buf.buf;
+	sq_ctxt = sq_ctxt_block->sq_ctxt;
+
+	hinic_qp_prepare_header(&sq_ctxt_block->hdr, HINIC_QP_CTXT_TYPE_SQ,
+				num_sqs, func_to_io->max_qps);
+	for (i = 0; i < num_sqs; i++) {
+		qp = &func_to_io->qps[i];
+
+		hinic_sq_prepare_ctxt(&sq_ctxt[i], &qp->sq,
+				      base_qpn + qp->q_id);
+	}
+
+	cmdq_buf.size = HINIC_SQ_CTXT_SIZE(num_sqs);
+
+	err = hinic_cmdq_direct_resp(&func_to_io->cmdqs, HINIC_MOD_L2NIC,
+				     IO_CMD_MODIFY_QUEUE_CTXT, &cmdq_buf,
+				     &out_param);
+	if ((err) || (out_param != 0)) {
+		dev_err(&pdev->dev, "Failed to set SQ ctxts\n");
+		err = -EFAULT;
+	}
+
+	hinic_free_cmdq_buf(&func_to_io->cmdqs, &cmdq_buf);
+	return err;
+}
+
+static int write_rq_ctxts(struct hinic_func_to_io *func_to_io, u16 base_qpn,
+			  u16 num_rqs)
+{
+	struct hinic_hwif *hwif = func_to_io->hwif;
+	struct hinic_rq_ctxt_block *rq_ctxt_block;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_cmdq_buf cmdq_buf;
+	struct hinic_rq_ctxt *rq_ctxt;
+	struct hinic_qp *qp;
+	u64 out_param;
+	int err, i;
+
+	err = hinic_alloc_cmdq_buf(&func_to_io->cmdqs, &cmdq_buf);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to allocate cmdq buf\n");
+		return err;
+	}
+
+	rq_ctxt_block = cmdq_buf.buf;
+	rq_ctxt = rq_ctxt_block->rq_ctxt;
+
+	hinic_qp_prepare_header(&rq_ctxt_block->hdr, HINIC_QP_CTXT_TYPE_RQ,
+				num_rqs, func_to_io->max_qps);
+	for (i = 0; i < num_rqs; i++) {
+		qp = &func_to_io->qps[i];
+
+		hinic_rq_prepare_ctxt(&rq_ctxt[i], &qp->rq,
+				      base_qpn + qp->q_id);
+	}
+
+	cmdq_buf.size = HINIC_RQ_CTXT_SIZE(num_rqs);
+
+	err = hinic_cmdq_direct_resp(&func_to_io->cmdqs, HINIC_MOD_L2NIC,
+				     IO_CMD_MODIFY_QUEUE_CTXT, &cmdq_buf,
+				     &out_param);
+	if ((err) || (out_param != 0)) {
+		dev_err(&pdev->dev, "Failed to set RQ ctxts\n");
+		err = -EFAULT;
+	}
+
+	hinic_free_cmdq_buf(&func_to_io->cmdqs, &cmdq_buf);
+	return err;
+}
+
+/**
+ * write_qp_ctxts - write the qp ctxt to HW
+ * @func_to_io: func to io channel that holds the IO components
+ * @base_qpn: first qp number
+ * @num_qps: number of qps to write
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int write_qp_ctxts(struct hinic_func_to_io *func_to_io, u16 base_qpn,
+			  u16 num_qps)
+{
+	return (write_sq_ctxts(func_to_io, base_qpn, num_qps) ||
+		write_rq_ctxts(func_to_io, base_qpn, num_qps));
+}
+
 /**
  * init_qp - Initialize a Queue Pair
  * @func_to_io: func to io channel that holds the IO components
@@ -265,8 +374,15 @@ int hinic_io_create_qps(struct hinic_func_to_io *func_to_io,
 		}
 	}
 
+	err = write_qp_ctxts(func_to_io, base_qpn, num_qps);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to init QP ctxts\n");
+		goto err_write_qp_ctxts;
+	}
+
 	return 0;
 
+err_write_qp_ctxts:
 err_init_qp:
 	for (j = 0; j < i; j++)
 		destroy_qp(func_to_io, &func_to_io->qps[j]);
@@ -331,6 +447,8 @@ int hinic_io_init(struct hinic_func_to_io *func_to_io,
 		  struct msix_entry *ceq_msix_entries)
 {
 	struct pci_dev *pdev = hwif->pdev;
+	enum hinic_cmdq_type cmdq, type;
+	void __iomem *db_area;
 	int err;
 
 	func_to_io->hwif = hwif;
@@ -351,8 +469,34 @@ int hinic_io_init(struct hinic_func_to_io *func_to_io,
 	}
 
 	init_db_area_idx(&func_to_io->free_db_area);
+
+	for (cmdq = HINIC_CMDQ_SYNC; cmdq < HINIC_MAX_CMDQ_TYPES; cmdq++) {
+		db_area = get_db_area(func_to_io);
+		if (IS_ERR(db_area)) {
+			dev_err(&pdev->dev, "Failed to get cmdq db area\n");
+			err = PTR_ERR(db_area);
+			goto err_db_area;
+		}
+
+		func_to_io->cmdq_db_area[cmdq] = db_area;
+	}
+
+	err = hinic_init_cmdqs(&func_to_io->cmdqs, hwif,
+			       func_to_io->cmdq_db_area);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to initialize cmdqs\n");
+		goto err_init_cmdqs;
+	}
+
 	return 0;
 
+err_init_cmdqs:
+err_db_area:
+	for (type = HINIC_CMDQ_SYNC; type < cmdq; type++)
+		return_db_area(func_to_io, func_to_io->cmdq_db_area[type]);
+
+	iounmap(func_to_io->db_base);
+
 err_db_ioremap:
 	hinic_wqs_free(&func_to_io->wqs);
 	return err;
@@ -364,6 +508,13 @@ int hinic_io_init(struct hinic_func_to_io *func_to_io,
  **/
 void hinic_io_free(struct hinic_func_to_io *func_to_io)
 {
+	enum hinic_cmdq_type cmdq;
+
+	hinic_free_cmdqs(&func_to_io->cmdqs);
+
+	for (cmdq = HINIC_CMDQ_SYNC; cmdq < HINIC_MAX_CMDQ_TYPES; cmdq++)
+		return_db_area(func_to_io, func_to_io->cmdq_db_area[cmdq]);
+
 	iounmap(func_to_io->db_base);
 	hinic_wqs_free(&func_to_io->wqs);
 }
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
index 2d85a38..60d77b34 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
@@ -23,6 +23,7 @@
 
 #include "hinic_hw_if.h"
 #include "hinic_hw_wq.h"
+#include "hinic_hw_cmdq.h"
 #include "hinic_hw_qp.h"
 
 #define HINIC_DB_PAGE_SIZE      SZ_4K
@@ -60,6 +61,10 @@ struct hinic_func_to_io {
 	dma_addr_t              ci_dma_base;
 
 	struct hinic_free_db_area       free_db_area;
+
+	void __iomem                    *cmdq_db_area[HINIC_MAX_CMDQ_TYPES];
+
+	struct hinic_cmdqs              cmdqs;
 };
 
 int hinic_io_create_qps(struct hinic_func_to_io *func_to_io,
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c
index 2b77b59..cfed040 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c
@@ -21,13 +21,172 @@
 #include <linux/vmalloc.h>
 #include <linux/errno.h>
 #include <linux/sizes.h>
+#include <linux/atomic.h>
+#include <asm/byteorder.h>
 
+#include "hinic_common.h"
 #include "hinic_hw_if.h"
 #include "hinic_hw_wq.h"
+#include "hinic_hw_qp_ctxt.h"
 #include "hinic_hw_qp.h"
 
 #define SQ_DB_OFF               SZ_2K
 
+/* The number of cache line to prefetch Until threshold state */
+#define WQ_PREFETCH_MAX         2
+/* The number of cache line to prefetch After threshold state */
+#define WQ_PREFETCH_MIN         1
+/* Threshold state */
+#define WQ_PREFETCH_THRESHOLD   256
+
+/* size of the SQ/RQ ctxt */
+#define Q_CTXT_SIZE             48
+#define CTXT_RSVD               240
+
+#define SQ_CTXT_OFFSET(max_sqs, max_rqs, q_id)  \
+		(((max_rqs) + (max_sqs)) * CTXT_RSVD + (q_id) * Q_CTXT_SIZE)
+
+#define RQ_CTXT_OFFSET(max_sqs, max_rqs, q_id)  \
+		(((max_rqs) + (max_sqs)) * CTXT_RSVD + (max_sqs) * Q_CTXT_SIZE \
+		 + (q_id) * Q_CTXT_SIZE)
+
+#define SIZE_16BYTES(size)      (ALIGN(size, 16) >> 4)
+
+void hinic_qp_prepare_header(struct hinic_qp_ctxt_header *qp_ctxt_hdr,
+			     enum hinic_qp_ctxt_type ctxt_type,
+			     int num_queues, int max_queues)
+{
+	qp_ctxt_hdr->num_queues = num_queues;
+	qp_ctxt_hdr->queue_type = ctxt_type;
+
+	if (ctxt_type == HINIC_QP_CTXT_TYPE_SQ)
+		qp_ctxt_hdr->addr_offset = SQ_CTXT_OFFSET(max_queues,
+							  max_queues, 0);
+	else
+		qp_ctxt_hdr->addr_offset = RQ_CTXT_OFFSET(max_queues,
+							  max_queues, 0);
+
+	qp_ctxt_hdr->addr_offset = SIZE_16BYTES(qp_ctxt_hdr->addr_offset);
+
+	hinic_cpu_to_be32(qp_ctxt_hdr, sizeof(*qp_ctxt_hdr));
+}
+
+void hinic_sq_prepare_ctxt(struct hinic_sq_ctxt *sq_ctxt,
+			   struct hinic_sq *sq, u16 global_qid)
+{
+	u32 wq_page_pfn_hi, wq_page_pfn_lo, wq_block_pfn_hi, wq_block_pfn_lo;
+	u64 wq_page_addr, wq_page_pfn, wq_block_pfn;
+	u16 pi_start, ci_start;
+	struct hinic_wq *wq;
+
+	wq = sq->wq;
+	ci_start = atomic_read(&wq->cons_idx);
+	pi_start = atomic_read(&wq->prod_idx);
+
+	/* Read the first page paddr from the WQ page paddr ptrs */
+	wq_page_addr = be64_to_cpu(*wq->block_vaddr);
+
+	wq_page_pfn = HINIC_WQ_PAGE_PFN(wq_page_addr);
+	wq_page_pfn_hi = upper_32_bits(wq_page_pfn);
+	wq_page_pfn_lo = lower_32_bits(wq_page_pfn);
+
+	wq_block_pfn = HINIC_WQ_BLOCK_PFN(wq->block_paddr);
+	wq_block_pfn_hi = upper_32_bits(wq_block_pfn);
+	wq_block_pfn_lo = lower_32_bits(wq_block_pfn);
+
+	sq_ctxt->ceq_attr = HINIC_SQ_CTXT_CEQ_ATTR_SET(global_qid,
+						       GLOBAL_SQ_ID) |
+			    HINIC_SQ_CTXT_CEQ_ATTR_SET(0, EN);
+
+	sq_ctxt->ci_wrapped = HINIC_SQ_CTXT_CI_SET(ci_start, IDX) |
+			      HINIC_SQ_CTXT_CI_SET(1, WRAPPED);
+
+	sq_ctxt->wq_hi_pfn_pi =
+			HINIC_SQ_CTXT_WQ_PAGE_SET(wq_page_pfn_hi, HI_PFN) |
+			HINIC_SQ_CTXT_WQ_PAGE_SET(pi_start, PI);
+
+	sq_ctxt->wq_lo_pfn = wq_page_pfn_lo;
+
+	sq_ctxt->pref_cache =
+		HINIC_SQ_CTXT_PREF_SET(WQ_PREFETCH_MIN, CACHE_MIN) |
+		HINIC_SQ_CTXT_PREF_SET(WQ_PREFETCH_MAX, CACHE_MAX) |
+		HINIC_SQ_CTXT_PREF_SET(WQ_PREFETCH_THRESHOLD, CACHE_THRESHOLD);
+
+	sq_ctxt->pref_wrapped = 1;
+
+	sq_ctxt->pref_wq_hi_pfn_ci =
+		HINIC_SQ_CTXT_PREF_SET(ci_start, CI) |
+		HINIC_SQ_CTXT_PREF_SET(wq_page_pfn_hi, WQ_HI_PFN);
+
+	sq_ctxt->pref_wq_lo_pfn = wq_page_pfn_lo;
+
+	sq_ctxt->wq_block_hi_pfn =
+		HINIC_SQ_CTXT_WQ_BLOCK_SET(wq_block_pfn_hi, HI_PFN);
+
+	sq_ctxt->wq_block_lo_pfn = wq_block_pfn_lo;
+
+	hinic_cpu_to_be32(sq_ctxt, sizeof(*sq_ctxt));
+}
+
+void hinic_rq_prepare_ctxt(struct hinic_rq_ctxt *rq_ctxt,
+			   struct hinic_rq *rq, u16 global_qid)
+{
+	u32 wq_page_pfn_hi, wq_page_pfn_lo, wq_block_pfn_hi, wq_block_pfn_lo;
+	u64 wq_page_addr, wq_page_pfn, wq_block_pfn;
+	u16 pi_start, ci_start;
+	struct hinic_wq *wq;
+
+	wq = rq->wq;
+	ci_start = atomic_read(&wq->cons_idx);
+	pi_start = atomic_read(&wq->prod_idx);
+
+	/* Read the first page paddr from the WQ page paddr ptrs */
+	wq_page_addr = be64_to_cpu(*wq->block_vaddr);
+
+	wq_page_pfn = HINIC_WQ_PAGE_PFN(wq_page_addr);
+	wq_page_pfn_hi = upper_32_bits(wq_page_pfn);
+	wq_page_pfn_lo = lower_32_bits(wq_page_pfn);
+
+	wq_block_pfn = HINIC_WQ_BLOCK_PFN(wq->block_paddr);
+	wq_block_pfn_hi = upper_32_bits(wq_block_pfn);
+	wq_block_pfn_lo = lower_32_bits(wq_block_pfn);
+
+	rq_ctxt->ceq_attr = HINIC_RQ_CTXT_CEQ_ATTR_SET(0, EN) |
+			    HINIC_RQ_CTXT_CEQ_ATTR_SET(1, WRAPPED);
+
+	rq_ctxt->pi_intr_attr = HINIC_RQ_CTXT_PI_SET(pi_start, IDX) |
+				HINIC_RQ_CTXT_PI_SET(rq->msix_entry, INTR);
+
+	rq_ctxt->wq_hi_pfn_ci = HINIC_RQ_CTXT_WQ_PAGE_SET(wq_page_pfn_hi,
+							  HI_PFN) |
+				HINIC_RQ_CTXT_WQ_PAGE_SET(ci_start, CI);
+
+	rq_ctxt->wq_lo_pfn = wq_page_pfn_lo;
+
+	rq_ctxt->pref_cache =
+		HINIC_RQ_CTXT_PREF_SET(WQ_PREFETCH_MIN, CACHE_MIN) |
+		HINIC_RQ_CTXT_PREF_SET(WQ_PREFETCH_MAX, CACHE_MAX) |
+		HINIC_RQ_CTXT_PREF_SET(WQ_PREFETCH_THRESHOLD, CACHE_THRESHOLD);
+
+	rq_ctxt->pref_wrapped = 1;
+
+	rq_ctxt->pref_wq_hi_pfn_ci =
+		HINIC_RQ_CTXT_PREF_SET(wq_page_pfn_hi, WQ_HI_PFN) |
+		HINIC_RQ_CTXT_PREF_SET(ci_start, CI);
+
+	rq_ctxt->pref_wq_lo_pfn = wq_page_pfn_lo;
+
+	rq_ctxt->pi_paddr_hi = upper_32_bits(rq->pi_dma_addr);
+	rq_ctxt->pi_paddr_lo = lower_32_bits(rq->pi_dma_addr);
+
+	rq_ctxt->wq_block_hi_pfn =
+		HINIC_RQ_CTXT_WQ_BLOCK_SET(wq_block_pfn_hi, HI_PFN);
+
+	rq_ctxt->wq_block_lo_pfn = wq_block_pfn_lo;
+
+	hinic_cpu_to_be32(rq_ctxt, sizeof(*rq_ctxt));
+}
+
 /**
  * alloc_sq_skb_arr - allocate sq array for saved skb
  * @sq: HW Send Queue
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
index c5ec30d..eb10bd9 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
@@ -24,6 +24,7 @@
 #include "hinic_hw_if.h"
 #include "hinic_hw_wqe.h"
 #include "hinic_hw_wq.h"
+#include "hinic_hw_qp_ctxt.h"
 
 #define HINIC_SQ_WQEBB_SIZE                     64
 #define HINIC_RQ_WQEBB_SIZE                     32
@@ -78,6 +79,16 @@ struct hinic_qp {
 	u16     q_id;
 };
 
+void hinic_qp_prepare_header(struct hinic_qp_ctxt_header *qp_ctxt_hdr,
+			     enum hinic_qp_ctxt_type ctxt_type,
+			     int num_queues, int max_queues);
+
+void hinic_sq_prepare_ctxt(struct hinic_sq_ctxt *sq_ctxt,
+			   struct hinic_sq *sq, u16 global_qid);
+
+void hinic_rq_prepare_ctxt(struct hinic_rq_ctxt *rq_ctxt,
+			   struct hinic_rq *rq, u16 global_qid);
+
 int hinic_init_sq(struct hinic_sq *sq, struct hinic_hwif *hwif,
 		  struct hinic_wq *wq, struct msix_entry *entry, void *ci_addr,
 		  dma_addr_t ci_dma_addr, void __iomem *db_base);
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp_ctxt.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp_ctxt.h
new file mode 100644
index 0000000..376abf0
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp_ctxt.h
@@ -0,0 +1,214 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#ifndef HINIC_HW_QP_CTXT_H
+#define HINIC_HW_QP_CTXT_H
+
+#include <linux/types.h>
+
+#include "hinic_hw_cmdq.h"
+
+#define HINIC_SQ_CTXT_CEQ_ATTR_GLOBAL_SQ_ID_SHIFT       13
+#define HINIC_SQ_CTXT_CEQ_ATTR_EN_SHIFT                 23
+
+#define HINIC_SQ_CTXT_CEQ_ATTR_GLOBAL_SQ_ID_MASK        0x3FF
+#define HINIC_SQ_CTXT_CEQ_ATTR_EN_MASK                  0x1
+
+#define HINIC_SQ_CTXT_CEQ_ATTR_SET(val, member)         \
+	(((u32)(val) & HINIC_SQ_CTXT_CEQ_ATTR_##member##_MASK) \
+	 << HINIC_SQ_CTXT_CEQ_ATTR_##member##_SHIFT)
+
+#define HINIC_SQ_CTXT_CI_IDX_SHIFT                      11
+#define HINIC_SQ_CTXT_CI_WRAPPED_SHIFT                  23
+
+#define HINIC_SQ_CTXT_CI_IDX_MASK                       0xFFF
+#define HINIC_SQ_CTXT_CI_WRAPPED_MASK                   0x1
+
+#define HINIC_SQ_CTXT_CI_SET(val, member)               \
+	(((u32)(val) & HINIC_SQ_CTXT_CI_##member##_MASK) \
+	 << HINIC_SQ_CTXT_CI_##member##_SHIFT)
+
+#define HINIC_SQ_CTXT_WQ_PAGE_HI_PFN_SHIFT              0
+#define HINIC_SQ_CTXT_WQ_PAGE_PI_SHIFT                  20
+
+#define HINIC_SQ_CTXT_WQ_PAGE_HI_PFN_MASK               0xFFFFF
+#define HINIC_SQ_CTXT_WQ_PAGE_PI_MASK                   0xFFF
+
+#define HINIC_SQ_CTXT_WQ_PAGE_SET(val, member)          \
+	(((u32)(val) & HINIC_SQ_CTXT_WQ_PAGE_##member##_MASK) \
+	 << HINIC_SQ_CTXT_WQ_PAGE_##member##_SHIFT)
+
+#define HINIC_SQ_CTXT_PREF_CACHE_THRESHOLD_SHIFT        0
+#define HINIC_SQ_CTXT_PREF_CACHE_MAX_SHIFT              14
+#define HINIC_SQ_CTXT_PREF_CACHE_MIN_SHIFT              25
+
+#define HINIC_SQ_CTXT_PREF_CACHE_THRESHOLD_MASK         0x3FFF
+#define HINIC_SQ_CTXT_PREF_CACHE_MAX_MASK               0x7FF
+#define HINIC_SQ_CTXT_PREF_CACHE_MIN_MASK               0x7F
+
+#define HINIC_SQ_CTXT_PREF_WQ_HI_PFN_SHIFT              0
+#define HINIC_SQ_CTXT_PREF_CI_SHIFT                     20
+
+#define HINIC_SQ_CTXT_PREF_WQ_HI_PFN_MASK               0xFFFFF
+#define HINIC_SQ_CTXT_PREF_CI_MASK                      0xFFF
+
+#define HINIC_SQ_CTXT_PREF_SET(val, member)             \
+	(((u32)(val) & HINIC_SQ_CTXT_PREF_##member##_MASK) \
+	 << HINIC_SQ_CTXT_PREF_##member##_SHIFT)
+
+#define HINIC_SQ_CTXT_WQ_BLOCK_HI_PFN_SHIFT             0
+
+#define HINIC_SQ_CTXT_WQ_BLOCK_HI_PFN_MASK              0x7FFFFF
+
+#define HINIC_SQ_CTXT_WQ_BLOCK_SET(val, member)         \
+	(((u32)(val) & HINIC_SQ_CTXT_WQ_BLOCK_##member##_MASK) \
+	 << HINIC_SQ_CTXT_WQ_BLOCK_##member##_SHIFT)
+
+#define HINIC_RQ_CTXT_CEQ_ATTR_EN_SHIFT                 0
+#define HINIC_RQ_CTXT_CEQ_ATTR_WRAPPED_SHIFT            1
+
+#define HINIC_RQ_CTXT_CEQ_ATTR_EN_MASK                  0x1
+#define HINIC_RQ_CTXT_CEQ_ATTR_WRAPPED_MASK             0x1
+
+#define HINIC_RQ_CTXT_CEQ_ATTR_SET(val, member)         \
+	(((u32)(val) & HINIC_RQ_CTXT_CEQ_ATTR_##member##_MASK) \
+	 << HINIC_RQ_CTXT_CEQ_ATTR_##member##_SHIFT)
+
+#define HINIC_RQ_CTXT_PI_IDX_SHIFT                      0
+#define HINIC_RQ_CTXT_PI_INTR_SHIFT                     22
+
+#define HINIC_RQ_CTXT_PI_IDX_MASK                       0xFFF
+#define HINIC_RQ_CTXT_PI_INTR_MASK                      0x3FF
+
+#define HINIC_RQ_CTXT_PI_SET(val, member)               \
+	(((u32)(val) & HINIC_RQ_CTXT_PI_##member##_MASK) << \
+	 HINIC_RQ_CTXT_PI_##member##_SHIFT)
+
+#define HINIC_RQ_CTXT_WQ_PAGE_HI_PFN_SHIFT              0
+#define HINIC_RQ_CTXT_WQ_PAGE_CI_SHIFT                  20
+
+#define HINIC_RQ_CTXT_WQ_PAGE_HI_PFN_MASK               0xFFFFF
+#define HINIC_RQ_CTXT_WQ_PAGE_CI_MASK                   0xFFF
+
+#define HINIC_RQ_CTXT_WQ_PAGE_SET(val, member)          \
+	(((u32)(val) & HINIC_RQ_CTXT_WQ_PAGE_##member##_MASK) << \
+	 HINIC_RQ_CTXT_WQ_PAGE_##member##_SHIFT)
+
+#define HINIC_RQ_CTXT_PREF_CACHE_THRESHOLD_SHIFT        0
+#define HINIC_RQ_CTXT_PREF_CACHE_MAX_SHIFT              14
+#define HINIC_RQ_CTXT_PREF_CACHE_MIN_SHIFT              25
+
+#define HINIC_RQ_CTXT_PREF_CACHE_THRESHOLD_MASK         0x3FFF
+#define HINIC_RQ_CTXT_PREF_CACHE_MAX_MASK               0x7FF
+#define HINIC_RQ_CTXT_PREF_CACHE_MIN_MASK               0x7F
+
+#define HINIC_RQ_CTXT_PREF_WQ_HI_PFN_SHIFT              0
+#define HINIC_RQ_CTXT_PREF_CI_SHIFT                     20
+
+#define HINIC_RQ_CTXT_PREF_WQ_HI_PFN_MASK               0xFFFFF
+#define HINIC_RQ_CTXT_PREF_CI_MASK                      0xFFF
+
+#define HINIC_RQ_CTXT_PREF_SET(val, member)             \
+	(((u32)(val) & HINIC_RQ_CTXT_PREF_##member##_MASK) << \
+	 HINIC_RQ_CTXT_PREF_##member##_SHIFT)
+
+#define HINIC_RQ_CTXT_WQ_BLOCK_HI_PFN_SHIFT             0
+
+#define HINIC_RQ_CTXT_WQ_BLOCK_HI_PFN_MASK              0x7FFFFF
+
+#define HINIC_RQ_CTXT_WQ_BLOCK_SET(val, member)         \
+	(((u32)(val) & HINIC_RQ_CTXT_WQ_BLOCK_##member##_MASK) << \
+	 HINIC_RQ_CTXT_WQ_BLOCK_##member##_SHIFT)
+
+#define HINIC_SQ_CTXT_SIZE(num_sqs) (sizeof(struct hinic_qp_ctxt_header) \
+				     + (num_sqs) * sizeof(struct hinic_sq_ctxt))
+
+#define HINIC_RQ_CTXT_SIZE(num_rqs) (sizeof(struct hinic_qp_ctxt_header) \
+				     + (num_rqs) * sizeof(struct hinic_rq_ctxt))
+
+#define HINIC_WQ_PAGE_PFN_SHIFT         12
+#define HINIC_WQ_BLOCK_PFN_SHIFT        9
+
+#define HINIC_WQ_PAGE_PFN(page_addr)    ((page_addr) >> HINIC_WQ_PAGE_PFN_SHIFT)
+#define HINIC_WQ_BLOCK_PFN(page_addr)   ((page_addr) >> \
+					 HINIC_WQ_BLOCK_PFN_SHIFT)
+
+#define HINIC_Q_CTXT_MAX                \
+		((HINIC_CMDQ_BUF_SIZE - sizeof(struct hinic_qp_ctxt_header)) \
+		 / sizeof(struct hinic_sq_ctxt))
+
+enum hinic_qp_ctxt_type {
+	HINIC_QP_CTXT_TYPE_SQ,
+	HINIC_QP_CTXT_TYPE_RQ
+};
+
+struct hinic_qp_ctxt_header {
+	u16     num_queues;
+	u16     queue_type;
+	u32     addr_offset;
+};
+
+struct hinic_sq_ctxt {
+	u32     ceq_attr;
+
+	u32     ci_wrapped;
+
+	u32     wq_hi_pfn_pi;
+	u32     wq_lo_pfn;
+
+	u32     pref_cache;
+	u32     pref_wrapped;
+	u32     pref_wq_hi_pfn_ci;
+	u32     pref_wq_lo_pfn;
+
+	u32     rsvd0;
+	u32     rsvd1;
+
+	u32     wq_block_hi_pfn;
+	u32     wq_block_lo_pfn;
+};
+
+struct hinic_rq_ctxt {
+	u32     ceq_attr;
+
+	u32     pi_intr_attr;
+
+	u32     wq_hi_pfn_ci;
+	u32     wq_lo_pfn;
+
+	u32     pref_cache;
+	u32     pref_wrapped;
+
+	u32     pref_wq_hi_pfn_ci;
+	u32     pref_wq_lo_pfn;
+
+	u32     pi_paddr_hi;
+	u32     pi_paddr_lo;
+
+	u32     wq_block_hi_pfn;
+	u32     wq_block_lo_pfn;
+};
+
+struct hinic_sq_ctxt_block {
+	struct hinic_qp_ctxt_header hdr;
+	struct hinic_sq_ctxt sq_ctxt[HINIC_Q_CTXT_MAX];
+};
+
+struct hinic_rq_ctxt_block {
+	struct hinic_qp_ctxt_header hdr;
+	struct hinic_rq_ctxt rq_ctxt[HINIC_Q_CTXT_MAX];
+};
+
+#endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h
index 7c114da..8ce259a 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h
@@ -72,6 +72,15 @@ struct hinic_wqs {
 	struct semaphore        alloc_blocks_lock;
 };
 
+struct hinic_cmdq_pages {
+	/* The addresses are 64 bit in the HW */
+	u64                     page_paddr;
+	u64                     *page_vaddr;
+	void                    **shadow_page_vaddr;
+
+	struct hinic_hwif       *hwif;
+};
+
 int hinic_wqs_alloc(struct hinic_wqs *wqs, int num_wqs,
 		    struct hinic_hwif *hwif);
 
-- 
1.9.1

^ permalink raw reply related

* [PATCH V7 net-next 12/22] net-next/hinic: Add qp resources
From: Aviad Krawczyk @ 2017-08-17 19:04 UTC (permalink / raw)
  To: davem
  Cc: linux-kernel, netdev, bc.y, victor.gissin, aviad.krawczyk,
	zhaochen6, tony.qu
In-Reply-To: <cover.1502996306.git.aviad.krawczyk@huawei.com>

Create the resources for queue pair operations: doorbell area,
consumer index address and producer index address.

Signed-off-by: Aviad Krawczyk <aviad.krawczyk@huawei.com>
Signed-off-by: Zhao Chen <zhaochen6@huawei.com>
---
 drivers/net/ethernet/huawei/hinic/Makefile      |   4 +-
 drivers/net/ethernet/huawei/hinic/hinic_hw_if.h |   1 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_io.c | 164 ++++++++++++++-
 drivers/net/ethernet/huawei/hinic/hinic_hw_io.h |  27 +++
 drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c | 266 ++++++++++++++++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h |  50 ++++-
 6 files changed, 507 insertions(+), 5 deletions(-)
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c

diff --git a/drivers/net/ethernet/huawei/hinic/Makefile b/drivers/net/ethernet/huawei/hinic/Makefile
index 0575a34..84815f7 100644
--- a/drivers/net/ethernet/huawei/hinic/Makefile
+++ b/drivers/net/ethernet/huawei/hinic/Makefile
@@ -1,5 +1,5 @@
 obj-$(CONFIG_HINIC) += hinic.o
 
 hinic-y := hinic_main.o hinic_tx.o hinic_rx.o hinic_port.o hinic_hw_dev.o \
-	   hinic_hw_io.o hinic_hw_wq.o hinic_hw_mgmt.o hinic_hw_api_cmd.o \
-	   hinic_hw_eqs.o hinic_hw_if.o
+	   hinic_hw_io.o hinic_hw_qp.o hinic_hw_wq.o hinic_hw_mgmt.o \
+	   hinic_hw_api_cmd.o hinic_hw_eqs.o hinic_hw_if.o
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_if.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_if.h
index 2280698..8f59195 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_if.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_if.h
@@ -137,6 +137,7 @@
 #define HINIC_IS_PPF(hwif)              (HINIC_FUNC_TYPE(hwif) == HINIC_PPF)
 
 #define HINIC_PCI_CFG_REGS_BAR          0
+#define HINIC_PCI_DB_BAR                4
 
 #define HINIC_PCIE_ST_DISABLE           0
 #define HINIC_PCIE_AT_DISABLE           0
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c
index 1bf944e..ad12cc7 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c
@@ -13,11 +13,16 @@
  *
  */
 
+#include <linux/kernel.h>
 #include <linux/types.h>
 #include <linux/pci.h>
 #include <linux/device.h>
 #include <linux/errno.h>
 #include <linux/slab.h>
+#include <linux/semaphore.h>
+#include <linux/dma-mapping.h>
+#include <linux/io.h>
+#include <linux/err.h>
 
 #include "hinic_hw_if.h"
 #include "hinic_hw_wqe.h"
@@ -25,6 +30,76 @@
 #include "hinic_hw_qp.h"
 #include "hinic_hw_io.h"
 
+#define CI_Q_ADDR_SIZE                  sizeof(u32)
+
+#define CI_ADDR(base_addr, q_id)        ((base_addr) + \
+					 (q_id) * CI_Q_ADDR_SIZE)
+
+#define CI_TABLE_SIZE(num_qps)          ((num_qps) * CI_Q_ADDR_SIZE)
+
+#define DB_IDX(db, db_base)             \
+	(((unsigned long)(db) - (unsigned long)(db_base)) / HINIC_DB_PAGE_SIZE)
+
+static void init_db_area_idx(struct hinic_free_db_area *free_db_area)
+{
+	int i;
+
+	for (i = 0; i < HINIC_DB_MAX_AREAS; i++)
+		free_db_area->db_idx[i] = i;
+
+	free_db_area->alloc_pos = 0;
+	free_db_area->return_pos = HINIC_DB_MAX_AREAS;
+
+	free_db_area->num_free = HINIC_DB_MAX_AREAS;
+
+	sema_init(&free_db_area->idx_lock, 1);
+}
+
+static void __iomem *get_db_area(struct hinic_func_to_io *func_to_io)
+{
+	struct hinic_free_db_area *free_db_area = &func_to_io->free_db_area;
+	int pos, idx;
+
+	down(&free_db_area->idx_lock);
+
+	free_db_area->num_free--;
+
+	if (free_db_area->num_free < 0) {
+		free_db_area->num_free++;
+		up(&free_db_area->idx_lock);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	pos = free_db_area->alloc_pos++;
+	pos &= HINIC_DB_MAX_AREAS - 1;
+
+	idx = free_db_area->db_idx[pos];
+
+	free_db_area->db_idx[pos] = -1;
+
+	up(&free_db_area->idx_lock);
+
+	return func_to_io->db_base + idx * HINIC_DB_PAGE_SIZE;
+}
+
+static void return_db_area(struct hinic_func_to_io *func_to_io,
+			   void __iomem *db_base)
+{
+	struct hinic_free_db_area *free_db_area = &func_to_io->free_db_area;
+	int pos, idx = DB_IDX(db_base, func_to_io->db_base);
+
+	down(&free_db_area->idx_lock);
+
+	pos = free_db_area->return_pos++;
+	pos &= HINIC_DB_MAX_AREAS - 1;
+
+	free_db_area->db_idx[pos] = idx;
+
+	free_db_area->num_free++;
+
+	up(&free_db_area->idx_lock);
+}
+
 /**
  * init_qp - Initialize a Queue Pair
  * @func_to_io: func to io channel that holds the IO components
@@ -42,6 +117,7 @@ static int init_qp(struct hinic_func_to_io *func_to_io,
 {
 	struct hinic_hwif *hwif = func_to_io->hwif;
 	struct pci_dev *pdev = hwif->pdev;
+	void __iomem *db_base;
 	int err;
 
 	qp->q_id = q_id;
@@ -62,8 +138,42 @@ static int init_qp(struct hinic_func_to_io *func_to_io,
 		goto err_rq_alloc;
 	}
 
+	db_base = get_db_area(func_to_io);
+	if (IS_ERR(db_base)) {
+		dev_err(&pdev->dev, "Failed to get DB area for SQ\n");
+		err = PTR_ERR(db_base);
+		goto err_get_db;
+	}
+
+	func_to_io->sq_db[q_id] = db_base;
+
+	err = hinic_init_sq(&qp->sq, hwif, &func_to_io->sq_wq[q_id],
+			    sq_msix_entry,
+			    CI_ADDR(func_to_io->ci_addr_base, q_id),
+			    CI_ADDR(func_to_io->ci_dma_base, q_id), db_base);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to init SQ\n");
+		goto err_sq_init;
+	}
+
+	err = hinic_init_rq(&qp->rq, hwif, &func_to_io->rq_wq[q_id],
+			    rq_msix_entry);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to init RQ\n");
+		goto err_rq_init;
+	}
+
 	return 0;
 
+err_rq_init:
+	hinic_clean_sq(&qp->sq);
+
+err_sq_init:
+	return_db_area(func_to_io, db_base);
+
+err_get_db:
+	hinic_wq_free(&func_to_io->wqs, &func_to_io->rq_wq[q_id]);
+
 err_rq_alloc:
 	hinic_wq_free(&func_to_io->wqs, &func_to_io->sq_wq[q_id]);
 	return err;
@@ -79,6 +189,11 @@ static void destroy_qp(struct hinic_func_to_io *func_to_io,
 {
 	int q_id = qp->q_id;
 
+	hinic_clean_rq(&qp->rq);
+	hinic_clean_sq(&qp->sq);
+
+	return_db_area(func_to_io, func_to_io->sq_db[q_id]);
+
 	hinic_wq_free(&func_to_io->wqs, &func_to_io->rq_wq[q_id]);
 	hinic_wq_free(&func_to_io->wqs, &func_to_io->sq_wq[q_id]);
 }
@@ -100,7 +215,8 @@ int hinic_io_create_qps(struct hinic_func_to_io *func_to_io,
 {
 	struct hinic_hwif *hwif = func_to_io->hwif;
 	struct pci_dev *pdev = hwif->pdev;
-	size_t qps_size, wq_size;
+	size_t qps_size, wq_size, db_size;
+	void *ci_addr_base;
 	int i, j, err;
 
 	qps_size = num_qps * sizeof(*func_to_io->qps);
@@ -122,6 +238,24 @@ int hinic_io_create_qps(struct hinic_func_to_io *func_to_io,
 		goto err_rq_wq;
 	}
 
+	db_size = num_qps * sizeof(*func_to_io->sq_db);
+	func_to_io->sq_db = devm_kzalloc(&pdev->dev, db_size, GFP_KERNEL);
+	if (!func_to_io->sq_db) {
+		err = -ENOMEM;
+		goto err_sq_db;
+	}
+
+	ci_addr_base = dma_zalloc_coherent(&pdev->dev, CI_TABLE_SIZE(num_qps),
+					   &func_to_io->ci_dma_base,
+					   GFP_KERNEL);
+	if (!ci_addr_base) {
+		dev_err(&pdev->dev, "Failed to allocate CI area\n");
+		err = -ENOMEM;
+		goto err_ci_base;
+	}
+
+	func_to_io->ci_addr_base = ci_addr_base;
+
 	for (i = 0; i < num_qps; i++) {
 		err = init_qp(func_to_io, &func_to_io->qps[i], i,
 			      &sq_msix_entries[i], &rq_msix_entries[i]);
@@ -137,6 +271,13 @@ int hinic_io_create_qps(struct hinic_func_to_io *func_to_io,
 	for (j = 0; j < i; j++)
 		destroy_qp(func_to_io, &func_to_io->qps[j]);
 
+	dma_free_coherent(&pdev->dev, CI_TABLE_SIZE(num_qps),
+			  func_to_io->ci_addr_base, func_to_io->ci_dma_base);
+
+err_ci_base:
+	devm_kfree(&pdev->dev, func_to_io->sq_db);
+
+err_sq_db:
 	devm_kfree(&pdev->dev, func_to_io->rq_wq);
 
 err_rq_wq:
@@ -156,11 +297,19 @@ void hinic_io_destroy_qps(struct hinic_func_to_io *func_to_io, int num_qps)
 {
 	struct hinic_hwif *hwif = func_to_io->hwif;
 	struct pci_dev *pdev = hwif->pdev;
+	size_t ci_table_size;
 	int i;
 
+	ci_table_size = CI_TABLE_SIZE(num_qps);
+
 	for (i = 0; i < num_qps; i++)
 		destroy_qp(func_to_io, &func_to_io->qps[i]);
 
+	dma_free_coherent(&pdev->dev, ci_table_size, func_to_io->ci_addr_base,
+			  func_to_io->ci_dma_base);
+
+	devm_kfree(&pdev->dev, func_to_io->sq_db);
+
 	devm_kfree(&pdev->dev, func_to_io->rq_wq);
 	devm_kfree(&pdev->dev, func_to_io->sq_wq);
 
@@ -194,7 +343,19 @@ int hinic_io_init(struct hinic_func_to_io *func_to_io,
 		return err;
 	}
 
+	func_to_io->db_base = pci_ioremap_bar(pdev, HINIC_PCI_DB_BAR);
+	if (!func_to_io->db_base) {
+		dev_err(&pdev->dev, "Failed to remap IO DB area\n");
+		err = -ENOMEM;
+		goto err_db_ioremap;
+	}
+
+	init_db_area_idx(&func_to_io->free_db_area);
 	return 0;
+
+err_db_ioremap:
+	hinic_wqs_free(&func_to_io->wqs);
+	return err;
 }
 
 /**
@@ -203,5 +364,6 @@ int hinic_io_init(struct hinic_func_to_io *func_to_io,
  **/
 void hinic_io_free(struct hinic_func_to_io *func_to_io)
 {
+	iounmap(func_to_io->db_base);
 	hinic_wqs_free(&func_to_io->wqs);
 }
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
index 6cacb8e..2d85a38 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
@@ -18,11 +18,30 @@
 
 #include <linux/types.h>
 #include <linux/pci.h>
+#include <linux/semaphore.h>
+#include <linux/sizes.h>
 
 #include "hinic_hw_if.h"
 #include "hinic_hw_wq.h"
 #include "hinic_hw_qp.h"
 
+#define HINIC_DB_PAGE_SIZE      SZ_4K
+#define HINIC_DB_SIZE           SZ_4M
+
+#define HINIC_DB_MAX_AREAS      (HINIC_DB_SIZE / HINIC_DB_PAGE_SIZE)
+
+struct hinic_free_db_area {
+	int             db_idx[HINIC_DB_MAX_AREAS];
+
+	int             alloc_pos;
+	int             return_pos;
+
+	int             num_free;
+
+	/* Lock for getting db area */
+	struct semaphore        idx_lock;
+};
+
 struct hinic_func_to_io {
 	struct hinic_hwif       *hwif;
 
@@ -33,6 +52,14 @@ struct hinic_func_to_io {
 
 	struct hinic_qp         *qps;
 	u16                     max_qps;
+
+	void __iomem            **sq_db;
+	void __iomem            *db_base;
+
+	void                    *ci_addr_base;
+	dma_addr_t              ci_dma_base;
+
+	struct hinic_free_db_area       free_db_area;
 };
 
 int hinic_io_create_qps(struct hinic_func_to_io *func_to_io,
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c
new file mode 100644
index 0000000..2b77b59
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.c
@@ -0,0 +1,266 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/pci.h>
+#include <linux/device.h>
+#include <linux/dma-mapping.h>
+#include <linux/vmalloc.h>
+#include <linux/errno.h>
+#include <linux/sizes.h>
+
+#include "hinic_hw_if.h"
+#include "hinic_hw_wq.h"
+#include "hinic_hw_qp.h"
+
+#define SQ_DB_OFF               SZ_2K
+
+/**
+ * alloc_sq_skb_arr - allocate sq array for saved skb
+ * @sq: HW Send Queue
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int alloc_sq_skb_arr(struct hinic_sq *sq)
+{
+	struct hinic_wq *wq = sq->wq;
+	size_t skb_arr_size;
+
+	skb_arr_size = wq->q_depth * sizeof(*sq->saved_skb);
+	sq->saved_skb = vzalloc(skb_arr_size);
+	if (!sq->saved_skb)
+		return -ENOMEM;
+
+	return 0;
+}
+
+/**
+ * free_sq_skb_arr - free sq array for saved skb
+ * @sq: HW Send Queue
+ **/
+static void free_sq_skb_arr(struct hinic_sq *sq)
+{
+	vfree(sq->saved_skb);
+}
+
+/**
+ * alloc_rq_skb_arr - allocate rq array for saved skb
+ * @rq: HW Receive Queue
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int alloc_rq_skb_arr(struct hinic_rq *rq)
+{
+	struct hinic_wq *wq = rq->wq;
+	size_t skb_arr_size;
+
+	skb_arr_size = wq->q_depth * sizeof(*rq->saved_skb);
+	rq->saved_skb = vzalloc(skb_arr_size);
+	if (!rq->saved_skb)
+		return -ENOMEM;
+
+	return 0;
+}
+
+/**
+ * free_rq_skb_arr - free rq array for saved skb
+ * @rq: HW Receive Queue
+ **/
+static void free_rq_skb_arr(struct hinic_rq *rq)
+{
+	vfree(rq->saved_skb);
+}
+
+/**
+ * hinic_init_sq - Initialize HW Send Queue
+ * @sq: HW Send Queue
+ * @hwif: HW Interface for accessing HW
+ * @wq: Work Queue for the data of the SQ
+ * @entry: msix entry for sq
+ * @ci_addr: address for reading the current HW consumer index
+ * @ci_dma_addr: dma address for reading the current HW consumer index
+ * @db_base: doorbell base address
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_init_sq(struct hinic_sq *sq, struct hinic_hwif *hwif,
+		  struct hinic_wq *wq, struct msix_entry *entry,
+		  void *ci_addr, dma_addr_t ci_dma_addr,
+		  void __iomem *db_base)
+{
+	sq->hwif = hwif;
+
+	sq->wq = wq;
+
+	sq->irq = entry->vector;
+	sq->msix_entry = entry->entry;
+
+	sq->hw_ci_addr = ci_addr;
+	sq->hw_ci_dma_addr = ci_dma_addr;
+
+	sq->db_base = db_base + SQ_DB_OFF;
+
+	return alloc_sq_skb_arr(sq);
+}
+
+/**
+ * hinic_clean_sq - Clean HW Send Queue's Resources
+ * @sq: Send Queue
+ **/
+void hinic_clean_sq(struct hinic_sq *sq)
+{
+	free_sq_skb_arr(sq);
+}
+
+/**
+ * alloc_rq_cqe - allocate rq completion queue elements
+ * @rq: HW Receive Queue
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int alloc_rq_cqe(struct hinic_rq *rq)
+{
+	struct hinic_hwif *hwif = rq->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	size_t cqe_dma_size, cqe_size;
+	struct hinic_wq *wq = rq->wq;
+	int j, i;
+
+	cqe_size = wq->q_depth * sizeof(*rq->cqe);
+	rq->cqe = vzalloc(cqe_size);
+	if (!rq->cqe)
+		return -ENOMEM;
+
+	cqe_dma_size = wq->q_depth * sizeof(*rq->cqe_dma);
+	rq->cqe_dma = vzalloc(cqe_dma_size);
+	if (!rq->cqe_dma)
+		goto err_cqe_dma_arr_alloc;
+
+	for (i = 0; i < wq->q_depth; i++) {
+		rq->cqe[i] = dma_zalloc_coherent(&pdev->dev,
+						 sizeof(*rq->cqe[i]),
+						 &rq->cqe_dma[i], GFP_KERNEL);
+		if (!rq->cqe[i])
+			goto err_cqe_alloc;
+	}
+
+	return 0;
+
+err_cqe_alloc:
+	for (j = 0; j < i; j++)
+		dma_free_coherent(&pdev->dev, sizeof(*rq->cqe[j]), rq->cqe[j],
+				  rq->cqe_dma[j]);
+
+	vfree(rq->cqe_dma);
+
+err_cqe_dma_arr_alloc:
+	vfree(rq->cqe);
+	return -ENOMEM;
+}
+
+/**
+ * free_rq_cqe - free rq completion queue elements
+ * @rq: HW Receive Queue
+ **/
+static void free_rq_cqe(struct hinic_rq *rq)
+{
+	struct hinic_hwif *hwif = rq->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_wq *wq = rq->wq;
+	int i;
+
+	for (i = 0; i < wq->q_depth; i++)
+		dma_free_coherent(&pdev->dev, sizeof(*rq->cqe[i]), rq->cqe[i],
+				  rq->cqe_dma[i]);
+
+	vfree(rq->cqe_dma);
+	vfree(rq->cqe);
+}
+
+/**
+ * hinic_init_rq - Initialize HW Receive Queue
+ * @rq: HW Receive Queue
+ * @hwif: HW Interface for accessing HW
+ * @wq: Work Queue for the data of the RQ
+ * @entry: msix entry for rq
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_init_rq(struct hinic_rq *rq, struct hinic_hwif *hwif,
+		  struct hinic_wq *wq, struct msix_entry *entry)
+{
+	struct pci_dev *pdev = hwif->pdev;
+	size_t pi_size;
+	int err;
+
+	rq->hwif = hwif;
+
+	rq->wq = wq;
+
+	rq->irq = entry->vector;
+	rq->msix_entry = entry->entry;
+
+	rq->buf_sz = HINIC_RX_BUF_SZ;
+
+	err = alloc_rq_skb_arr(rq);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to allocate rq priv data\n");
+		return err;
+	}
+
+	err = alloc_rq_cqe(rq);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to allocate rq cqe\n");
+		goto err_alloc_rq_cqe;
+	}
+
+	/* HW requirements: Must be at least 32 bit */
+	pi_size = ALIGN(sizeof(*rq->pi_virt_addr), sizeof(u32));
+	rq->pi_virt_addr = dma_zalloc_coherent(&pdev->dev, pi_size,
+					       &rq->pi_dma_addr, GFP_KERNEL);
+	if (!rq->pi_virt_addr) {
+		dev_err(&pdev->dev, "Failed to allocate PI address\n");
+		err = -ENOMEM;
+		goto err_pi_virt;
+	}
+
+	return 0;
+
+err_pi_virt:
+	free_rq_cqe(rq);
+
+err_alloc_rq_cqe:
+	free_rq_skb_arr(rq);
+	return err;
+}
+
+/**
+ * hinic_clean_rq - Clean HW Receive Queue's Resources
+ * @rq: HW Receive Queue
+ **/
+void hinic_clean_rq(struct hinic_rq *rq)
+{
+	struct hinic_hwif *hwif = rq->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	size_t pi_size;
+
+	pi_size = ALIGN(sizeof(*rq->pi_virt_addr), sizeof(u32));
+	dma_free_coherent(&pdev->dev, pi_size, rq->pi_virt_addr,
+			  rq->pi_dma_addr);
+
+	free_rq_cqe(rq);
+	free_rq_skb_arr(rq);
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
index 4031728..c5ec30d 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
@@ -18,6 +18,12 @@
 
 #include <linux/types.h>
 #include <linux/sizes.h>
+#include <linux/pci.h>
+#include <linux/skbuff.h>
+
+#include "hinic_hw_if.h"
+#include "hinic_hw_wqe.h"
+#include "hinic_hw_wq.h"
 
 #define HINIC_SQ_WQEBB_SIZE                     64
 #define HINIC_RQ_WQEBB_SIZE                     32
@@ -28,12 +34,41 @@
 #define HINIC_SQ_DEPTH                          SZ_4K
 #define HINIC_RQ_DEPTH                          SZ_4K
 
+#define HINIC_RX_BUF_SZ                         2048
+
 struct hinic_sq {
-	/* should be implemented */
+	struct hinic_hwif       *hwif;
+
+	struct hinic_wq         *wq;
+
+	u32                     irq;
+	u16                     msix_entry;
+
+	void                    *hw_ci_addr;
+	dma_addr_t              hw_ci_dma_addr;
+
+	void __iomem            *db_base;
+
+	struct sk_buff          **saved_skb;
 };
 
 struct hinic_rq {
-	/* should be implemented */
+	struct hinic_hwif       *hwif;
+
+	struct hinic_wq         *wq;
+
+	u32                     irq;
+	u16                     msix_entry;
+
+	size_t                  buf_sz;
+
+	struct sk_buff          **saved_skb;
+
+	struct hinic_rq_cqe     **cqe;
+	dma_addr_t              *cqe_dma;
+
+	u16                     *pi_virt_addr;
+	dma_addr_t              pi_dma_addr;
 };
 
 struct hinic_qp {
@@ -43,4 +78,15 @@ struct hinic_qp {
 	u16     q_id;
 };
 
+int hinic_init_sq(struct hinic_sq *sq, struct hinic_hwif *hwif,
+		  struct hinic_wq *wq, struct msix_entry *entry, void *ci_addr,
+		  dma_addr_t ci_dma_addr, void __iomem *db_base);
+
+void hinic_clean_sq(struct hinic_sq *sq);
+
+int hinic_init_rq(struct hinic_rq *rq, struct hinic_hwif *hwif,
+		  struct hinic_wq *wq, struct msix_entry *entry);
+
+void hinic_clean_rq(struct hinic_rq *rq);
+
 #endif
-- 
1.9.1

^ permalink raw reply related

* [PATCH V7 net-next 11/22] net-next/hinic: Add wq
From: Aviad Krawczyk @ 2017-08-17 19:04 UTC (permalink / raw)
  To: davem
  Cc: linux-kernel, netdev, bc.y, victor.gissin, aviad.krawczyk,
	zhaochen6, tony.qu
In-Reply-To: <cover.1502996306.git.aviad.krawczyk@huawei.com>

Create work queues for being used by the queue pairs.

Signed-off-by: Aviad Krawczyk <aviad.krawczyk@huawei.com>
Signed-off-by: Zhao Chen <zhaochen6@huawei.com>
---
 drivers/net/ethernet/huawei/hinic/Makefile       |   4 +-
 drivers/net/ethernet/huawei/hinic/hinic_common.h |  25 ++
 drivers/net/ethernet/huawei/hinic/hinic_hw_io.c  |  69 ++-
 drivers/net/ethernet/huawei/hinic/hinic_hw_io.h  |   6 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h  |  14 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c  | 517 +++++++++++++++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h  |  86 ++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_wqe.h | 253 +++++++++++
 8 files changed, 969 insertions(+), 5 deletions(-)
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_common.h
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_hw_wqe.h

diff --git a/drivers/net/ethernet/huawei/hinic/Makefile b/drivers/net/ethernet/huawei/hinic/Makefile
index f60c449..0575a34 100644
--- a/drivers/net/ethernet/huawei/hinic/Makefile
+++ b/drivers/net/ethernet/huawei/hinic/Makefile
@@ -1,5 +1,5 @@
 obj-$(CONFIG_HINIC) += hinic.o
 
 hinic-y := hinic_main.o hinic_tx.o hinic_rx.o hinic_port.o hinic_hw_dev.o \
-	   hinic_hw_io.o hinic_hw_mgmt.o hinic_hw_api_cmd.o hinic_hw_eqs.o \
-	   hinic_hw_if.o
+	   hinic_hw_io.o hinic_hw_wq.o hinic_hw_mgmt.o hinic_hw_api_cmd.o \
+	   hinic_hw_eqs.o hinic_hw_if.o
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_common.h b/drivers/net/ethernet/huawei/hinic/hinic_common.h
new file mode 100644
index 0000000..6a83c15
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_common.h
@@ -0,0 +1,25 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#ifndef HINIC_COMMON_H
+#define HINIC_COMMON_H
+
+struct hinic_sge {
+	u32             hi_addr;
+	u32             lo_addr;
+	u32             len;
+};
+
+#endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c
index ebe28ee..1bf944e 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c
@@ -20,6 +20,8 @@
 #include <linux/slab.h>
 
 #include "hinic_hw_if.h"
+#include "hinic_hw_wqe.h"
+#include "hinic_hw_wq.h"
 #include "hinic_hw_qp.h"
 #include "hinic_hw_io.h"
 
@@ -38,8 +40,33 @@ static int init_qp(struct hinic_func_to_io *func_to_io,
 		   struct msix_entry *sq_msix_entry,
 		   struct msix_entry *rq_msix_entry)
 {
-	/* should be implemented */
+	struct hinic_hwif *hwif = func_to_io->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	int err;
+
+	qp->q_id = q_id;
+
+	err = hinic_wq_allocate(&func_to_io->wqs, &func_to_io->sq_wq[q_id],
+				HINIC_SQ_WQEBB_SIZE, HINIC_SQ_PAGE_SIZE,
+				HINIC_SQ_DEPTH, HINIC_SQ_WQE_MAX_SIZE);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to allocate WQ for SQ\n");
+		return err;
+	}
+
+	err = hinic_wq_allocate(&func_to_io->wqs, &func_to_io->rq_wq[q_id],
+				HINIC_RQ_WQEBB_SIZE, HINIC_RQ_PAGE_SIZE,
+				HINIC_RQ_DEPTH, HINIC_RQ_WQE_SIZE);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to allocate WQ for RQ\n");
+		goto err_rq_alloc;
+	}
+
 	return 0;
+
+err_rq_alloc:
+	hinic_wq_free(&func_to_io->wqs, &func_to_io->sq_wq[q_id]);
+	return err;
 }
 
 /**
@@ -50,7 +77,10 @@ static int init_qp(struct hinic_func_to_io *func_to_io,
 static void destroy_qp(struct hinic_func_to_io *func_to_io,
 		       struct hinic_qp *qp)
 {
-	/* should be implemented */
+	int q_id = qp->q_id;
+
+	hinic_wq_free(&func_to_io->wqs, &func_to_io->rq_wq[q_id]);
+	hinic_wq_free(&func_to_io->wqs, &func_to_io->sq_wq[q_id]);
 }
 
 /**
@@ -70,7 +100,7 @@ int hinic_io_create_qps(struct hinic_func_to_io *func_to_io,
 {
 	struct hinic_hwif *hwif = func_to_io->hwif;
 	struct pci_dev *pdev = hwif->pdev;
-	size_t qps_size;
+	size_t qps_size, wq_size;
 	int i, j, err;
 
 	qps_size = num_qps * sizeof(*func_to_io->qps);
@@ -78,6 +108,20 @@ int hinic_io_create_qps(struct hinic_func_to_io *func_to_io,
 	if (!func_to_io->qps)
 		return -ENOMEM;
 
+	wq_size = num_qps * sizeof(*func_to_io->sq_wq);
+	func_to_io->sq_wq = devm_kzalloc(&pdev->dev, wq_size, GFP_KERNEL);
+	if (!func_to_io->sq_wq) {
+		err = -ENOMEM;
+		goto err_sq_wq;
+	}
+
+	wq_size = num_qps * sizeof(*func_to_io->rq_wq);
+	func_to_io->rq_wq = devm_kzalloc(&pdev->dev, wq_size, GFP_KERNEL);
+	if (!func_to_io->rq_wq) {
+		err = -ENOMEM;
+		goto err_rq_wq;
+	}
+
 	for (i = 0; i < num_qps; i++) {
 		err = init_qp(func_to_io, &func_to_io->qps[i], i,
 			      &sq_msix_entries[i], &rq_msix_entries[i]);
@@ -93,6 +137,12 @@ int hinic_io_create_qps(struct hinic_func_to_io *func_to_io,
 	for (j = 0; j < i; j++)
 		destroy_qp(func_to_io, &func_to_io->qps[j]);
 
+	devm_kfree(&pdev->dev, func_to_io->rq_wq);
+
+err_rq_wq:
+	devm_kfree(&pdev->dev, func_to_io->sq_wq);
+
+err_sq_wq:
 	devm_kfree(&pdev->dev, func_to_io->qps);
 	return err;
 }
@@ -111,6 +161,9 @@ void hinic_io_destroy_qps(struct hinic_func_to_io *func_to_io, int num_qps)
 	for (i = 0; i < num_qps; i++)
 		destroy_qp(func_to_io, &func_to_io->qps[i]);
 
+	devm_kfree(&pdev->dev, func_to_io->rq_wq);
+	devm_kfree(&pdev->dev, func_to_io->sq_wq);
+
 	devm_kfree(&pdev->dev, func_to_io->qps);
 }
 
@@ -128,10 +181,19 @@ int hinic_io_init(struct hinic_func_to_io *func_to_io,
 		  struct hinic_hwif *hwif, u16 max_qps, int num_ceqs,
 		  struct msix_entry *ceq_msix_entries)
 {
+	struct pci_dev *pdev = hwif->pdev;
+	int err;
+
 	func_to_io->hwif = hwif;
 	func_to_io->qps = NULL;
 	func_to_io->max_qps = max_qps;
 
+	err = hinic_wqs_alloc(&func_to_io->wqs, 2 * max_qps, hwif);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to allocate WQS for IO\n");
+		return err;
+	}
+
 	return 0;
 }
 
@@ -141,4 +203,5 @@ int hinic_io_init(struct hinic_func_to_io *func_to_io,
  **/
 void hinic_io_free(struct hinic_func_to_io *func_to_io)
 {
+	hinic_wqs_free(&func_to_io->wqs);
 }
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
index 7cdcffd..6cacb8e 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
@@ -20,11 +20,17 @@
 #include <linux/pci.h>
 
 #include "hinic_hw_if.h"
+#include "hinic_hw_wq.h"
 #include "hinic_hw_qp.h"
 
 struct hinic_func_to_io {
 	struct hinic_hwif       *hwif;
 
+	struct hinic_wqs        wqs;
+
+	struct hinic_wq         *sq_wq;
+	struct hinic_wq         *rq_wq;
+
 	struct hinic_qp         *qps;
 	u16                     max_qps;
 };
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
index 64330fb..4031728 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
@@ -16,6 +16,18 @@
 #ifndef HINIC_HW_QP_H
 #define HINIC_HW_QP_H
 
+#include <linux/types.h>
+#include <linux/sizes.h>
+
+#define HINIC_SQ_WQEBB_SIZE                     64
+#define HINIC_RQ_WQEBB_SIZE                     32
+
+#define HINIC_SQ_PAGE_SIZE                      SZ_4K
+#define HINIC_RQ_PAGE_SIZE                      SZ_4K
+
+#define HINIC_SQ_DEPTH                          SZ_4K
+#define HINIC_RQ_DEPTH                          SZ_4K
+
 struct hinic_sq {
 	/* should be implemented */
 };
@@ -27,6 +39,8 @@ struct hinic_rq {
 struct hinic_qp {
 	struct hinic_sq         sq;
 	struct hinic_rq         rq;
+
+	u16     q_id;
 };
 
 #endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c
new file mode 100644
index 0000000..88d370e
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.c
@@ -0,0 +1,517 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/pci.h>
+#include <linux/device.h>
+#include <linux/dma-mapping.h>
+#include <linux/slab.h>
+#include <linux/atomic.h>
+#include <linux/semaphore.h>
+#include <linux/errno.h>
+#include <linux/vmalloc.h>
+#include <asm/byteorder.h>
+
+#include "hinic_hw_if.h"
+#include "hinic_hw_wq.h"
+
+#define WQS_BLOCKS_PER_PAGE             4
+
+#define WQ_BLOCK_SIZE                   4096
+#define WQS_PAGE_SIZE                   (WQS_BLOCKS_PER_PAGE * WQ_BLOCK_SIZE)
+
+#define WQS_MAX_NUM_BLOCKS              128
+#define WQS_FREE_BLOCKS_SIZE(wqs)       (WQS_MAX_NUM_BLOCKS * \
+					 sizeof((wqs)->free_blocks[0]))
+
+#define WQ_SIZE(wq)                     ((wq)->q_depth * (wq)->wqebb_size)
+
+#define WQ_PAGE_ADDR_SIZE               sizeof(u64)
+#define WQ_MAX_PAGES                    (WQ_BLOCK_SIZE / WQ_PAGE_ADDR_SIZE)
+
+#define WQ_BASE_VADDR(wqs, wq)          \
+			((void *)((wqs)->page_vaddr[(wq)->page_idx]) \
+				+ (wq)->block_idx * WQ_BLOCK_SIZE)
+
+#define WQ_BASE_PADDR(wqs, wq)          \
+			((wqs)->page_paddr[(wq)->page_idx] \
+				+ (wq)->block_idx * WQ_BLOCK_SIZE)
+
+#define WQ_BASE_ADDR(wqs, wq)           \
+			((void *)((wqs)->shadow_page_vaddr[(wq)->page_idx]) \
+				+ (wq)->block_idx * WQ_BLOCK_SIZE)
+
+/**
+ * queue_alloc_page - allocate page for Queue
+ * @hwif: HW interface for allocating DMA
+ * @vaddr: virtual address will be returned in this address
+ * @paddr: physical address will be returned in this address
+ * @shadow_vaddr: VM area will be return here for holding WQ page addresses
+ * @page_sz: page size of each WQ page
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int queue_alloc_page(struct hinic_hwif *hwif, u64 **vaddr, u64 *paddr,
+			    void ***shadow_vaddr, size_t page_sz)
+{
+	struct pci_dev *pdev = hwif->pdev;
+	dma_addr_t dma_addr;
+
+	*vaddr = dma_zalloc_coherent(&pdev->dev, page_sz, &dma_addr,
+				     GFP_KERNEL);
+	if (!*vaddr) {
+		dev_err(&pdev->dev, "Failed to allocate dma for wqs page\n");
+		return -ENOMEM;
+	}
+
+	*paddr = (u64)dma_addr;
+
+	/* use vzalloc for big mem */
+	*shadow_vaddr = vzalloc(page_sz);
+	if (!*shadow_vaddr)
+		goto err_shadow_vaddr;
+
+	return 0;
+
+err_shadow_vaddr:
+	dma_free_coherent(&pdev->dev, page_sz, *vaddr, dma_addr);
+	return -ENOMEM;
+}
+
+/**
+ * wqs_allocate_page - allocate page for WQ set
+ * @wqs: Work Queue Set
+ * @page_idx: the page index of the page will be allocated
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int wqs_allocate_page(struct hinic_wqs *wqs, int page_idx)
+{
+	return queue_alloc_page(wqs->hwif, &wqs->page_vaddr[page_idx],
+				&wqs->page_paddr[page_idx],
+				&wqs->shadow_page_vaddr[page_idx],
+				WQS_PAGE_SIZE);
+}
+
+/**
+ * wqs_free_page - free page of WQ set
+ * @wqs: Work Queue Set
+ * @page_idx: the page index of the page will be freed
+ **/
+static void wqs_free_page(struct hinic_wqs *wqs, int page_idx)
+{
+	struct hinic_hwif *hwif = wqs->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+
+	dma_free_coherent(&pdev->dev, WQS_PAGE_SIZE,
+			  wqs->page_vaddr[page_idx],
+			  (dma_addr_t)wqs->page_paddr[page_idx]);
+	vfree(wqs->shadow_page_vaddr[page_idx]);
+}
+
+static int alloc_page_arrays(struct hinic_wqs *wqs)
+{
+	struct hinic_hwif *hwif = wqs->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	size_t size;
+
+	size = wqs->num_pages * sizeof(*wqs->page_paddr);
+	wqs->page_paddr = devm_kzalloc(&pdev->dev, size, GFP_KERNEL);
+	if (!wqs->page_paddr)
+		return -ENOMEM;
+
+	size = wqs->num_pages * sizeof(*wqs->page_vaddr);
+	wqs->page_vaddr = devm_kzalloc(&pdev->dev, size, GFP_KERNEL);
+	if (!wqs->page_vaddr)
+		goto err_page_vaddr;
+
+	size = wqs->num_pages * sizeof(*wqs->shadow_page_vaddr);
+	wqs->shadow_page_vaddr = devm_kzalloc(&pdev->dev, size, GFP_KERNEL);
+	if (!wqs->shadow_page_vaddr)
+		goto err_page_shadow_vaddr;
+
+	return 0;
+
+err_page_shadow_vaddr:
+	devm_kfree(&pdev->dev, wqs->page_vaddr);
+
+err_page_vaddr:
+	devm_kfree(&pdev->dev, wqs->page_paddr);
+	return -ENOMEM;
+}
+
+static void free_page_arrays(struct hinic_wqs *wqs)
+{
+	struct hinic_hwif *hwif = wqs->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+
+	devm_kfree(&pdev->dev, wqs->shadow_page_vaddr);
+	devm_kfree(&pdev->dev, wqs->page_vaddr);
+	devm_kfree(&pdev->dev, wqs->page_paddr);
+}
+
+static int wqs_next_block(struct hinic_wqs *wqs, int *page_idx,
+			  int *block_idx)
+{
+	int pos;
+
+	down(&wqs->alloc_blocks_lock);
+
+	wqs->num_free_blks--;
+
+	if (wqs->num_free_blks < 0) {
+		wqs->num_free_blks++;
+		up(&wqs->alloc_blocks_lock);
+		return -ENOMEM;
+	}
+
+	pos = wqs->alloc_blk_pos++;
+	pos &= WQS_MAX_NUM_BLOCKS - 1;
+
+	*page_idx = wqs->free_blocks[pos].page_idx;
+	*block_idx = wqs->free_blocks[pos].block_idx;
+
+	wqs->free_blocks[pos].page_idx = -1;
+	wqs->free_blocks[pos].block_idx = -1;
+
+	up(&wqs->alloc_blocks_lock);
+	return 0;
+}
+
+static void wqs_return_block(struct hinic_wqs *wqs, int page_idx,
+			     int block_idx)
+{
+	int pos;
+
+	down(&wqs->alloc_blocks_lock);
+
+	pos = wqs->return_blk_pos++;
+	pos &= WQS_MAX_NUM_BLOCKS - 1;
+
+	wqs->free_blocks[pos].page_idx = page_idx;
+	wqs->free_blocks[pos].block_idx = block_idx;
+
+	wqs->num_free_blks++;
+
+	up(&wqs->alloc_blocks_lock);
+}
+
+static void init_wqs_blocks_arr(struct hinic_wqs *wqs)
+{
+	int page_idx, blk_idx, pos = 0;
+
+	for (page_idx = 0; page_idx < wqs->num_pages; page_idx++) {
+		for (blk_idx = 0; blk_idx < WQS_BLOCKS_PER_PAGE; blk_idx++) {
+			wqs->free_blocks[pos].page_idx = page_idx;
+			wqs->free_blocks[pos].block_idx = blk_idx;
+			pos++;
+		}
+	}
+
+	wqs->alloc_blk_pos = 0;
+	wqs->return_blk_pos = pos;
+	wqs->num_free_blks = pos;
+
+	sema_init(&wqs->alloc_blocks_lock, 1);
+}
+
+/**
+ * hinic_wqs_alloc - allocate Work Queues set
+ * @wqs: Work Queue Set
+ * @max_wqs: maximum wqs to allocate
+ * @hwif: HW interface for use for the allocation
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_wqs_alloc(struct hinic_wqs *wqs, int max_wqs,
+		    struct hinic_hwif *hwif)
+{
+	struct pci_dev *pdev = hwif->pdev;
+	int err, i, page_idx;
+
+	max_wqs = ALIGN(max_wqs, WQS_BLOCKS_PER_PAGE);
+
+	if (max_wqs > WQS_MAX_NUM_BLOCKS)  {
+		dev_err(&pdev->dev, "Invalid max_wqs = %d\n", max_wqs);
+		return -EINVAL;
+	}
+
+	wqs->hwif = hwif;
+	wqs->num_pages = max_wqs / WQS_BLOCKS_PER_PAGE;
+
+	if (alloc_page_arrays(wqs)) {
+		dev_err(&pdev->dev,
+			"Failed to allocate mem for page addresses\n");
+		return -ENOMEM;
+	}
+
+	for (page_idx = 0; page_idx < wqs->num_pages; page_idx++) {
+		err = wqs_allocate_page(wqs, page_idx);
+		if (err) {
+			dev_err(&pdev->dev, "Failed wq page allocation\n");
+			goto err_wq_allocate_page;
+		}
+	}
+
+	wqs->free_blocks = devm_kzalloc(&pdev->dev, WQS_FREE_BLOCKS_SIZE(wqs),
+					GFP_KERNEL);
+	if (!wqs->free_blocks) {
+		err = -ENOMEM;
+		goto err_alloc_blocks;
+	}
+
+	init_wqs_blocks_arr(wqs);
+	return 0;
+
+err_alloc_blocks:
+err_wq_allocate_page:
+	for (i = 0; i < page_idx; i++)
+		wqs_free_page(wqs, i);
+
+	free_page_arrays(wqs);
+	return err;
+}
+
+/**
+ * hinic_wqs_free - free Work Queues set
+ * @wqs: Work Queue Set
+ **/
+void hinic_wqs_free(struct hinic_wqs *wqs)
+{
+	struct hinic_hwif *hwif = wqs->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	int page_idx;
+
+	devm_kfree(&pdev->dev, wqs->free_blocks);
+
+	for (page_idx = 0; page_idx < wqs->num_pages; page_idx++)
+		wqs_free_page(wqs, page_idx);
+
+	free_page_arrays(wqs);
+}
+
+/**
+ * alloc_wqes_shadow - allocate WQE shadows for WQ
+ * @wq: WQ to allocate shadows for
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int alloc_wqes_shadow(struct hinic_wq *wq)
+{
+	struct hinic_hwif *hwif = wq->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	size_t size;
+
+	size = wq->num_q_pages * wq->max_wqe_size;
+	wq->shadow_wqe = devm_kzalloc(&pdev->dev, size, GFP_KERNEL);
+	if (!wq->shadow_wqe)
+		return -ENOMEM;
+
+	size = wq->num_q_pages * sizeof(wq->prod_idx);
+	wq->shadow_idx = devm_kzalloc(&pdev->dev, size, GFP_KERNEL);
+	if (!wq->shadow_idx)
+		goto err_shadow_idx;
+
+	return 0;
+
+err_shadow_idx:
+	devm_kfree(&pdev->dev, wq->shadow_wqe);
+	return -ENOMEM;
+}
+
+/**
+ * free_wqes_shadow - free WQE shadows of WQ
+ * @wq: WQ to free shadows from
+ **/
+static void free_wqes_shadow(struct hinic_wq *wq)
+{
+	struct hinic_hwif *hwif = wq->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+
+	devm_kfree(&pdev->dev, wq->shadow_idx);
+	devm_kfree(&pdev->dev, wq->shadow_wqe);
+}
+
+/**
+ * free_wq_pages - free pages of WQ
+ * @hwif: HW interface for releasing dma addresses
+ * @wq: WQ to free pages from
+ * @num_q_pages: number pages to free
+ **/
+static void free_wq_pages(struct hinic_wq *wq, struct hinic_hwif *hwif,
+			  int num_q_pages)
+{
+	struct pci_dev *pdev = hwif->pdev;
+	int i;
+
+	for (i = 0; i < num_q_pages; i++) {
+		void **vaddr = &wq->shadow_block_vaddr[i];
+		u64 *paddr = &wq->block_vaddr[i];
+		dma_addr_t dma_addr;
+
+		dma_addr = (dma_addr_t)be64_to_cpu(*paddr);
+		dma_free_coherent(&pdev->dev, wq->wq_page_size, *vaddr,
+				  dma_addr);
+	}
+
+	free_wqes_shadow(wq);
+}
+
+/**
+ * alloc_wq_pages - alloc pages for WQ
+ * @hwif: HW interface for allocating dma addresses
+ * @wq: WQ to allocate pages for
+ * @max_pages: maximum pages allowed
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int alloc_wq_pages(struct hinic_wq *wq, struct hinic_hwif *hwif,
+			  int max_pages)
+{
+	struct pci_dev *pdev = hwif->pdev;
+	int i, err, num_q_pages;
+
+	num_q_pages = ALIGN(WQ_SIZE(wq), wq->wq_page_size) / wq->wq_page_size;
+	if (num_q_pages > max_pages) {
+		dev_err(&pdev->dev, "Number wq pages exceeds the limit\n");
+		return -EINVAL;
+	}
+
+	if (num_q_pages & (num_q_pages - 1)) {
+		dev_err(&pdev->dev, "Number wq pages must be power of 2\n");
+		return -EINVAL;
+	}
+
+	wq->num_q_pages = num_q_pages;
+
+	err = alloc_wqes_shadow(wq);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to allocate wqe shadow\n");
+		return err;
+	}
+
+	for (i = 0; i < num_q_pages; i++) {
+		void **vaddr = &wq->shadow_block_vaddr[i];
+		u64 *paddr = &wq->block_vaddr[i];
+		dma_addr_t dma_addr;
+
+		*vaddr = dma_zalloc_coherent(&pdev->dev, wq->wq_page_size,
+					     &dma_addr, GFP_KERNEL);
+		if (!*vaddr) {
+			dev_err(&pdev->dev, "Failed to allocate wq page\n");
+			goto err_alloc_wq_pages;
+		}
+
+		/* HW uses Big Endian Format */
+		*paddr = cpu_to_be64(dma_addr);
+	}
+
+	return 0;
+
+err_alloc_wq_pages:
+	free_wq_pages(wq, hwif, i);
+	return -ENOMEM;
+}
+
+/**
+ * hinic_wq_allocate - Allocate the WQ resources from the WQS
+ * @wqs: WQ set from which to allocate the WQ resources
+ * @wq: WQ to allocate resources for it from the WQ set
+ * @wqebb_size: Work Queue Block Byte Size
+ * @wq_page_size: the page size in the Work Queue
+ * @q_depth: number of wqebbs in WQ
+ * @max_wqe_size: maximum WQE size that will be used in the WQ
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_wq_allocate(struct hinic_wqs *wqs, struct hinic_wq *wq,
+		      u16 wqebb_size, u16 wq_page_size, u16 q_depth,
+		      u16 max_wqe_size)
+{
+	struct hinic_hwif *hwif = wqs->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	u16 num_wqebbs_per_page;
+	int err;
+
+	if (wqebb_size == 0) {
+		dev_err(&pdev->dev, "wqebb_size must be > 0\n");
+		return -EINVAL;
+	}
+
+	if (wq_page_size == 0) {
+		dev_err(&pdev->dev, "wq_page_size must be > 0\n");
+		return -EINVAL;
+	}
+
+	if (q_depth & (q_depth - 1)) {
+		dev_err(&pdev->dev, "WQ q_depth must be power of 2\n");
+		return -EINVAL;
+	}
+
+	num_wqebbs_per_page = ALIGN(wq_page_size, wqebb_size) / wqebb_size;
+
+	if (num_wqebbs_per_page & (num_wqebbs_per_page - 1)) {
+		dev_err(&pdev->dev, "num wqebbs per page must be power of 2\n");
+		return -EINVAL;
+	}
+
+	wq->hwif = hwif;
+
+	err = wqs_next_block(wqs, &wq->page_idx, &wq->block_idx);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to get free wqs next block\n");
+		return err;
+	}
+
+	wq->wqebb_size = wqebb_size;
+	wq->wq_page_size = wq_page_size;
+	wq->q_depth = q_depth;
+	wq->max_wqe_size = max_wqe_size;
+	wq->num_wqebbs_per_page = num_wqebbs_per_page;
+
+	wq->block_vaddr = WQ_BASE_VADDR(wqs, wq);
+	wq->shadow_block_vaddr = WQ_BASE_ADDR(wqs, wq);
+	wq->block_paddr = WQ_BASE_PADDR(wqs, wq);
+
+	err = alloc_wq_pages(wq, wqs->hwif, WQ_MAX_PAGES);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to allocate wq pages\n");
+		goto err_alloc_wq_pages;
+	}
+
+	atomic_set(&wq->cons_idx, 0);
+	atomic_set(&wq->prod_idx, 0);
+	atomic_set(&wq->delta, q_depth);
+	wq->mask = q_depth - 1;
+
+	return 0;
+
+err_alloc_wq_pages:
+	wqs_return_block(wqs, wq->page_idx, wq->block_idx);
+	return err;
+}
+
+/**
+ * hinic_wq_free - Free the WQ resources to the WQS
+ * @wqs: WQ set to free the WQ resources to it
+ * @wq: WQ to free its resources to the WQ set resources
+ **/
+void hinic_wq_free(struct hinic_wqs *wqs, struct hinic_wq *wq)
+{
+	free_wq_pages(wq, wqs->hwif, wq->num_q_pages);
+
+	wqs_return_block(wqs, wq->page_idx, wq->block_idx);
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h
new file mode 100644
index 0000000..7c114da
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_wq.h
@@ -0,0 +1,86 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#ifndef HINIC_HW_WQ_H
+#define HINIC_HW_WQ_H
+
+#include <linux/types.h>
+#include <linux/semaphore.h>
+#include <linux/atomic.h>
+
+#include "hinic_hw_if.h"
+
+struct hinic_free_block {
+	int     page_idx;
+	int     block_idx;
+};
+
+struct hinic_wq {
+	struct hinic_hwif       *hwif;
+
+	int             page_idx;
+	int             block_idx;
+
+	u16             wqebb_size;
+	u16             wq_page_size;
+	u16             q_depth;
+	u16             max_wqe_size;
+	u16             num_wqebbs_per_page;
+
+	/* The addresses are 64 bit in the HW */
+	u64             block_paddr;
+	void            **shadow_block_vaddr;
+	u64             *block_vaddr;
+
+	int             num_q_pages;
+	u8              *shadow_wqe;
+	u16             *shadow_idx;
+
+	atomic_t        cons_idx;
+	atomic_t        prod_idx;
+	atomic_t        delta;
+	u16             mask;
+};
+
+struct hinic_wqs {
+	struct hinic_hwif       *hwif;
+	int                     num_pages;
+
+	/* The addresses are 64 bit in the HW */
+	u64                     *page_paddr;
+	u64                     **page_vaddr;
+	void                    ***shadow_page_vaddr;
+
+	struct hinic_free_block *free_blocks;
+	int                     alloc_blk_pos;
+	int                     return_blk_pos;
+	int                     num_free_blks;
+
+	/* Lock for getting a free block from the WQ set */
+	struct semaphore        alloc_blocks_lock;
+};
+
+int hinic_wqs_alloc(struct hinic_wqs *wqs, int num_wqs,
+		    struct hinic_hwif *hwif);
+
+void hinic_wqs_free(struct hinic_wqs *wqs);
+
+int hinic_wq_allocate(struct hinic_wqs *wqs, struct hinic_wq *wq,
+		      u16 wqebb_size, u16 wq_page_size, u16 q_depth,
+		      u16 max_wqe_size);
+
+void hinic_wq_free(struct hinic_wqs *wqs, struct hinic_wq *wq);
+
+#endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_wqe.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_wqe.h
new file mode 100644
index 0000000..d727c4d
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_wqe.h
@@ -0,0 +1,253 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#ifndef HINIC_HW_WQE_H
+#define HINIC_HW_WQE_H
+
+#include "hinic_common.h"
+
+#define HINIC_SQ_CTRL_BUFDESC_SECT_LEN_SHIFT    0
+#define HINIC_SQ_CTRL_TASKSECT_LEN_SHIFT        16
+#define HINIC_SQ_CTRL_DATA_FORMAT_SHIFT         22
+#define HINIC_SQ_CTRL_LEN_SHIFT                 29
+
+#define HINIC_SQ_CTRL_BUFDESC_SECT_LEN_MASK     0xFF
+#define HINIC_SQ_CTRL_TASKSECT_LEN_MASK         0x1F
+#define HINIC_SQ_CTRL_DATA_FORMAT_MASK          0x1
+#define HINIC_SQ_CTRL_LEN_MASK                  0x3
+
+#define HINIC_SQ_CTRL_QUEUE_INFO_MSS_SHIFT      13
+
+#define HINIC_SQ_CTRL_QUEUE_INFO_MSS_MASK       0x3FFF
+
+#define HINIC_SQ_CTRL_SET(val, member)          \
+		(((u32)(val) & HINIC_SQ_CTRL_##member##_MASK) \
+		 << HINIC_SQ_CTRL_##member##_SHIFT)
+
+#define HINIC_SQ_CTRL_GET(val, member)          \
+		(((val) >> HINIC_SQ_CTRL_##member##_SHIFT) \
+		 & HINIC_SQ_CTRL_##member##_MASK)
+
+#define HINIC_SQ_TASK_INFO0_L2HDR_LEN_SHIFT     0
+#define HINIC_SQ_TASK_INFO0_L4_OFFLOAD_SHIFT    8
+#define HINIC_SQ_TASK_INFO0_INNER_L3TYPE_SHIFT  10
+#define HINIC_SQ_TASK_INFO0_VLAN_OFFLOAD_SHIFT  12
+#define HINIC_SQ_TASK_INFO0_PARSE_FLAG_SHIFT    13
+/* 1 bit reserved */
+#define HINIC_SQ_TASK_INFO0_TSO_FLAG_SHIFT      15
+#define HINIC_SQ_TASK_INFO0_VLAN_TAG_SHIFT      16
+
+#define HINIC_SQ_TASK_INFO0_L2HDR_LEN_MASK      0xFF
+#define HINIC_SQ_TASK_INFO0_L4_OFFLOAD_MASK     0x3
+#define HINIC_SQ_TASK_INFO0_INNER_L3TYPE_MASK   0x3
+#define HINIC_SQ_TASK_INFO0_VLAN_OFFLOAD_MASK   0x1
+#define HINIC_SQ_TASK_INFO0_PARSE_FLAG_MASK     0x1
+/* 1 bit reserved */
+#define HINIC_SQ_TASK_INFO0_TSO_FLAG_MASK       0x1
+#define HINIC_SQ_TASK_INFO0_VLAN_TAG_MASK       0xFFFF
+
+#define HINIC_SQ_TASK_INFO0_SET(val, member)    \
+		(((u32)(val) & HINIC_SQ_TASK_INFO0_##member##_MASK) <<  \
+		 HINIC_SQ_TASK_INFO0_##member##_SHIFT)
+
+/* 8 bits reserved */
+#define HINIC_SQ_TASK_INFO1_MEDIA_TYPE_SHIFT    8
+#define HINIC_SQ_TASK_INFO1_INNER_L4_LEN_SHIFT  16
+#define HINIC_SQ_TASK_INFO1_INNER_L3_LEN_SHIFT  24
+
+/* 8 bits reserved */
+#define HINIC_SQ_TASK_INFO1_MEDIA_TYPE_MASK     0xFF
+#define HINIC_SQ_TASK_INFO1_INNER_L4_LEN_MASK   0xFF
+#define HINIC_SQ_TASK_INFO1_INNER_L3_LEN_MASK   0xFF
+
+#define HINIC_SQ_TASK_INFO1_SET(val, member)    \
+		(((u32)(val) & HINIC_SQ_TASK_INFO1_##member##_MASK) <<  \
+		 HINIC_SQ_TASK_INFO1_##member##_SHIFT)
+
+#define HINIC_SQ_TASK_INFO2_TUNNEL_L4_LEN_SHIFT 0
+#define HINIC_SQ_TASK_INFO2_OUTER_L3_LEN_SHIFT  12
+#define HINIC_SQ_TASK_INFO2_TUNNEL_L4TYPE_SHIFT 19
+/* 1 bit reserved */
+#define HINIC_SQ_TASK_INFO2_OUTER_L3TYPE_SHIFT  22
+/* 8 bits reserved */
+
+#define HINIC_SQ_TASK_INFO2_TUNNEL_L4_LEN_MASK  0xFFF
+#define HINIC_SQ_TASK_INFO2_OUTER_L3_LEN_MASK   0x7F
+#define HINIC_SQ_TASK_INFO2_TUNNEL_L4TYPE_MASK  0x3
+/* 1 bit reserved */
+#define HINIC_SQ_TASK_INFO2_OUTER_L3TYPE_MASK   0x3
+/* 8 bits reserved */
+
+#define HINIC_SQ_TASK_INFO2_SET(val, member)    \
+		(((u32)(val) & HINIC_SQ_TASK_INFO2_##member##_MASK) <<  \
+		 HINIC_SQ_TASK_INFO2_##member##_SHIFT)
+
+/* 31 bits reserved */
+#define HINIC_SQ_TASK_INFO4_L2TYPE_SHIFT        31
+
+/* 31 bits reserved */
+#define HINIC_SQ_TASK_INFO4_L2TYPE_MASK         0x1
+
+#define HINIC_SQ_TASK_INFO4_SET(val, member)    \
+		(((u32)(val) & HINIC_SQ_TASK_INFO4_##member##_MASK) << \
+		 HINIC_SQ_TASK_INFO4_##member##_SHIFT)
+
+#define HINIC_RQ_CQE_STATUS_RXDONE_SHIFT        31
+
+#define HINIC_RQ_CQE_STATUS_RXDONE_MASK         0x1
+
+#define HINIC_RQ_CQE_STATUS_GET(val, member)    \
+		(((val) >> HINIC_RQ_CQE_STATUS_##member##_SHIFT) & \
+		 HINIC_RQ_CQE_STATUS_##member##_MASK)
+
+#define HINIC_RQ_CQE_STATUS_CLEAR(val, member)  \
+		((val) & (~(HINIC_RQ_CQE_STATUS_##member##_MASK << \
+		 HINIC_RQ_CQE_STATUS_##member##_SHIFT)))
+
+#define HINIC_RQ_CQE_SGE_LEN_SHIFT              16
+
+#define HINIC_RQ_CQE_SGE_LEN_MASK               0xFFFF
+
+#define HINIC_RQ_CQE_SGE_GET(val, member)       \
+		(((val) >> HINIC_RQ_CQE_SGE_##member##_SHIFT) & \
+		 HINIC_RQ_CQE_SGE_##member##_MASK)
+
+#define HINIC_RQ_CTRL_BUFDESC_SECT_LEN_SHIFT    0
+#define HINIC_RQ_CTRL_COMPLETE_FORMAT_SHIFT     15
+#define HINIC_RQ_CTRL_COMPLETE_LEN_SHIFT        27
+#define HINIC_RQ_CTRL_LEN_SHIFT                 29
+
+#define HINIC_RQ_CTRL_BUFDESC_SECT_LEN_MASK     0xFF
+#define HINIC_RQ_CTRL_COMPLETE_FORMAT_MASK      0x1
+#define HINIC_RQ_CTRL_COMPLETE_LEN_MASK         0x3
+#define HINIC_RQ_CTRL_LEN_MASK                  0x3
+
+#define HINIC_RQ_CTRL_SET(val, member)          \
+		(((u32)(val) & HINIC_RQ_CTRL_##member##_MASK) << \
+		 HINIC_RQ_CTRL_##member##_SHIFT)
+
+#define HINIC_SQ_WQE_SIZE(nr_sges)              \
+		(sizeof(struct hinic_sq_ctrl) + \
+		 sizeof(struct hinic_sq_task) + \
+		 (nr_sges) * sizeof(struct hinic_sq_bufdesc))
+
+#define HINIC_MAX_SQ_BUFDESCS           17
+
+#define HINIC_SQ_WQE_MAX_SIZE           320
+#define HINIC_RQ_WQE_SIZE               32
+
+enum hinic_l4offload_type {
+	HINIC_L4_OFF_DISABLE            = 0,
+	HINIC_TCP_OFFLOAD_ENABLE        = 1,
+	HINIC_SCTP_OFFLOAD_ENABLE       = 2,
+	HINIC_UDP_OFFLOAD_ENABLE        = 3,
+};
+
+enum hinic_vlan_offload {
+	HINIC_VLAN_OFF_DISABLE = 0,
+	HINIC_VLAN_OFF_ENABLE  = 1,
+};
+
+enum hinic_pkt_parsed {
+	HINIC_PKT_NOT_PARSED = 0,
+	HINIC_PKT_PARSED     = 1,
+};
+
+enum hinic_outer_l3type {
+	HINIC_OUTER_L3TYPE_UNKNOWN              = 0,
+	HINIC_OUTER_L3TYPE_IPV6                 = 1,
+	HINIC_OUTER_L3TYPE_IPV4_NO_CHKSUM       = 2,
+	HINIC_OUTER_L3TYPE_IPV4_CHKSUM          = 3,
+};
+
+enum hinic_media_type {
+	HINIC_MEDIA_UNKNOWN = 0,
+};
+
+enum hinic_l2type {
+	HINIC_L2TYPE_ETH = 0,
+};
+
+enum hinc_tunnel_l4type {
+	HINIC_TUNNEL_L4TYPE_UNKNOWN = 0,
+};
+
+struct hinic_sq_ctrl {
+	u32     ctrl_info;
+	u32     queue_info;
+};
+
+struct hinic_sq_task {
+	u32     pkt_info0;
+	u32     pkt_info1;
+	u32     pkt_info2;
+	u32     ufo_v6_identify;
+	u32     pkt_info4;
+	u32     zero_pad;
+};
+
+struct hinic_sq_bufdesc {
+	struct hinic_sge sge;
+	u32     rsvd;
+};
+
+struct hinic_sq_wqe {
+	struct hinic_sq_ctrl            ctrl;
+	struct hinic_sq_task            task;
+	struct hinic_sq_bufdesc         buf_descs[HINIC_MAX_SQ_BUFDESCS];
+};
+
+struct hinic_rq_cqe {
+	u32     status;
+	u32     len;
+
+	u32     rsvd2;
+	u32     rsvd3;
+	u32     rsvd4;
+	u32     rsvd5;
+	u32     rsvd6;
+	u32     rsvd7;
+};
+
+struct hinic_rq_ctrl {
+	u32     ctrl_info;
+};
+
+struct hinic_rq_cqe_sect {
+	struct hinic_sge        sge;
+	u32                     rsvd;
+};
+
+struct hinic_rq_bufdesc {
+	u32     hi_addr;
+	u32     lo_addr;
+};
+
+struct hinic_rq_wqe {
+	struct hinic_rq_ctrl            ctrl;
+	u32                             rsvd;
+	struct hinic_rq_cqe_sect        cqe_sect;
+	struct hinic_rq_bufdesc         buf_desc;
+};
+
+struct hinic_hw_wqe {
+	/* HW Format */
+	union {
+		struct hinic_sq_wqe     sq_wqe;
+		struct hinic_rq_wqe     rq_wqe;
+	};
+};
+
+#endif
-- 
1.9.1

^ permalink raw reply related

* [PATCH V7 net-next 10/22] net-next/hinic: Add logical Txq and Rxq
From: Aviad Krawczyk @ 2017-08-17 19:04 UTC (permalink / raw)
  To: davem
  Cc: linux-kernel, netdev, bc.y, victor.gissin, aviad.krawczyk,
	zhaochen6, tony.qu
In-Reply-To: <cover.1502996306.git.aviad.krawczyk@huawei.com>

Create the logical queues of the nic.

Signed-off-by: Aviad Krawczyk <aviad.krawczyk@huawei.com>
Signed-off-by: Zhao Chen <zhaochen6@huawei.com>
---
 drivers/net/ethernet/huawei/hinic/Makefile       |   5 +-
 drivers/net/ethernet/huawei/hinic/hinic_dev.h    |   5 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c | 131 +++++++++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h |  20 +++
 drivers/net/ethernet/huawei/hinic/hinic_hw_io.c  | 144 +++++++++++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_io.h  |  46 ++++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h  |  32 +++++
 drivers/net/ethernet/huawei/hinic/hinic_main.c   | 172 ++++++++++++++++++++++-
 drivers/net/ethernet/huawei/hinic/hinic_rx.c     |  72 ++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_rx.h     |  46 ++++++
 drivers/net/ethernet/huawei/hinic/hinic_tx.c     |  75 ++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_tx.h     |  49 +++++++
 12 files changed, 793 insertions(+), 4 deletions(-)
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_hw_io.c
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_rx.c
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_rx.h
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_tx.c
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_tx.h

diff --git a/drivers/net/ethernet/huawei/hinic/Makefile b/drivers/net/ethernet/huawei/hinic/Makefile
index dbb1b9d..f60c449 100644
--- a/drivers/net/ethernet/huawei/hinic/Makefile
+++ b/drivers/net/ethernet/huawei/hinic/Makefile
@@ -1,4 +1,5 @@
 obj-$(CONFIG_HINIC) += hinic.o
 
-hinic-y := hinic_main.o hinic_port.o hinic_hw_dev.o hinic_hw_mgmt.o \
-	   hinic_hw_api_cmd.o hinic_hw_eqs.o hinic_hw_if.o
+hinic-y := hinic_main.o hinic_tx.o hinic_rx.o hinic_port.o hinic_hw_dev.o \
+	   hinic_hw_io.o hinic_hw_mgmt.o hinic_hw_api_cmd.o hinic_hw_eqs.o \
+	   hinic_hw_if.o
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_dev.h b/drivers/net/ethernet/huawei/hinic/hinic_dev.h
index 5c5b4e9..5b8231d 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_dev.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_dev.h
@@ -23,6 +23,8 @@
 #include <linux/bitops.h>
 
 #include "hinic_hw_dev.h"
+#include "hinic_tx.h"
+#include "hinic_rx.h"
 
 #define HINIC_DRV_NAME          "hinic"
 
@@ -49,6 +51,9 @@ struct hinic_dev {
 
 	struct hinic_rx_mode_work       rx_mode_work;
 	struct workqueue_struct         *workq;
+
+	struct hinic_txq                *txqs;
+	struct hinic_rxq                *rxqs;
 };
 
 #endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
index 75fd6d2..d113908 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
@@ -25,6 +25,8 @@
 #include "hinic_hw_if.h"
 #include "hinic_hw_eqs.h"
 #include "hinic_hw_mgmt.h"
+#include "hinic_hw_qp.h"
+#include "hinic_hw_io.h"
 #include "hinic_hw_dev.h"
 
 #define MAX_IRQS(max_qps, num_aeqs, num_ceqs)   \
@@ -230,6 +232,99 @@ int hinic_port_msg_cmd(struct hinic_hwdev *hwdev, enum hinic_port_cmd cmd,
 }
 
 /**
+ * get_base_qpn - get the first qp number
+ * @hwdev: the NIC HW device
+ * @base_qpn: returned qp number
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int get_base_qpn(struct hinic_hwdev *hwdev, u16 *base_qpn)
+{
+	struct hinic_cmd_base_qpn cmd_base_qpn;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	u16 out_size;
+	int err;
+
+	cmd_base_qpn.func_idx = HINIC_HWIF_FUNC_IDX(hwif);
+
+	err = hinic_port_msg_cmd(hwdev, HINIC_PORT_CMD_GET_GLOBAL_QPN,
+				 &cmd_base_qpn, sizeof(cmd_base_qpn),
+				 &cmd_base_qpn, &out_size);
+	if (err || (out_size != sizeof(cmd_base_qpn)) || cmd_base_qpn.status) {
+		dev_err(&pdev->dev, "Failed to get base qpn, status = %d\n",
+			cmd_base_qpn.status);
+		return -EFAULT;
+	}
+
+	*base_qpn = cmd_base_qpn.qpn;
+	return 0;
+}
+
+/**
+ * hinic_hwdev_ifup - Preparing the HW for passing IO
+ * @hwdev: the NIC HW device
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_hwdev_ifup(struct hinic_hwdev *hwdev)
+{
+	struct hinic_func_to_io *func_to_io = &hwdev->func_to_io;
+	struct hinic_cap *nic_cap = &hwdev->nic_cap;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	int err, num_aeqs, num_ceqs, num_qps;
+	struct msix_entry *sq_msix_entries;
+	struct msix_entry *rq_msix_entries;
+	struct pci_dev *pdev = hwif->pdev;
+	u16 base_qpn;
+
+	err = get_base_qpn(hwdev, &base_qpn);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to get global base qp number\n");
+		return err;
+	}
+
+	num_aeqs = HINIC_HWIF_NUM_AEQS(hwif);
+	num_ceqs = HINIC_HWIF_NUM_CEQS(hwif);
+	err = hinic_io_init(func_to_io, hwif, nic_cap->max_qps, 0, NULL);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to init IO channel\n");
+		return err;
+	}
+
+	num_qps = nic_cap->num_qps;
+	sq_msix_entries = &hwdev->msix_entries[num_aeqs + num_ceqs];
+	rq_msix_entries = &hwdev->msix_entries[num_aeqs + num_ceqs + num_qps];
+
+	err = hinic_io_create_qps(func_to_io, base_qpn, num_qps,
+				  sq_msix_entries, rq_msix_entries);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to create QPs\n");
+		goto err_create_qps;
+	}
+
+	return 0;
+
+err_create_qps:
+	hinic_io_free(func_to_io);
+	return err;
+}
+
+/**
+ * hinic_hwdev_ifdown - Closing the HW for passing IO
+ * @hwdev: the NIC HW device
+ *
+ **/
+void hinic_hwdev_ifdown(struct hinic_hwdev *hwdev)
+{
+	struct hinic_func_to_io *func_to_io = &hwdev->func_to_io;
+	struct hinic_cap *nic_cap = &hwdev->nic_cap;
+
+	hinic_io_destroy_qps(func_to_io, nic_cap->num_qps);
+	hinic_io_free(func_to_io);
+}
+
+/**
  * hinic_hwdev_cb_register - register callback handler for MGMT events
  * @hwdev: the NIC HW device
  * @cmd: the mgmt event
@@ -495,3 +590,39 @@ int hinic_hwdev_num_qps(struct hinic_hwdev *hwdev)
 
 	return nic_cap->num_qps;
 }
+
+/**
+ * hinic_hwdev_get_sq - get SQ
+ * @hwdev: the NIC HW device
+ * @i: the position of the SQ
+ *
+ * Return: the SQ in the i position
+ **/
+struct hinic_sq *hinic_hwdev_get_sq(struct hinic_hwdev *hwdev, int i)
+{
+	struct hinic_func_to_io *func_to_io = &hwdev->func_to_io;
+	struct hinic_qp *qp = &func_to_io->qps[i];
+
+	if (i >= hinic_hwdev_num_qps(hwdev))
+		return NULL;
+
+	return &qp->sq;
+}
+
+/**
+ * hinic_hwdev_get_sq - get RQ
+ * @hwdev: the NIC HW device
+ * @i: the position of the RQ
+ *
+ * Return: the RQ in the i position
+ **/
+struct hinic_rq *hinic_hwdev_get_rq(struct hinic_hwdev *hwdev, int i)
+{
+	struct hinic_func_to_io *func_to_io = &hwdev->func_to_io;
+	struct hinic_qp *qp = &func_to_io->qps[i];
+
+	if (i >= hinic_hwdev_num_qps(hwdev))
+		return NULL;
+
+	return &qp->rq;
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h
index 1cd8159..81c2c6e 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h
@@ -23,6 +23,8 @@
 #include "hinic_hw_if.h"
 #include "hinic_hw_eqs.h"
 #include "hinic_hw_mgmt.h"
+#include "hinic_hw_qp.h"
+#include "hinic_hw_io.h"
 
 #define HINIC_MAX_QPS   32
 
@@ -72,11 +74,21 @@ enum hinic_cb_state {
 	HINIC_CB_RUNNING = BIT(1),
 };
 
+struct hinic_cmd_base_qpn {
+	u8      status;
+	u8      version;
+	u8      rsvd0[6];
+
+	u16     func_idx;
+	u16     qpn;
+};
+
 struct hinic_hwdev {
 	struct hinic_hwif               *hwif;
 	struct msix_entry               *msix_entries;
 
 	struct hinic_aeqs               aeqs;
+	struct hinic_func_to_io         func_to_io;
 
 	struct hinic_cap                nic_cap;
 };
@@ -111,10 +123,18 @@ int hinic_port_msg_cmd(struct hinic_hwdev *hwdev, enum hinic_port_cmd cmd,
 		       void *buf_in, u16 in_size, void *buf_out,
 		       u16 *out_size);
 
+int hinic_hwdev_ifup(struct hinic_hwdev *hwdev);
+
+void hinic_hwdev_ifdown(struct hinic_hwdev *hwdev);
+
 struct hinic_hwdev *hinic_init_hwdev(struct pci_dev *pdev);
 
 void hinic_free_hwdev(struct hinic_hwdev *hwdev);
 
 int hinic_hwdev_num_qps(struct hinic_hwdev *hwdev);
 
+struct hinic_sq *hinic_hwdev_get_sq(struct hinic_hwdev *hwdev, int i);
+
+struct hinic_rq *hinic_hwdev_get_rq(struct hinic_hwdev *hwdev, int i);
+
 #endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c
new file mode 100644
index 0000000..ebe28ee
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.c
@@ -0,0 +1,144 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#include <linux/types.h>
+#include <linux/pci.h>
+#include <linux/device.h>
+#include <linux/errno.h>
+#include <linux/slab.h>
+
+#include "hinic_hw_if.h"
+#include "hinic_hw_qp.h"
+#include "hinic_hw_io.h"
+
+/**
+ * init_qp - Initialize a Queue Pair
+ * @func_to_io: func to io channel that holds the IO components
+ * @qp: pointer to the qp to initialize
+ * @q_id: the id of the qp
+ * @sq_msix_entry: msix entry for sq
+ * @rq_msix_entry: msix entry for rq
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int init_qp(struct hinic_func_to_io *func_to_io,
+		   struct hinic_qp *qp, int q_id,
+		   struct msix_entry *sq_msix_entry,
+		   struct msix_entry *rq_msix_entry)
+{
+	/* should be implemented */
+	return 0;
+}
+
+/**
+ * destroy_qp - Clean the resources of a Queue Pair
+ * @func_to_io: func to io channel that holds the IO components
+ * @qp: pointer to the qp to clean
+ **/
+static void destroy_qp(struct hinic_func_to_io *func_to_io,
+		       struct hinic_qp *qp)
+{
+	/* should be implemented */
+}
+
+/**
+ * hinic_io_create_qps - Create Queue Pairs
+ * @func_to_io: func to io channel that holds the IO components
+ * @base_qpn: base qp number
+ * @num_qps: number queue pairs to create
+ * @sq_msix_entry: msix entries for sq
+ * @rq_msix_entry: msix entries for rq
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_io_create_qps(struct hinic_func_to_io *func_to_io,
+			u16 base_qpn, int num_qps,
+			struct msix_entry *sq_msix_entries,
+			struct msix_entry *rq_msix_entries)
+{
+	struct hinic_hwif *hwif = func_to_io->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	size_t qps_size;
+	int i, j, err;
+
+	qps_size = num_qps * sizeof(*func_to_io->qps);
+	func_to_io->qps = devm_kzalloc(&pdev->dev, qps_size, GFP_KERNEL);
+	if (!func_to_io->qps)
+		return -ENOMEM;
+
+	for (i = 0; i < num_qps; i++) {
+		err = init_qp(func_to_io, &func_to_io->qps[i], i,
+			      &sq_msix_entries[i], &rq_msix_entries[i]);
+		if (err) {
+			dev_err(&pdev->dev, "Failed to create QP %d\n", i);
+			goto err_init_qp;
+		}
+	}
+
+	return 0;
+
+err_init_qp:
+	for (j = 0; j < i; j++)
+		destroy_qp(func_to_io, &func_to_io->qps[j]);
+
+	devm_kfree(&pdev->dev, func_to_io->qps);
+	return err;
+}
+
+/**
+ * hinic_io_destroy_qps - Destroy the IO Queue Pairs
+ * @func_to_io: func to io channel that holds the IO components
+ * @num_qps: number queue pairs to destroy
+ **/
+void hinic_io_destroy_qps(struct hinic_func_to_io *func_to_io, int num_qps)
+{
+	struct hinic_hwif *hwif = func_to_io->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	int i;
+
+	for (i = 0; i < num_qps; i++)
+		destroy_qp(func_to_io, &func_to_io->qps[i]);
+
+	devm_kfree(&pdev->dev, func_to_io->qps);
+}
+
+/**
+ * hinic_io_init - Initialize the IO components
+ * @func_to_io: func to io channel that holds the IO components
+ * @hwif: HW interface for accessing IO
+ * @max_qps: maximum QPs in HW
+ * @num_ceqs: number completion event queues
+ * @ceq_msix_entries: msix entries for ceqs
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_io_init(struct hinic_func_to_io *func_to_io,
+		  struct hinic_hwif *hwif, u16 max_qps, int num_ceqs,
+		  struct msix_entry *ceq_msix_entries)
+{
+	func_to_io->hwif = hwif;
+	func_to_io->qps = NULL;
+	func_to_io->max_qps = max_qps;
+
+	return 0;
+}
+
+/**
+ * hinic_io_free - Free the IO components
+ * @func_to_io: func to io channel that holds the IO components
+ **/
+void hinic_io_free(struct hinic_func_to_io *func_to_io)
+{
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
new file mode 100644
index 0000000..7cdcffd
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_io.h
@@ -0,0 +1,46 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#ifndef HINIC_HW_IO_H
+#define HINIC_HW_IO_H
+
+#include <linux/types.h>
+#include <linux/pci.h>
+
+#include "hinic_hw_if.h"
+#include "hinic_hw_qp.h"
+
+struct hinic_func_to_io {
+	struct hinic_hwif       *hwif;
+
+	struct hinic_qp         *qps;
+	u16                     max_qps;
+};
+
+int hinic_io_create_qps(struct hinic_func_to_io *func_to_io,
+			u16 base_qpn, int num_qps,
+			struct msix_entry *sq_msix_entries,
+			struct msix_entry *rq_msix_entries);
+
+void hinic_io_destroy_qps(struct hinic_func_to_io *func_to_io,
+			  int num_qps);
+
+int hinic_io_init(struct hinic_func_to_io *func_to_io,
+		  struct hinic_hwif *hwif, u16 max_qps, int num_ceqs,
+		  struct msix_entry *ceq_msix_entries);
+
+void hinic_io_free(struct hinic_func_to_io *func_to_io);
+
+#endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
new file mode 100644
index 0000000..64330fb
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_qp.h
@@ -0,0 +1,32 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#ifndef HINIC_HW_QP_H
+#define HINIC_HW_QP_H
+
+struct hinic_sq {
+	/* should be implemented */
+};
+
+struct hinic_rq {
+	/* should be implemented */
+};
+
+struct hinic_qp {
+	struct hinic_sq         sq;
+	struct hinic_rq         rq;
+};
+
+#endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_main.c b/drivers/net/ethernet/huawei/hinic/hinic_main.c
index 3fe8a6f..914ce52 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_main.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_main.c
@@ -31,8 +31,11 @@
 #include <linux/delay.h>
 #include <linux/err.h>
 
+#include "hinic_hw_qp.h"
 #include "hinic_hw_dev.h"
 #include "hinic_port.h"
+#include "hinic_tx.h"
+#include "hinic_rx.h"
 #include "hinic_dev.h"
 
 MODULE_AUTHOR("Huawei Technologies CO., Ltd");
@@ -57,17 +60,164 @@
 
 static int change_mac_addr(struct net_device *netdev, const u8 *addr);
 
+/**
+ * create_txqs - Create the Logical Tx Queues of specific NIC device
+ * @nic_dev: the specific NIC device
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int create_txqs(struct hinic_dev *nic_dev)
+{
+	int err, i, j, num_txqs = hinic_hwdev_num_qps(nic_dev->hwdev);
+	struct net_device *netdev = nic_dev->netdev;
+	size_t txq_size;
+
+	if (nic_dev->txqs)
+		return -EINVAL;
+
+	txq_size = num_txqs * sizeof(*nic_dev->txqs);
+	nic_dev->txqs = devm_kzalloc(&netdev->dev, txq_size, GFP_KERNEL);
+	if (!nic_dev->txqs)
+		return -ENOMEM;
+
+	for (i = 0; i < num_txqs; i++) {
+		struct hinic_sq *sq = hinic_hwdev_get_sq(nic_dev->hwdev, i);
+
+		err = hinic_init_txq(&nic_dev->txqs[i], sq, netdev);
+		if (err) {
+			netif_err(nic_dev, drv, netdev,
+				  "Failed to init Txq\n");
+			goto err_init_txq;
+		}
+	}
+
+	return 0;
+
+err_init_txq:
+	for (j = 0; j < i; j++)
+		hinic_clean_txq(&nic_dev->txqs[j]);
+
+	devm_kfree(&netdev->dev, nic_dev->txqs);
+	return err;
+}
+
+/**
+ * free_txqs - Free the Logical Tx Queues of specific NIC device
+ * @nic_dev: the specific NIC device
+ **/
+static void free_txqs(struct hinic_dev *nic_dev)
+{
+	int i, num_txqs = hinic_hwdev_num_qps(nic_dev->hwdev);
+	struct net_device *netdev = nic_dev->netdev;
+
+	if (!nic_dev->txqs)
+		return;
+
+	for (i = 0; i < num_txqs; i++)
+		hinic_clean_txq(&nic_dev->txqs[i]);
+
+	devm_kfree(&netdev->dev, nic_dev->txqs);
+	nic_dev->txqs = NULL;
+}
+
+/**
+ * create_txqs - Create the Logical Rx Queues of specific NIC device
+ * @nic_dev: the specific NIC device
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int create_rxqs(struct hinic_dev *nic_dev)
+{
+	int err, i, j, num_rxqs = hinic_hwdev_num_qps(nic_dev->hwdev);
+	struct net_device *netdev = nic_dev->netdev;
+	size_t rxq_size;
+
+	if (nic_dev->rxqs)
+		return -EINVAL;
+
+	rxq_size = num_rxqs * sizeof(*nic_dev->rxqs);
+	nic_dev->rxqs = devm_kzalloc(&netdev->dev, rxq_size, GFP_KERNEL);
+	if (!nic_dev->rxqs)
+		return -ENOMEM;
+
+	for (i = 0; i < num_rxqs; i++) {
+		struct hinic_rq *rq = hinic_hwdev_get_rq(nic_dev->hwdev, i);
+
+		err = hinic_init_rxq(&nic_dev->rxqs[i], rq, netdev);
+		if (err) {
+			netif_err(nic_dev, drv, netdev,
+				  "Failed to init rxq\n");
+			goto err_init_rxq;
+		}
+	}
+
+	return 0;
+
+err_init_rxq:
+	for (j = 0; j < i; j++)
+		hinic_clean_rxq(&nic_dev->rxqs[j]);
+
+	devm_kfree(&netdev->dev, nic_dev->rxqs);
+	return err;
+}
+
+/**
+ * free_txqs - Free the Logical Rx Queues of specific NIC device
+ * @nic_dev: the specific NIC device
+ **/
+static void free_rxqs(struct hinic_dev *nic_dev)
+{
+	int i, num_rxqs = hinic_hwdev_num_qps(nic_dev->hwdev);
+	struct net_device *netdev = nic_dev->netdev;
+
+	if (!nic_dev->rxqs)
+		return;
+
+	for (i = 0; i < num_rxqs; i++)
+		hinic_clean_rxq(&nic_dev->rxqs[i]);
+
+	devm_kfree(&netdev->dev, nic_dev->rxqs);
+	nic_dev->rxqs = NULL;
+}
+
 static int hinic_open(struct net_device *netdev)
 {
 	struct hinic_dev *nic_dev = netdev_priv(netdev);
 	enum hinic_port_link_state link_state;
-	int err, ret;
+	int err, ret, num_qps;
+
+	if (!(nic_dev->flags & HINIC_INTF_UP)) {
+		err = hinic_hwdev_ifup(nic_dev->hwdev);
+		if (err) {
+			netif_err(nic_dev, drv, netdev,
+				  "Failed - HW interface up\n");
+			return err;
+		}
+	}
+
+	err = create_txqs(nic_dev);
+	if (err) {
+		netif_err(nic_dev, drv, netdev,
+			  "Failed to create Tx queues\n");
+		goto err_create_txqs;
+	}
+
+	err = create_rxqs(nic_dev);
+	if (err) {
+		netif_err(nic_dev, drv, netdev,
+			  "Failed to create Rx queues\n");
+		goto err_create_rxqs;
+	}
+
+	num_qps = hinic_hwdev_num_qps(nic_dev->hwdev);
+	netif_set_real_num_tx_queues(netdev, num_qps);
+	netif_set_real_num_rx_queues(netdev, num_qps);
 
 	err = hinic_port_set_state(nic_dev, HINIC_PORT_ENABLE);
 	if (err) {
 		netif_err(nic_dev, drv, netdev,
 			  "Failed to set port state\n");
-		return err;
+		goto err_port_state;
 	}
 
 	/* Wait up to 3 sec between port enable to link state */
@@ -104,6 +254,16 @@ static int hinic_open(struct net_device *netdev)
 	if (ret)
 		netif_warn(nic_dev, drv, netdev,
 			   "Failed to revert port state\n");
+
+err_port_state:
+	free_rxqs(nic_dev);
+
+err_create_rxqs:
+	free_txqs(nic_dev);
+
+err_create_txqs:
+	if (!(nic_dev->flags & HINIC_INTF_UP))
+		hinic_hwdev_ifdown(nic_dev->hwdev);
 	return err;
 }
 
@@ -130,6 +290,12 @@ static int hinic_close(struct net_device *netdev)
 		return err;
 	}
 
+	free_rxqs(nic_dev);
+	free_txqs(nic_dev);
+
+	if (flags & HINIC_INTF_UP)
+		hinic_hwdev_ifdown(nic_dev->hwdev);
+
 	netif_info(nic_dev, drv, netdev, "HINIC_INTF is DOWN\n");
 	return 0;
 }
@@ -496,6 +662,8 @@ static int nic_dev_init(struct pci_dev *pdev)
 	nic_dev->hwdev = hwdev;
 	nic_dev->msg_enable = MSG_ENABLE_DEFAULT;
 	nic_dev->flags = 0;
+	nic_dev->txqs = NULL;
+	nic_dev->rxqs = NULL;
 
 	sema_init(&nic_dev->mgmt_lock, 1);
 
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_rx.c b/drivers/net/ethernet/huawei/hinic/hinic_rx.c
new file mode 100644
index 0000000..173fe8b
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_rx.c
@@ -0,0 +1,72 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#include <linux/netdevice.h>
+#include <linux/u64_stats_sync.h>
+
+#include "hinic_hw_qp.h"
+#include "hinic_rx.h"
+
+/**
+ * hinic_rxq_clean_stats - Clean the statistics of specific queue
+ * @rxq: Logical Rx Queue
+ **/
+void hinic_rxq_clean_stats(struct hinic_rxq *rxq)
+{
+	struct hinic_rxq_stats *rxq_stats = &rxq->rxq_stats;
+
+	u64_stats_update_begin(&rxq_stats->syncp);
+	rxq_stats->pkts = 0;
+	rxq_stats->bytes = 0;
+	u64_stats_update_end(&rxq_stats->syncp);
+}
+
+/**
+ * rxq_stats_init - Initialize the statistics of specific queue
+ * @rxq: Logical Rx Queue
+ **/
+static void rxq_stats_init(struct hinic_rxq *rxq)
+{
+	struct hinic_rxq_stats *rxq_stats = &rxq->rxq_stats;
+
+	u64_stats_init(&rxq_stats->syncp);
+	hinic_rxq_clean_stats(rxq);
+}
+
+/**
+ * hinic_init_rxq - Initialize the Rx Queue
+ * @rxq: Logical Rx Queue
+ * @rq: Hardware Rx Queue to connect the Logical queue with
+ * @netdev: network device to connect the Logical queue with
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_init_rxq(struct hinic_rxq *rxq, struct hinic_rq *rq,
+		   struct net_device *netdev)
+{
+	rxq->netdev = netdev;
+	rxq->rq = rq;
+
+	rxq_stats_init(rxq);
+	return 0;
+}
+
+/**
+ * hinic_clean_rxq - Clean the Rx Queue
+ * @rxq: Logical Rx Queue
+ **/
+void hinic_clean_rxq(struct hinic_rxq *rxq)
+{
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_rx.h b/drivers/net/ethernet/huawei/hinic/hinic_rx.h
new file mode 100644
index 0000000..fbd0246
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_rx.h
@@ -0,0 +1,46 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#ifndef HINIC_RX_H
+#define HINIC_RX_H
+
+#include <linux/types.h>
+#include <linux/netdevice.h>
+#include <linux/u64_stats_sync.h>
+
+#include "hinic_hw_qp.h"
+
+struct hinic_rxq_stats {
+	u64                     pkts;
+	u64                     bytes;
+
+	struct u64_stats_sync   syncp;
+};
+
+struct hinic_rxq {
+	struct net_device       *netdev;
+	struct hinic_rq         *rq;
+
+	struct hinic_rxq_stats  rxq_stats;
+};
+
+void hinic_rxq_clean_stats(struct hinic_rxq *rxq);
+
+int hinic_init_rxq(struct hinic_rxq *rxq, struct hinic_rq *rq,
+		   struct net_device *netdev);
+
+void hinic_clean_rxq(struct hinic_rxq *rxq);
+
+#endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_tx.c b/drivers/net/ethernet/huawei/hinic/hinic_tx.c
new file mode 100644
index 0000000..9c27fb2
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_tx.c
@@ -0,0 +1,75 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#include <linux/netdevice.h>
+#include <linux/u64_stats_sync.h>
+
+#include "hinic_hw_qp.h"
+#include "hinic_tx.h"
+
+/**
+ * hinic_txq_clean_stats - Clean the statistics of specific queue
+ * @txq: Logical Tx Queue
+ **/
+void hinic_txq_clean_stats(struct hinic_txq *txq)
+{
+	struct hinic_txq_stats *txq_stats = &txq->txq_stats;
+
+	u64_stats_update_begin(&txq_stats->syncp);
+	txq_stats->pkts = 0;
+	txq_stats->bytes = 0;
+	txq_stats->tx_busy = 0;
+	txq_stats->tx_wake = 0;
+	txq_stats->tx_dropped = 0;
+	u64_stats_update_end(&txq_stats->syncp);
+}
+
+/**
+ * txq_stats_init - Initialize the statistics of specific queue
+ * @txq: Logical Tx Queue
+ **/
+static void txq_stats_init(struct hinic_txq *txq)
+{
+	struct hinic_txq_stats *txq_stats = &txq->txq_stats;
+
+	u64_stats_init(&txq_stats->syncp);
+	hinic_txq_clean_stats(txq);
+}
+
+/**
+ * hinic_init_txq - Initialize the Tx Queue
+ * @txq: Logical Tx Queue
+ * @sq: Hardware Tx Queue to connect the Logical queue with
+ * @netdev: network device to connect the Logical queue with
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_init_txq(struct hinic_txq *txq, struct hinic_sq *sq,
+		   struct net_device *netdev)
+{
+	txq->netdev = netdev;
+	txq->sq = sq;
+
+	txq_stats_init(txq);
+	return 0;
+}
+
+/**
+ * hinic_clean_txq - Clean the Tx Queue
+ * @txq: Logical Tx Queue
+ **/
+void hinic_clean_txq(struct hinic_txq *txq)
+{
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_tx.h b/drivers/net/ethernet/huawei/hinic/hinic_tx.h
new file mode 100644
index 0000000..bbdb4b6
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_tx.h
@@ -0,0 +1,49 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#ifndef HINIC_TX_H
+#define HINIC_TX_H
+
+#include <linux/types.h>
+#include <linux/netdevice.h>
+#include <linux/u64_stats_sync.h>
+
+#include "hinic_hw_qp.h"
+
+struct hinic_txq_stats {
+	u64     pkts;
+	u64     bytes;
+	u64     tx_busy;
+	u64     tx_wake;
+	u64     tx_dropped;
+
+	struct u64_stats_sync   syncp;
+};
+
+struct hinic_txq {
+	struct net_device       *netdev;
+	struct hinic_sq         *sq;
+
+	struct hinic_txq_stats  txq_stats;
+};
+
+void hinic_txq_clean_stats(struct hinic_txq *txq);
+
+int hinic_init_txq(struct hinic_txq *txq, struct hinic_sq *sq,
+		   struct net_device *netdev);
+
+void hinic_clean_txq(struct hinic_txq *txq);
+
+#endif
-- 
1.9.1

^ permalink raw reply related

* [PATCH V7 net-next 09/22] net-next/hinic: Add Rx mode and link event handler
From: Aviad Krawczyk @ 2017-08-17 19:04 UTC (permalink / raw)
  To: davem
  Cc: linux-kernel, netdev, bc.y, victor.gissin, aviad.krawczyk,
	zhaochen6, tony.qu
In-Reply-To: <cover.1502996306.git.aviad.krawczyk@huawei.com>

Add port management message for setting Rx mode in the card,
used for rx_mode netdev operation.
The link event handler is used for getting a notification about the
link state.

Signed-off-by: Aviad Krawczyk <aviad.krawczyk@huawei.com>
Signed-off-by: Zhao Chen <zhaochen6@huawei.com>
---
 drivers/net/ethernet/huawei/hinic/hinic_dev.h     |  17 ++
 drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h  |   2 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c  | 118 +++++++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h  |  37 +++
 drivers/net/ethernet/huawei/hinic/hinic_hw_if.c   |  17 ++
 drivers/net/ethernet/huawei/hinic/hinic_hw_if.h   |  17 ++
 drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.c |  64 ++++-
 drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.h |  28 +++
 drivers/net/ethernet/huawei/hinic/hinic_main.c    | 284 ++++++++++++++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_port.c    |  92 +++++++
 drivers/net/ethernet/huawei/hinic/hinic_port.h    |  66 +++++
 11 files changed, 741 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/huawei/hinic/hinic_dev.h b/drivers/net/ethernet/huawei/hinic/hinic_dev.h
index e54a45c..5c5b4e9 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_dev.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_dev.h
@@ -19,19 +19,36 @@
 #include <linux/netdevice.h>
 #include <linux/types.h>
 #include <linux/semaphore.h>
+#include <linux/workqueue.h>
+#include <linux/bitops.h>
 
 #include "hinic_hw_dev.h"
 
 #define HINIC_DRV_NAME          "hinic"
 
+enum hinic_flags {
+	HINIC_LINK_UP = BIT(0),
+	HINIC_INTF_UP = BIT(1),
+};
+
+struct hinic_rx_mode_work {
+	struct work_struct      work;
+	u32                     rx_mode;
+};
+
 struct hinic_dev {
 	struct net_device               *netdev;
 	struct hinic_hwdev              *hwdev;
 
 	u32                             msg_enable;
 
+	unsigned int                    flags;
+
 	struct semaphore                mgmt_lock;
 	unsigned long                   *vlan_bitmap;
+
+	struct hinic_rx_mode_work       rx_mode_work;
+	struct workqueue_struct         *workq;
 };
 
 #endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h
index 52eb89c..1f57301 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h
@@ -20,6 +20,8 @@
 #define HINIC_CSR_FUNC_ATTR0_ADDR                       0x0
 #define HINIC_CSR_FUNC_ATTR1_ADDR                       0x4
 
+#define HINIC_CSR_FUNC_ATTR5_ADDR                       0x14
+
 #define HINIC_DMA_ATTR_BASE                             0xC80
 #define HINIC_ELECTION_BASE                             0x4200
 
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
index 6bb6c33..75fd6d2 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
@@ -230,6 +230,114 @@ int hinic_port_msg_cmd(struct hinic_hwdev *hwdev, enum hinic_port_cmd cmd,
 }
 
 /**
+ * hinic_hwdev_cb_register - register callback handler for MGMT events
+ * @hwdev: the NIC HW device
+ * @cmd: the mgmt event
+ * @handle: private data for the handler
+ * @handler: event handler
+ **/
+void hinic_hwdev_cb_register(struct hinic_hwdev *hwdev,
+			     enum hinic_mgmt_msg_cmd cmd, void *handle,
+			     void (*handler)(void *handle, void *buf_in,
+					     u16 in_size, void *buf_out,
+					     u16 *out_size))
+{
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_pfhwdev *pfhwdev;
+	struct hinic_nic_cb *nic_cb;
+	u8 cmd_cb;
+
+	if (!HINIC_IS_PF(hwif) && !HINIC_IS_PPF(hwif)) {
+		dev_err(&pdev->dev, "unsupported PCI Function type\n");
+		return;
+	}
+
+	pfhwdev = container_of(hwdev, struct hinic_pfhwdev, hwdev);
+
+	cmd_cb = cmd - HINIC_MGMT_MSG_CMD_BASE;
+	nic_cb = &pfhwdev->nic_cb[cmd_cb];
+
+	nic_cb->handler = handler;
+	nic_cb->handle = handle;
+	nic_cb->cb_state = HINIC_CB_ENABLED;
+}
+
+/**
+ * hinic_hwdev_cb_unregister - unregister callback handler for MGMT events
+ * @hwdev: the NIC HW device
+ * @cmd: the mgmt event
+ **/
+void hinic_hwdev_cb_unregister(struct hinic_hwdev *hwdev,
+			       enum hinic_mgmt_msg_cmd cmd)
+{
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_pfhwdev *pfhwdev;
+	struct hinic_nic_cb *nic_cb;
+	u8 cmd_cb;
+
+	if (!HINIC_IS_PF(hwif) && !HINIC_IS_PPF(hwif)) {
+		dev_err(&pdev->dev, "unsupported PCI Function type\n");
+		return;
+	}
+
+	pfhwdev = container_of(hwdev, struct hinic_pfhwdev, hwdev);
+
+	cmd_cb = cmd - HINIC_MGMT_MSG_CMD_BASE;
+	nic_cb = &pfhwdev->nic_cb[cmd_cb];
+
+	nic_cb->cb_state &= ~HINIC_CB_ENABLED;
+
+	while (nic_cb->cb_state & HINIC_CB_RUNNING)
+		schedule();
+
+	nic_cb->handler = NULL;
+}
+
+/**
+ * nic_mgmt_msg_handler - nic mgmt event handler
+ * @handle: private data for the handler
+ * @buf_in: input buffer
+ * @in_size: input size
+ * @buf_out: output buffer
+ * @out_size: returned output size
+ **/
+static void nic_mgmt_msg_handler(void *handle, u8 cmd, void *buf_in,
+				 u16 in_size, void *buf_out, u16 *out_size)
+{
+	struct hinic_pfhwdev *pfhwdev = (struct hinic_pfhwdev *)handle;
+	struct hinic_hwdev *hwdev = &pfhwdev->hwdev;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_nic_cb *nic_cb;
+	enum hinic_cb_state cb_state;
+	u8 cmd_cb;
+
+	if ((cmd < HINIC_MGMT_MSG_CMD_BASE) ||
+	    (cmd >= HINIC_MGMT_MSG_CMD_MAX)) {
+		dev_err(&pdev->dev, "unknown L2NIC event, cmd = %d\n", cmd);
+		return;
+	}
+
+	cmd_cb = (enum hinic_mgmt_msg_cmd)cmd - HINIC_MGMT_MSG_CMD_BASE;
+
+	nic_cb = &pfhwdev->nic_cb[cmd_cb];
+
+	cb_state = cmpxchg(&nic_cb->cb_state,
+			   HINIC_CB_ENABLED,
+			   HINIC_CB_ENABLED | HINIC_CB_RUNNING);
+
+	if ((cb_state == HINIC_CB_ENABLED) && (nic_cb->handler))
+		nic_cb->handler(nic_cb->handle, buf_in,
+				in_size, buf_out, out_size);
+	else
+		dev_err(&pdev->dev, "Unhandled NIC Event %d\n", cmd);
+
+	nic_cb->cb_state &= ~HINIC_CB_RUNNING;
+}
+
+/**
  * init_pfhwdev - Initialize the extended components of PF
  * @pfhwdev: the HW device for PF
  *
@@ -248,6 +356,10 @@ static int init_pfhwdev(struct hinic_pfhwdev *pfhwdev)
 		return err;
 	}
 
+	hinic_register_mgmt_msg_cb(&pfhwdev->pf_to_mgmt, HINIC_MOD_L2NIC,
+				   pfhwdev, nic_mgmt_msg_handler);
+
+	hinic_set_pf_action(hwif, HINIC_PF_MGMT_ACTIVE);
 	return 0;
 }
 
@@ -257,6 +369,12 @@ static int init_pfhwdev(struct hinic_pfhwdev *pfhwdev)
  **/
 static void free_pfhwdev(struct hinic_pfhwdev *pfhwdev)
 {
+	struct hinic_hwdev *hwdev = &pfhwdev->hwdev;
+
+	hinic_set_pf_action(hwdev->hwif, HINIC_PF_MGMT_INIT);
+
+	hinic_unregister_mgmt_msg_cb(&pfhwdev->pf_to_mgmt, HINIC_MOD_L2NIC);
+
 	hinic_pf_to_mgmt_free(&pfhwdev->pf_to_mgmt);
 }
 
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h
index ee9e76a..1cd8159 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h
@@ -18,6 +18,7 @@
 
 #include <linux/pci.h>
 #include <linux/types.h>
+#include <linux/bitops.h>
 
 #include "hinic_hw_if.h"
 #include "hinic_hw_eqs.h"
@@ -25,6 +26,9 @@
 
 #define HINIC_MAX_QPS   32
 
+#define HINIC_MGMT_NUM_MSG_CMD  (HINIC_MGMT_MSG_CMD_MAX - \
+				 HINIC_MGMT_MSG_CMD_BASE)
+
 struct hinic_cap {
 	u16     max_qps;
 	u16     num_qps;
@@ -55,6 +59,19 @@ enum hinic_port_cmd {
 	HINIC_PORT_CMD_GET_CAP          = 170,
 };
 
+enum hinic_mgmt_msg_cmd {
+	HINIC_MGMT_MSG_CMD_BASE         = 160,
+
+	HINIC_MGMT_MSG_CMD_LINK_STATUS  = 160,
+
+	HINIC_MGMT_MSG_CMD_MAX,
+};
+
+enum hinic_cb_state {
+	HINIC_CB_ENABLED = BIT(0),
+	HINIC_CB_RUNNING = BIT(1),
+};
+
 struct hinic_hwdev {
 	struct hinic_hwif               *hwif;
 	struct msix_entry               *msix_entries;
@@ -64,12 +81,32 @@ struct hinic_hwdev {
 	struct hinic_cap                nic_cap;
 };
 
+struct hinic_nic_cb {
+	void    (*handler)(void *handle, void *buf_in,
+			   u16 in_size, void *buf_out,
+			   u16 *out_size);
+
+	void            *handle;
+	unsigned long   cb_state;
+};
+
 struct hinic_pfhwdev {
 	struct hinic_hwdev              hwdev;
 
 	struct hinic_pf_to_mgmt         pf_to_mgmt;
+
+	struct hinic_nic_cb             nic_cb[HINIC_MGMT_NUM_MSG_CMD];
 };
 
+void hinic_hwdev_cb_register(struct hinic_hwdev *hwdev,
+			     enum hinic_mgmt_msg_cmd cmd, void *handle,
+			     void (*handler)(void *handle, void *buf_in,
+					     u16 in_size, void *buf_out,
+					     u16 *out_size));
+
+void hinic_hwdev_cb_unregister(struct hinic_hwdev *hwdev,
+			       enum hinic_mgmt_msg_cmd cmd);
+
 int hinic_port_msg_cmd(struct hinic_hwdev *hwdev, enum hinic_port_cmd cmd,
 		       void *buf_in, u16 in_size, void *buf_out,
 		       u16 *out_size);
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_if.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_if.c
index 2b9410c..d4e6ec4 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_if.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_if.c
@@ -117,6 +117,23 @@ int hinic_msix_attr_cnt_clear(struct hinic_hwif *hwif, u16 msix_index)
 }
 
 /**
+ * hinic_set_pf_action - set action on pf channel
+ * @hwif: the HW interface of a pci function device
+ * @action: action on pf channel
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+void hinic_set_pf_action(struct hinic_hwif *hwif, enum hinic_pf_action action)
+{
+	u32 attr5 = hinic_hwif_read_reg(hwif, HINIC_CSR_FUNC_ATTR5_ADDR);
+
+	attr5 = HINIC_FA5_CLEAR(attr5, PF_ACTION);
+	attr5 |= HINIC_FA5_SET(action, PF_ACTION);
+
+	hinic_hwif_write_reg(hwif, HINIC_CSR_FUNC_ATTR5_ADDR, attr5);
+}
+
+/**
  * hwif_ready - test if the HW is ready for use
  * @hwif: the HW interface of a pci function device
  *
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_if.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_if.h
index 707a046..2280698 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_if.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_if.h
@@ -73,6 +73,15 @@
 #define HINIC_FA1_GET(val, member)                              \
 	(((val) >> HINIC_FA1_##member##_SHIFT) & HINIC_FA1_##member##_MASK)
 
+#define HINIC_FA5_PF_ACTION_SHIFT                               0
+#define HINIC_FA5_PF_ACTION_MASK                                0xFFFF
+
+#define HINIC_FA5_SET(val, member)                              \
+	(((u32)(val) & HINIC_FA5_##member##_MASK) << HINIC_FA5_##member##_SHIFT)
+
+#define HINIC_FA5_CLEAR(val, member)                            \
+	((val) & (~(HINIC_FA5_##member##_MASK << HINIC_FA5_##member##_SHIFT)))
+
 #define HINIC_PPF_ELECTION_IDX_SHIFT                            0
 #define HINIC_PPF_ELECTION_IDX_MASK                             0x1F
 
@@ -166,6 +175,12 @@ enum hinic_node_id {
 	HINIC_NODE_ID_MGMT = 21,
 };
 
+enum hinic_pf_action {
+	HINIC_PF_MGMT_INIT = 0x0,
+
+	HINIC_PF_MGMT_ACTIVE = 0x11,
+};
+
 struct hinic_func_attr {
 	u16                     func_idx;
 	u8                      pf_idx;
@@ -212,6 +227,8 @@ int hinic_msix_attr_get(struct hinic_hwif *hwif, u16 msix_index,
 
 int hinic_msix_attr_cnt_clear(struct hinic_hwif *hwif, u16 msix_index);
 
+void hinic_set_pf_action(struct hinic_hwif *hwif, enum hinic_pf_action action);
+
 int hinic_init_hwif(struct hinic_hwif *hwif, struct pci_dev *pdev);
 
 void hinic_free_hwif(struct hinic_hwif *hwif);
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.c
index 147c404..434e134 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.c
@@ -73,6 +73,46 @@ enum msg_ack_type {
 };
 
 /**
+ * hinic_register_mgmt_msg_cb - register msg handler for a msg from a module
+ * @pf_to_mgmt: PF to MGMT channel
+ * @mod: module in the chip that this handler will handle its messages
+ * @handle: private data for the callback
+ * @callback: the handler that will handle messages
+ **/
+void hinic_register_mgmt_msg_cb(struct hinic_pf_to_mgmt *pf_to_mgmt,
+				enum hinic_mod_type mod,
+				void *handle,
+				void (*callback)(void *handle,
+						 u8 cmd, void *buf_in,
+						 u16 in_size, void *buf_out,
+						 u16 *out_size))
+{
+	struct hinic_mgmt_cb *mgmt_cb = &pf_to_mgmt->mgmt_cb[mod];
+
+	mgmt_cb->cb = callback;
+	mgmt_cb->handle = handle;
+	mgmt_cb->state = HINIC_MGMT_CB_ENABLED;
+}
+
+/**
+ * hinic_unregister_mgmt_msg_cb - unregister msg handler for a msg from a module
+ * @pf_to_mgmt: PF to MGMT channel
+ * @mod: module in the chip that this handler handles its messages
+ **/
+void hinic_unregister_mgmt_msg_cb(struct hinic_pf_to_mgmt *pf_to_mgmt,
+				  enum hinic_mod_type mod)
+{
+	struct hinic_mgmt_cb *mgmt_cb = &pf_to_mgmt->mgmt_cb[mod];
+
+	mgmt_cb->state &= ~HINIC_MGMT_CB_ENABLED;
+
+	while (mgmt_cb->state & HINIC_MGMT_CB_RUNNING)
+		schedule();
+
+	mgmt_cb->cb = NULL;
+}
+
+/**
  * prepare_header - prepare the header of the message
  * @pf_to_mgmt: PF to MGMT channel
  * @msg_len: the length of the message
@@ -337,9 +377,31 @@ static void mgmt_recv_msg_handler(struct hinic_pf_to_mgmt *pf_to_mgmt,
 	struct hinic_hwif *hwif = pf_to_mgmt->hwif;
 	struct pci_dev *pdev = hwif->pdev;
 	u8 *buf_out = recv_msg->buf_out;
+	struct hinic_mgmt_cb *mgmt_cb;
+	unsigned long cb_state;
 	u16 out_size = 0;
 
-	dev_err(&pdev->dev, "No MGMT msg handler, mod = %d\n", recv_msg->mod);
+	if (recv_msg->mod >= HINIC_MOD_MAX) {
+		dev_err(&pdev->dev, "Unknown MGMT MSG module = %d\n",
+			recv_msg->mod);
+		return;
+	}
+
+	mgmt_cb = &pf_to_mgmt->mgmt_cb[recv_msg->mod];
+
+	cb_state = cmpxchg(&mgmt_cb->state,
+			   HINIC_MGMT_CB_ENABLED,
+			   HINIC_MGMT_CB_ENABLED | HINIC_MGMT_CB_RUNNING);
+
+	if ((cb_state == HINIC_MGMT_CB_ENABLED) && (mgmt_cb->cb))
+		mgmt_cb->cb(mgmt_cb->handle, recv_msg->cmd,
+			    recv_msg->msg, recv_msg->msg_len,
+			    buf_out, &out_size);
+	else
+		dev_err(&pdev->dev, "No MGMT msg handler, mod = %d\n",
+			recv_msg->mod);
+
+	mgmt_cb->state &= ~HINIC_MGMT_CB_RUNNING;
 
 	if (!recv_msg->async_mgmt_to_pf)
 		/* MGMT sent sync msg, send the response */
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.h
index eca7ad8..8021406 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_mgmt.h
@@ -19,6 +19,7 @@
 #include <linux/types.h>
 #include <linux/semaphore.h>
 #include <linux/completion.h>
+#include <linux/bitops.h>
 
 #include "hinic_hw_if.h"
 #include "hinic_hw_api_cmd.h"
@@ -67,6 +68,11 @@ enum hinic_cfg_cmd {
 	HINIC_CFG_NIC_CAP = 0,
 };
 
+enum hinic_mgmt_cb_state {
+	HINIC_MGMT_CB_ENABLED = BIT(0),
+	HINIC_MGMT_CB_RUNNING = BIT(1),
+};
+
 struct hinic_recv_msg {
 	u8                      *msg;
 	u8                      *buf_out;
@@ -81,6 +87,15 @@ struct hinic_recv_msg {
 	u16                     msg_id;
 };
 
+struct hinic_mgmt_cb {
+	void    (*cb)(void *handle, u8 cmd,
+		      void *buf_in, u16 in_size,
+		      void *buf_out, u16 *out_size);
+
+	void            *handle;
+	unsigned long   state;
+};
+
 struct hinic_pf_to_mgmt {
 	struct hinic_hwif               *hwif;
 
@@ -92,8 +107,21 @@ struct hinic_pf_to_mgmt {
 	struct hinic_recv_msg           recv_msg_from_mgmt;
 
 	struct hinic_api_cmd_chain      *cmd_chain[HINIC_API_CMD_MAX];
+
+	struct hinic_mgmt_cb            mgmt_cb[HINIC_MOD_MAX];
 };
 
+void hinic_register_mgmt_msg_cb(struct hinic_pf_to_mgmt *pf_to_mgmt,
+				enum hinic_mod_type mod,
+				void *handle,
+				void (*callback)(void *handle,
+						 u8 cmd, void *buf_in,
+						 u16 in_size, void *buf_out,
+						 u16 *out_size));
+
+void hinic_unregister_mgmt_msg_cb(struct hinic_pf_to_mgmt *pf_to_mgmt,
+				  enum hinic_mod_type mod);
+
 int hinic_msg_to_mgmt(struct hinic_pf_to_mgmt *pf_to_mgmt,
 		      enum hinic_mod_type mod, u8 cmd,
 		      void *buf_in, u16 in_size, void *buf_out, u16 *out_size,
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_main.c b/drivers/net/ethernet/huawei/hinic/hinic_main.c
index f46c2bc..3fe8a6f 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_main.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_main.c
@@ -24,9 +24,11 @@
 #include <linux/slab.h>
 #include <linux/if_vlan.h>
 #include <linux/semaphore.h>
+#include <linux/workqueue.h>
 #include <net/ip.h>
 #include <linux/bitops.h>
 #include <linux/bitmap.h>
+#include <linux/delay.h>
 #include <linux/err.h>
 
 #include "hinic_hw_dev.h"
@@ -39,12 +41,99 @@
 
 #define PCI_DEVICE_ID_HI1822_PF         0x1822
 
+#define HINIC_WQ_NAME                   "hinic_dev"
+
 #define MSG_ENABLE_DEFAULT              (NETIF_MSG_DRV | NETIF_MSG_PROBE | \
 					 NETIF_MSG_IFUP |                  \
 					 NETIF_MSG_TX_ERR | NETIF_MSG_RX_ERR)
 
 #define VLAN_BITMAP_SIZE(nic_dev)       (ALIGN(VLAN_N_VID, 8) / 8)
 
+#define work_to_rx_mode_work(work)      \
+		container_of(work, struct hinic_rx_mode_work, work)
+
+#define rx_mode_work_to_nic_dev(rx_mode_work) \
+		container_of(rx_mode_work, struct hinic_dev, rx_mode_work)
+
+static int change_mac_addr(struct net_device *netdev, const u8 *addr);
+
+static int hinic_open(struct net_device *netdev)
+{
+	struct hinic_dev *nic_dev = netdev_priv(netdev);
+	enum hinic_port_link_state link_state;
+	int err, ret;
+
+	err = hinic_port_set_state(nic_dev, HINIC_PORT_ENABLE);
+	if (err) {
+		netif_err(nic_dev, drv, netdev,
+			  "Failed to set port state\n");
+		return err;
+	}
+
+	/* Wait up to 3 sec between port enable to link state */
+	msleep(3000);
+
+	down(&nic_dev->mgmt_lock);
+
+	err = hinic_port_link_state(nic_dev, &link_state);
+	if (err) {
+		netif_err(nic_dev, drv, netdev, "Failed to get link state\n");
+		goto err_port_link;
+	}
+
+	if (link_state == HINIC_LINK_STATE_UP)
+		nic_dev->flags |= HINIC_LINK_UP;
+
+	nic_dev->flags |= HINIC_INTF_UP;
+
+	if ((nic_dev->flags & (HINIC_LINK_UP | HINIC_INTF_UP)) ==
+	    (HINIC_LINK_UP | HINIC_INTF_UP)) {
+		netif_info(nic_dev, drv, netdev, "link + intf UP\n");
+		netif_carrier_on(netdev);
+		netif_tx_wake_all_queues(netdev);
+	}
+
+	up(&nic_dev->mgmt_lock);
+
+	netif_info(nic_dev, drv, netdev, "HINIC_INTF is UP\n");
+	return 0;
+
+err_port_link:
+	up(&nic_dev->mgmt_lock);
+	ret = hinic_port_set_state(nic_dev, HINIC_PORT_DISABLE);
+	if (ret)
+		netif_warn(nic_dev, drv, netdev,
+			   "Failed to revert port state\n");
+	return err;
+}
+
+static int hinic_close(struct net_device *netdev)
+{
+	struct hinic_dev *nic_dev = netdev_priv(netdev);
+	unsigned int flags;
+	int err;
+
+	down(&nic_dev->mgmt_lock);
+
+	flags = nic_dev->flags;
+	nic_dev->flags &= ~HINIC_INTF_UP;
+
+	netif_carrier_off(netdev);
+	netif_tx_disable(netdev);
+
+	up(&nic_dev->mgmt_lock);
+
+	err = hinic_port_set_state(nic_dev, HINIC_PORT_DISABLE);
+	if (err) {
+		netif_err(nic_dev, drv, netdev, "Failed to set port state\n");
+		nic_dev->flags |= (flags & HINIC_INTF_UP);
+		return err;
+	}
+
+	netif_info(nic_dev, drv, netdev, "HINIC_INTF is DOWN\n");
+	return 0;
+}
+
 static int hinic_change_mtu(struct net_device *netdev, int new_mtu)
 {
 	struct hinic_dev *nic_dev = netdev_priv(netdev);
@@ -118,6 +207,77 @@ static int hinic_set_mac_addr(struct net_device *netdev, void *addr)
 	return err;
 }
 
+/**
+ * add_mac_addr - add mac address to network device
+ * @netdev: network device
+ * @addr: mac address to add
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int add_mac_addr(struct net_device *netdev, const u8 *addr)
+{
+	struct hinic_dev *nic_dev = netdev_priv(netdev);
+	u16 vid = 0;
+	int err;
+
+	if (!is_valid_ether_addr(addr))
+		return -EADDRNOTAVAIL;
+
+	netif_info(nic_dev, drv, netdev, "set mac addr = %02x %02x %02x %02x %02x %02x\n",
+		   addr[0], addr[1], addr[2], addr[3], addr[4], addr[5]);
+
+	down(&nic_dev->mgmt_lock);
+
+	do {
+		err = hinic_port_add_mac(nic_dev, addr, vid);
+		if (err) {
+			netif_err(nic_dev, drv, netdev, "Failed to add mac\n");
+			break;
+		}
+
+		vid = find_next_bit(nic_dev->vlan_bitmap, VLAN_N_VID, vid + 1);
+	} while (vid != VLAN_N_VID);
+
+	up(&nic_dev->mgmt_lock);
+	return err;
+}
+
+/**
+ * remove_mac_addr - remove mac address from network device
+ * @netdev: network device
+ * @addr: mac address to remove
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int remove_mac_addr(struct net_device *netdev, const u8 *addr)
+{
+	struct hinic_dev *nic_dev = netdev_priv(netdev);
+	u16 vid = 0;
+	int err;
+
+	if (!is_valid_ether_addr(addr))
+		return -EADDRNOTAVAIL;
+
+	netif_info(nic_dev, drv, netdev, "remove mac addr = %02x %02x %02x %02x %02x %02x\n",
+		   addr[0], addr[1], addr[2], addr[3], addr[4], addr[5]);
+
+	down(&nic_dev->mgmt_lock);
+
+	do {
+		err = hinic_port_del_mac(nic_dev, addr, vid);
+		if (err) {
+			netif_err(nic_dev, drv, netdev,
+				  "Failed to delete mac\n");
+			break;
+		}
+
+		vid = find_next_bit(nic_dev->vlan_bitmap, VLAN_N_VID, vid + 1);
+	} while (vid != VLAN_N_VID);
+
+	up(&nic_dev->mgmt_lock);
+	return err;
+}
+
 static int hinic_vlan_rx_add_vid(struct net_device *netdev,
 				 __always_unused __be16 proto, u16 vid)
 {
@@ -182,12 +342,56 @@ static int hinic_vlan_rx_kill_vid(struct net_device *netdev,
 	return err;
 }
 
+static void set_rx_mode(struct work_struct *work)
+{
+	struct hinic_rx_mode_work *rx_mode_work = work_to_rx_mode_work(work);
+	struct hinic_dev *nic_dev = rx_mode_work_to_nic_dev(rx_mode_work);
+
+	netif_info(nic_dev, drv, nic_dev->netdev, "set rx mode work\n");
+
+	hinic_port_set_rx_mode(nic_dev, rx_mode_work->rx_mode);
+
+	__dev_uc_sync(nic_dev->netdev, add_mac_addr, remove_mac_addr);
+	__dev_mc_sync(nic_dev->netdev, add_mac_addr, remove_mac_addr);
+}
+
+static void hinic_set_rx_mode(struct net_device *netdev)
+{
+	struct hinic_dev *nic_dev = netdev_priv(netdev);
+	struct hinic_rx_mode_work *rx_mode_work;
+	u32 rx_mode;
+
+	rx_mode_work = &nic_dev->rx_mode_work;
+
+	rx_mode = HINIC_RX_MODE_UC |
+		  HINIC_RX_MODE_MC |
+		  HINIC_RX_MODE_BC;
+
+	if (netdev->flags & IFF_PROMISC)
+		rx_mode |= HINIC_RX_MODE_PROMISC;
+	else if (netdev->flags & IFF_ALLMULTI)
+		rx_mode |= HINIC_RX_MODE_MC_ALL;
+
+	rx_mode_work->rx_mode = rx_mode;
+
+	queue_work(nic_dev->workq, &rx_mode_work->work);
+}
+
+netdev_tx_t hinic_xmit_frame(struct sk_buff *skb, struct net_device *netdev)
+{
+	return NETDEV_TX_BUSY;
+}
+
 static const struct net_device_ops hinic_netdev_ops = {
+	.ndo_open = hinic_open,
+	.ndo_stop = hinic_close,
 	.ndo_change_mtu = hinic_change_mtu,
 	.ndo_set_mac_address = hinic_set_mac_addr,
 	.ndo_validate_addr = eth_validate_addr,
 	.ndo_vlan_rx_add_vid = hinic_vlan_rx_add_vid,
 	.ndo_vlan_rx_kill_vid = hinic_vlan_rx_kill_vid,
+	.ndo_set_rx_mode = hinic_set_rx_mode,
+	.ndo_start_xmit = hinic_xmit_frame,
 	/* more operations should be filled */
 };
 
@@ -201,6 +405,57 @@ static void netdev_features_init(struct net_device *netdev)
 }
 
 /**
+ * link_status_event_handler - link event handler
+ * @handle: nic device for the handler
+ * @buf_in: input buffer
+ * @in_size: input size
+ * @buf_in: output buffer
+ * @out_size: returned output size
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static void link_status_event_handler(void *handle, void *buf_in, u16 in_size,
+				      void *buf_out, u16 *out_size)
+{
+	struct hinic_port_link_status *link_status, *ret_link_status;
+	struct hinic_dev *nic_dev = (struct hinic_dev *)handle;
+
+	link_status = buf_in;
+
+	if (link_status->link == HINIC_LINK_STATE_UP) {
+		down(&nic_dev->mgmt_lock);
+
+		nic_dev->flags |= HINIC_LINK_UP;
+
+		if ((nic_dev->flags & (HINIC_LINK_UP | HINIC_INTF_UP)) ==
+		    (HINIC_LINK_UP | HINIC_INTF_UP)) {
+			netif_carrier_on(nic_dev->netdev);
+			netif_tx_wake_all_queues(nic_dev->netdev);
+		}
+
+		up(&nic_dev->mgmt_lock);
+
+		netif_info(nic_dev, drv, nic_dev->netdev, "HINIC_Link is UP\n");
+	} else {
+		down(&nic_dev->mgmt_lock);
+
+		nic_dev->flags &= ~HINIC_LINK_UP;
+
+		netif_carrier_off(nic_dev->netdev);
+		netif_tx_disable(nic_dev->netdev);
+
+		up(&nic_dev->mgmt_lock);
+
+		netif_info(nic_dev, drv, nic_dev->netdev, "HINIC_Link is DOWN\n");
+	}
+
+	ret_link_status = buf_out;
+	ret_link_status->status = 0;
+
+	*out_size = sizeof(*ret_link_status);
+}
+
+/**
  * nic_dev_init - Initialize the NIC device
  * @pdev: the NIC pci device
  *
@@ -208,6 +463,7 @@ static void netdev_features_init(struct net_device *netdev)
  **/
 static int nic_dev_init(struct pci_dev *pdev)
 {
+	struct hinic_rx_mode_work *rx_mode_work;
 	struct hinic_dev *nic_dev;
 	struct net_device *netdev;
 	struct hinic_hwdev *hwdev;
@@ -239,6 +495,7 @@ static int nic_dev_init(struct pci_dev *pdev)
 	nic_dev->netdev = netdev;
 	nic_dev->hwdev = hwdev;
 	nic_dev->msg_enable = MSG_ENABLE_DEFAULT;
+	nic_dev->flags = 0;
 
 	sema_init(&nic_dev->mgmt_lock, 1);
 
@@ -250,6 +507,12 @@ static int nic_dev_init(struct pci_dev *pdev)
 		goto err_vlan_bitmap;
 	}
 
+	nic_dev->workq = create_singlethread_workqueue(HINIC_WQ_NAME);
+	if (!nic_dev->workq) {
+		err = -ENOMEM;
+		goto err_workq;
+	}
+
 	pci_set_drvdata(pdev, netdev);
 
 	err = hinic_port_get_mac(nic_dev, netdev->dev_addr);
@@ -268,10 +531,16 @@ static int nic_dev_init(struct pci_dev *pdev)
 		goto err_set_mtu;
 	}
 
+	rx_mode_work = &nic_dev->rx_mode_work;
+	INIT_WORK(&rx_mode_work->work, set_rx_mode);
+
 	netdev_features_init(netdev);
 
 	netif_carrier_off(netdev);
 
+	hinic_hwdev_cb_register(nic_dev->hwdev, HINIC_MGMT_MSG_CMD_LINK_STATUS,
+				nic_dev, link_status_event_handler);
+
 	err = register_netdev(netdev);
 	if (err) {
 		dev_err(&pdev->dev, "Failed to register netdev\n");
@@ -281,10 +550,16 @@ static int nic_dev_init(struct pci_dev *pdev)
 	return 0;
 
 err_reg_netdev:
+	hinic_hwdev_cb_unregister(nic_dev->hwdev,
+				  HINIC_MGMT_MSG_CMD_LINK_STATUS);
+	cancel_work_sync(&rx_mode_work->work);
+
 err_set_mtu:
 err_add_mac:
 	pci_set_drvdata(pdev, NULL);
+	destroy_workqueue(nic_dev->workq);
 
+err_workq:
 err_vlan_bitmap:
 	free_netdev(netdev);
 
@@ -357,11 +632,20 @@ static void hinic_remove(struct pci_dev *pdev)
 {
 	struct net_device *netdev = pci_get_drvdata(pdev);
 	struct hinic_dev *nic_dev = netdev_priv(netdev);
+	struct hinic_rx_mode_work *rx_mode_work;
 
 	unregister_netdev(netdev);
 
+	hinic_hwdev_cb_unregister(nic_dev->hwdev,
+				  HINIC_MGMT_MSG_CMD_LINK_STATUS);
+
+	rx_mode_work = &nic_dev->rx_mode_work;
+	cancel_work_sync(&rx_mode_work->work);
+
 	pci_set_drvdata(pdev, NULL);
 
+	destroy_workqueue(nic_dev->workq);
+
 	hinic_free_hwdev(nic_dev->hwdev);
 
 	free_netdev(netdev);
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_port.c b/drivers/net/ethernet/huawei/hinic/hinic_port.c
index 5b249e8..0dafede 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_port.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_port.c
@@ -222,3 +222,95 @@ int hinic_port_del_vlan(struct hinic_dev *nic_dev, u16 vlan_id)
 				 &port_vlan_cmd, sizeof(port_vlan_cmd),
 				 NULL, NULL);
 }
+
+/**
+ * hinic_port_set_rx_mode - set rx mode in the nic device
+ * @nic_dev: nic device
+ * @rx_mode: the rx mode to set
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_port_set_rx_mode(struct hinic_dev *nic_dev, u32 rx_mode)
+{
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	struct hinic_port_rx_mode_cmd rx_mode_cmd;
+
+	rx_mode_cmd.func_idx = HINIC_HWIF_FUNC_IDX(hwdev->hwif);
+	rx_mode_cmd.rx_mode = rx_mode;
+
+	return hinic_port_msg_cmd(hwdev, HINIC_PORT_CMD_SET_RX_MODE,
+				  &rx_mode_cmd, sizeof(rx_mode_cmd),
+				  NULL, NULL);
+}
+
+/**
+ * hinic_port_link_state - get the link state
+ * @nic_dev: nic device
+ * @link_state: the returned link state
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_port_link_state(struct hinic_dev *nic_dev,
+			  enum hinic_port_link_state *link_state)
+{
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct hinic_port_link_cmd link_cmd;
+	struct pci_dev *pdev = hwif->pdev;
+	u16 out_size;
+	int err;
+
+	if (!HINIC_IS_PF(hwif) && !HINIC_IS_PPF(hwif)) {
+		dev_err(&pdev->dev, "unsupported PCI Function type\n");
+		return -EINVAL;
+	}
+
+	link_cmd.func_idx = HINIC_HWIF_FUNC_IDX(hwif);
+
+	err = hinic_port_msg_cmd(hwdev, HINIC_PORT_CMD_GET_LINK_STATE,
+				 &link_cmd, sizeof(link_cmd),
+				 &link_cmd, &out_size);
+	if (err || (out_size != sizeof(link_cmd)) || link_cmd.status) {
+		dev_err(&pdev->dev, "Failed to get link state, ret = %d\n",
+			link_cmd.status);
+		return -EINVAL;
+	}
+
+	*link_state = link_cmd.state;
+	return 0;
+}
+
+/**
+ * hinic_port_set_state - set port state
+ * @nic_dev: nic device
+ * @state: the state to set
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_port_set_state(struct hinic_dev *nic_dev, enum hinic_port_state state)
+{
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	struct hinic_port_state_cmd port_state;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	u16 out_size;
+	int err;
+
+	if (!HINIC_IS_PF(hwif) && !HINIC_IS_PPF(hwif)) {
+		dev_err(&pdev->dev, "unsupported PCI Function type\n");
+		return -EINVAL;
+	}
+
+	port_state.state = state;
+
+	err = hinic_port_msg_cmd(hwdev, HINIC_PORT_CMD_SET_PORT_STATE,
+				 &port_state, sizeof(port_state),
+				 &port_state, &out_size);
+	if (err || (out_size != sizeof(port_state)) || port_state.status) {
+		dev_err(&pdev->dev, "Failed to set port state, ret = %d\n",
+			port_state.status);
+		return -EFAULT;
+	}
+
+	return 0;
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_port.h b/drivers/net/ethernet/huawei/hinic/hinic_port.h
index 4cafb94..3a8da8e 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_port.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_port.h
@@ -18,9 +18,28 @@
 
 #include <linux/types.h>
 #include <linux/etherdevice.h>
+#include <linux/bitops.h>
 
 #include "hinic_dev.h"
 
+enum hinic_rx_mode {
+	HINIC_RX_MODE_UC        = BIT(0),
+	HINIC_RX_MODE_MC        = BIT(1),
+	HINIC_RX_MODE_BC        = BIT(2),
+	HINIC_RX_MODE_MC_ALL    = BIT(3),
+	HINIC_RX_MODE_PROMISC   = BIT(4),
+};
+
+enum hinic_port_link_state {
+	HINIC_LINK_STATE_DOWN,
+	HINIC_LINK_STATE_UP,
+};
+
+enum hinic_port_state {
+	HINIC_PORT_DISABLE      = 0,
+	HINIC_PORT_ENABLE       = 3,
+};
+
 struct hinic_port_mac_cmd {
 	u8              status;
 	u8              version;
@@ -51,6 +70,45 @@ struct hinic_port_vlan_cmd {
 	u16     vlan_id;
 };
 
+struct hinic_port_rx_mode_cmd {
+	u8      status;
+	u8      version;
+	u8      rsvd0[6];
+
+	u16     func_idx;
+	u16     rsvd;
+	u32     rx_mode;
+};
+
+struct hinic_port_link_cmd {
+	u8      status;
+	u8      version;
+	u8      rsvd0[6];
+
+	u16     func_idx;
+	u8      state;
+	u8      rsvd1;
+};
+
+struct hinic_port_state_cmd {
+	u8      status;
+	u8      version;
+	u8      rsvd0[6];
+
+	u8      state;
+	u8      rsvd1[3];
+};
+
+struct hinic_port_link_status {
+	u8      status;
+	u8      version;
+	u8      rsvd0[6];
+
+	u16     rsvd1;
+	u8      link;
+	u8      rsvd2;
+};
+
 int hinic_port_add_mac(struct hinic_dev *nic_dev, const u8 *addr,
 		       u16 vlan_id);
 
@@ -65,4 +123,12 @@ int hinic_port_del_mac(struct hinic_dev *nic_dev, const u8 *addr,
 
 int hinic_port_del_vlan(struct hinic_dev *nic_dev, u16 vlan_id);
 
+int hinic_port_set_rx_mode(struct hinic_dev *nic_dev, u32 rx_mode);
+
+int hinic_port_link_state(struct hinic_dev *nic_dev,
+			  enum hinic_port_link_state *link_state);
+
+int hinic_port_set_state(struct hinic_dev *nic_dev,
+			 enum hinic_port_state state);
+
 #endif
-- 
1.9.1

^ permalink raw reply related

* [PATCH V7 net-next 08/22] net-next/hinic: Add port management commands
From: Aviad Krawczyk @ 2017-08-17 19:04 UTC (permalink / raw)
  To: davem
  Cc: linux-kernel, netdev, bc.y, victor.gissin, aviad.krawczyk,
	zhaochen6, tony.qu
In-Reply-To: <cover.1502996306.git.aviad.krawczyk@huawei.com>

Add the port management commands that are sent as management messages.
The port management commands are used for netdev operations.

Signed-off-by: Aviad Krawczyk <aviad.krawczyk@huawei.com>
Signed-off-by: Zhao Chen <zhaochen6@huawei.com>
---
 drivers/net/ethernet/huawei/hinic/Makefile       |   4 +-
 drivers/net/ethernet/huawei/hinic/hinic_dev.h    |   4 +
 drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c |  30 +++
 drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h |  29 +++
 drivers/net/ethernet/huawei/hinic/hinic_main.c   | 195 +++++++++++++++++++-
 drivers/net/ethernet/huawei/hinic/hinic_port.c   | 224 +++++++++++++++++++++++
 drivers/net/ethernet/huawei/hinic/hinic_port.h   |  68 +++++++
 7 files changed, 551 insertions(+), 3 deletions(-)
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_port.c
 create mode 100644 drivers/net/ethernet/huawei/hinic/hinic_port.h

diff --git a/drivers/net/ethernet/huawei/hinic/Makefile b/drivers/net/ethernet/huawei/hinic/Makefile
index beba90a..dbb1b9d 100644
--- a/drivers/net/ethernet/huawei/hinic/Makefile
+++ b/drivers/net/ethernet/huawei/hinic/Makefile
@@ -1,4 +1,4 @@
 obj-$(CONFIG_HINIC) += hinic.o
 
-hinic-y := hinic_main.o hinic_hw_dev.o hinic_hw_mgmt.o hinic_hw_api_cmd.o \
-	   hinic_hw_eqs.o hinic_hw_if.o
+hinic-y := hinic_main.o hinic_port.o hinic_hw_dev.o hinic_hw_mgmt.o \
+	   hinic_hw_api_cmd.o hinic_hw_eqs.o hinic_hw_if.o
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_dev.h b/drivers/net/ethernet/huawei/hinic/hinic_dev.h
index 6c2c896..e54a45c 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_dev.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_dev.h
@@ -18,6 +18,7 @@
 
 #include <linux/netdevice.h>
 #include <linux/types.h>
+#include <linux/semaphore.h>
 
 #include "hinic_hw_dev.h"
 
@@ -28,6 +29,9 @@ struct hinic_dev {
 	struct hinic_hwdev              *hwdev;
 
 	u32                             msg_enable;
+
+	struct semaphore                mgmt_lock;
+	unsigned long                   *vlan_bitmap;
 };
 
 #endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
index d430e60..6bb6c33 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.c
@@ -200,6 +200,36 @@ static void disable_msix(struct hinic_hwdev *hwdev)
 }
 
 /**
+ * hinic_port_msg_cmd - send port msg to mgmt
+ * @hwdev: the NIC HW device
+ * @cmd: the port command
+ * @buf_in: input buffer
+ * @in_size: input size
+ * @buf_out: output buffer
+ * @out_size: returned output size
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_port_msg_cmd(struct hinic_hwdev *hwdev, enum hinic_port_cmd cmd,
+		       void *buf_in, u16 in_size, void *buf_out, u16 *out_size)
+{
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_pfhwdev *pfhwdev;
+
+	if (!HINIC_IS_PF(hwif) && !HINIC_IS_PPF(hwif)) {
+		dev_err(&pdev->dev, "unsupported PCI Function type\n");
+		return -EINVAL;
+	}
+
+	pfhwdev = container_of(hwdev, struct hinic_pfhwdev, hwdev);
+
+	return hinic_msg_to_mgmt(&pfhwdev->pf_to_mgmt, HINIC_MOD_L2NIC, cmd,
+				 buf_in, in_size, buf_out, out_size,
+				 HINIC_MGMT_MSG_SYNC);
+}
+
+/**
  * init_pfhwdev - Initialize the extended components of PF
  * @pfhwdev: the HW device for PF
  *
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h
index feb60138..ee9e76a 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_dev.h
@@ -30,6 +30,31 @@ struct hinic_cap {
 	u16     num_qps;
 };
 
+enum hinic_port_cmd {
+	HINIC_PORT_CMD_CHANGE_MTU       = 2,
+
+	HINIC_PORT_CMD_ADD_VLAN         = 3,
+	HINIC_PORT_CMD_DEL_VLAN         = 4,
+
+	HINIC_PORT_CMD_SET_MAC          = 9,
+	HINIC_PORT_CMD_GET_MAC          = 10,
+	HINIC_PORT_CMD_DEL_MAC          = 11,
+
+	HINIC_PORT_CMD_SET_RX_MODE      = 12,
+
+	HINIC_PORT_CMD_GET_LINK_STATE   = 24,
+
+	HINIC_PORT_CMD_SET_PORT_STATE   = 41,
+
+	HINIC_PORT_CMD_FWCTXT_INIT      = 69,
+
+	HINIC_PORT_CMD_SET_FUNC_STATE   = 93,
+
+	HINIC_PORT_CMD_GET_GLOBAL_QPN   = 102,
+
+	HINIC_PORT_CMD_GET_CAP          = 170,
+};
+
 struct hinic_hwdev {
 	struct hinic_hwif               *hwif;
 	struct msix_entry               *msix_entries;
@@ -45,6 +70,10 @@ struct hinic_pfhwdev {
 	struct hinic_pf_to_mgmt         pf_to_mgmt;
 };
 
+int hinic_port_msg_cmd(struct hinic_hwdev *hwdev, enum hinic_port_cmd cmd,
+		       void *buf_in, u16 in_size, void *buf_out,
+		       u16 *out_size);
+
 struct hinic_hwdev *hinic_init_hwdev(struct pci_dev *pdev);
 
 void hinic_free_hwdev(struct hinic_hwdev *hwdev);
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_main.c b/drivers/net/ethernet/huawei/hinic/hinic_main.c
index 10df850..f46c2bc 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_main.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_main.c
@@ -13,6 +13,7 @@
  *
  */
 
+#include <linux/kernel.h>
 #include <linux/module.h>
 #include <linux/pci.h>
 #include <linux/device.h>
@@ -20,9 +21,16 @@
 #include <linux/types.h>
 #include <linux/etherdevice.h>
 #include <linux/netdevice.h>
+#include <linux/slab.h>
+#include <linux/if_vlan.h>
+#include <linux/semaphore.h>
+#include <net/ip.h>
+#include <linux/bitops.h>
+#include <linux/bitmap.h>
 #include <linux/err.h>
 
 #include "hinic_hw_dev.h"
+#include "hinic_port.h"
 #include "hinic_dev.h"
 
 MODULE_AUTHOR("Huawei Technologies CO., Ltd");
@@ -35,10 +43,163 @@
 					 NETIF_MSG_IFUP |                  \
 					 NETIF_MSG_TX_ERR | NETIF_MSG_RX_ERR)
 
+#define VLAN_BITMAP_SIZE(nic_dev)       (ALIGN(VLAN_N_VID, 8) / 8)
+
+static int hinic_change_mtu(struct net_device *netdev, int new_mtu)
+{
+	struct hinic_dev *nic_dev = netdev_priv(netdev);
+	int err;
+
+	netif_info(nic_dev, drv, netdev, "set_mtu = %d\n", new_mtu);
+
+	err = hinic_port_set_mtu(nic_dev, new_mtu);
+	if (err)
+		netif_err(nic_dev, drv, netdev, "Failed to set port mtu\n");
+	else
+		netdev->mtu = new_mtu;
+
+	return err;
+}
+
+/**
+ * change_mac_addr - change the main mac address of network device
+ * @netdev: network device
+ * @addr: mac address to set
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int change_mac_addr(struct net_device *netdev, const u8 *addr)
+{
+	struct hinic_dev *nic_dev = netdev_priv(netdev);
+	u16 vid = 0;
+	int err;
+
+	if (!is_valid_ether_addr(addr))
+		return -EADDRNOTAVAIL;
+
+	netif_info(nic_dev, drv, netdev, "change mac addr = %02x %02x %02x %02x %02x %02x\n",
+		   addr[0], addr[1], addr[2], addr[3], addr[4], addr[5]);
+
+	down(&nic_dev->mgmt_lock);
+
+	do {
+		err = hinic_port_del_mac(nic_dev, netdev->dev_addr, vid);
+		if (err) {
+			netif_err(nic_dev, drv, netdev,
+				  "Failed to delete mac\n");
+			break;
+		}
+
+		err = hinic_port_add_mac(nic_dev, addr, vid);
+		if (err) {
+			netif_err(nic_dev, drv, netdev, "Failed to add mac\n");
+			break;
+		}
+
+		vid = find_next_bit(nic_dev->vlan_bitmap, VLAN_N_VID, vid + 1);
+	} while (vid != VLAN_N_VID);
+
+	up(&nic_dev->mgmt_lock);
+	return err;
+}
+
+static int hinic_set_mac_addr(struct net_device *netdev, void *addr)
+{
+	unsigned char new_mac[ETH_ALEN];
+	struct sockaddr *saddr = addr;
+	int err;
+
+	memcpy(new_mac, saddr->sa_data, ETH_ALEN);
+
+	err = change_mac_addr(netdev, new_mac);
+	if (!err)
+		memcpy(netdev->dev_addr, new_mac, ETH_ALEN);
+
+	return err;
+}
+
+static int hinic_vlan_rx_add_vid(struct net_device *netdev,
+				 __always_unused __be16 proto, u16 vid)
+{
+	struct hinic_dev *nic_dev = netdev_priv(netdev);
+	int ret, err;
+
+	netif_info(nic_dev, drv, netdev, "add vid = %d\n", vid);
+
+	down(&nic_dev->mgmt_lock);
+
+	err = hinic_port_add_vlan(nic_dev, vid);
+	if (err) {
+		netif_err(nic_dev, drv, netdev, "Failed to add vlan\n");
+		goto err_vlan_add;
+	}
+
+	err = hinic_port_add_mac(nic_dev, netdev->dev_addr, vid);
+	if (err) {
+		netif_err(nic_dev, drv, netdev, "Failed to set mac\n");
+		goto err_add_mac;
+	}
+
+	bitmap_set(nic_dev->vlan_bitmap, vid, 1);
+
+	up(&nic_dev->mgmt_lock);
+	return 0;
+
+err_add_mac:
+	ret = hinic_port_del_vlan(nic_dev, vid);
+	if (ret)
+		netif_err(nic_dev, drv, netdev,
+			  "Failed to revert by removing vlan\n");
+
+err_vlan_add:
+	up(&nic_dev->mgmt_lock);
+	return err;
+}
+
+static int hinic_vlan_rx_kill_vid(struct net_device *netdev,
+				  __always_unused __be16 proto, u16 vid)
+{
+	struct hinic_dev *nic_dev = netdev_priv(netdev);
+	int err;
+
+	netif_info(nic_dev, drv, netdev, "remove vid = %d\n", vid);
+
+	down(&nic_dev->mgmt_lock);
+
+	err = hinic_port_del_vlan(nic_dev, vid);
+	if (err) {
+		netif_err(nic_dev, drv, netdev, "Failed to delete vlan\n");
+		goto err_del_vlan;
+	}
+
+	bitmap_clear(nic_dev->vlan_bitmap, vid, 1);
+
+	up(&nic_dev->mgmt_lock);
+	return 0;
+
+err_del_vlan:
+	up(&nic_dev->mgmt_lock);
+	return err;
+}
+
 static const struct net_device_ops hinic_netdev_ops = {
-	/* Operations are empty, should be filled */
+	.ndo_change_mtu = hinic_change_mtu,
+	.ndo_set_mac_address = hinic_set_mac_addr,
+	.ndo_validate_addr = eth_validate_addr,
+	.ndo_vlan_rx_add_vid = hinic_vlan_rx_add_vid,
+	.ndo_vlan_rx_kill_vid = hinic_vlan_rx_kill_vid,
+	/* more operations should be filled */
 };
 
+static void netdev_features_init(struct net_device *netdev)
+{
+	netdev->hw_features = NETIF_F_SG | NETIF_F_HIGHDMA;
+
+	netdev->vlan_features = netdev->hw_features;
+
+	netdev->features = netdev->hw_features | NETIF_F_HW_VLAN_CTAG_FILTER;
+}
+
 /**
  * nic_dev_init - Initialize the NIC device
  * @pdev: the NIC pci device
@@ -79,8 +240,36 @@ static int nic_dev_init(struct pci_dev *pdev)
 	nic_dev->hwdev = hwdev;
 	nic_dev->msg_enable = MSG_ENABLE_DEFAULT;
 
+	sema_init(&nic_dev->mgmt_lock, 1);
+
+	nic_dev->vlan_bitmap = devm_kzalloc(&pdev->dev,
+					    VLAN_BITMAP_SIZE(nic_dev),
+					    GFP_KERNEL);
+	if (!nic_dev->vlan_bitmap) {
+		err = -ENOMEM;
+		goto err_vlan_bitmap;
+	}
+
 	pci_set_drvdata(pdev, netdev);
 
+	err = hinic_port_get_mac(nic_dev, netdev->dev_addr);
+	if (err)
+		dev_warn(&pdev->dev, "Failed to get mac address\n");
+
+	err = hinic_port_add_mac(nic_dev, netdev->dev_addr, 0);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to add mac\n");
+		goto err_add_mac;
+	}
+
+	err = hinic_port_set_mtu(nic_dev, netdev->mtu);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to set mtu\n");
+		goto err_set_mtu;
+	}
+
+	netdev_features_init(netdev);
+
 	netif_carrier_off(netdev);
 
 	err = register_netdev(netdev);
@@ -92,7 +281,11 @@ static int nic_dev_init(struct pci_dev *pdev)
 	return 0;
 
 err_reg_netdev:
+err_set_mtu:
+err_add_mac:
 	pci_set_drvdata(pdev, NULL);
+
+err_vlan_bitmap:
 	free_netdev(netdev);
 
 err_alloc_etherdev:
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_port.c b/drivers/net/ethernet/huawei/hinic/hinic_port.c
new file mode 100644
index 0000000..5b249e8
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_port.c
@@ -0,0 +1,224 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#include <linux/types.h>
+#include <linux/netdevice.h>
+#include <linux/etherdevice.h>
+#include <linux/if_vlan.h>
+#include <linux/pci.h>
+#include <linux/device.h>
+#include <linux/errno.h>
+
+#include "hinic_hw_if.h"
+#include "hinic_hw_dev.h"
+#include "hinic_port.h"
+#include "hinic_dev.h"
+
+#define HINIC_MIN_MTU_SIZE              256
+#define HINIC_MAX_JUMBO_FRAME_SIZE      15872
+
+enum mac_op {
+	MAC_DEL,
+	MAC_SET,
+};
+
+/**
+ * change_mac - change(add or delete) mac address
+ * @nic_dev: nic device
+ * @addr: mac address
+ * @vlan_id: vlan number to set with the mac
+ * @op: add or delete the mac
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+static int change_mac(struct hinic_dev *nic_dev, const u8 *addr,
+		      u16 vlan_id, enum mac_op op)
+{
+	struct net_device *netdev = nic_dev->netdev;
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	struct hinic_port_mac_cmd port_mac_cmd;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	enum hinic_port_cmd cmd;
+	u16 out_size;
+	int err;
+
+	if (vlan_id >= VLAN_N_VID) {
+		netif_err(nic_dev, drv, netdev, "Invalid VLAN number\n");
+		return -EINVAL;
+	}
+
+	if (op == MAC_SET)
+		cmd = HINIC_PORT_CMD_SET_MAC;
+	else
+		cmd = HINIC_PORT_CMD_DEL_MAC;
+
+	port_mac_cmd.func_idx = HINIC_HWIF_FUNC_IDX(hwif);
+	port_mac_cmd.vlan_id = vlan_id;
+	memcpy(port_mac_cmd.mac, addr, ETH_ALEN);
+
+	err = hinic_port_msg_cmd(hwdev, cmd, &port_mac_cmd,
+				 sizeof(port_mac_cmd),
+				 &port_mac_cmd, &out_size);
+	if (err || (out_size != sizeof(port_mac_cmd)) || port_mac_cmd.status) {
+		dev_err(&pdev->dev, "Failed to change MAC, ret = %d\n",
+			port_mac_cmd.status);
+		return -EFAULT;
+	}
+
+	return 0;
+}
+
+/**
+ * hinic_port_add_mac - add mac address
+ * @nic_dev: nic device
+ * @addr: mac address
+ * @vlan_id: vlan number to set with the mac
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_port_add_mac(struct hinic_dev *nic_dev,
+		       const u8 *addr, u16 vlan_id)
+{
+	return change_mac(nic_dev, addr, vlan_id, MAC_SET);
+}
+
+/**
+ * hinic_port_del_mac - remove mac address
+ * @nic_dev: nic device
+ * @addr: mac address
+ * @vlan_id: vlan number that is connected to the mac
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_port_del_mac(struct hinic_dev *nic_dev, const u8 *addr,
+		       u16 vlan_id)
+{
+	return change_mac(nic_dev, addr, vlan_id, MAC_DEL);
+}
+
+/**
+ * hinic_port_get_mac - get the mac address of the nic device
+ * @nic_dev: nic device
+ * @addr: returned mac address
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_port_get_mac(struct hinic_dev *nic_dev, u8 *addr)
+{
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	struct hinic_port_mac_cmd port_mac_cmd;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	u16 out_size;
+	int err;
+
+	port_mac_cmd.func_idx = HINIC_HWIF_FUNC_IDX(hwif);
+
+	err = hinic_port_msg_cmd(hwdev, HINIC_PORT_CMD_GET_MAC,
+				 &port_mac_cmd, sizeof(port_mac_cmd),
+				 &port_mac_cmd, &out_size);
+	if (err || (out_size != sizeof(port_mac_cmd)) || port_mac_cmd.status) {
+		dev_err(&pdev->dev, "Failed to get mac, ret = %d\n",
+			port_mac_cmd.status);
+		return -EFAULT;
+	}
+
+	memcpy(addr, port_mac_cmd.mac, ETH_ALEN);
+	return 0;
+}
+
+/**
+ * hinic_port_set_mtu - set mtu
+ * @nic_dev: nic device
+ * @new_mtu: new mtu
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_port_set_mtu(struct hinic_dev *nic_dev, int new_mtu)
+{
+	struct net_device *netdev = nic_dev->netdev;
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	struct hinic_port_mtu_cmd port_mtu_cmd;
+	struct hinic_hwif *hwif = hwdev->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	int err, max_frame;
+	u16 out_size;
+
+	if (new_mtu < HINIC_MIN_MTU_SIZE) {
+		netif_err(nic_dev, drv, netdev, "mtu < MIN MTU size");
+		return -EINVAL;
+	}
+
+	max_frame = new_mtu + ETH_HLEN + ETH_FCS_LEN;
+	if (max_frame > HINIC_MAX_JUMBO_FRAME_SIZE) {
+		netif_err(nic_dev, drv, netdev, "mtu > MAX MTU size");
+		return -EINVAL;
+	}
+
+	port_mtu_cmd.func_idx = HINIC_HWIF_FUNC_IDX(hwif);
+	port_mtu_cmd.mtu = new_mtu;
+
+	err = hinic_port_msg_cmd(hwdev, HINIC_PORT_CMD_CHANGE_MTU,
+				 &port_mtu_cmd, sizeof(port_mtu_cmd),
+				 &port_mtu_cmd, &out_size);
+	if (err || (out_size != sizeof(port_mtu_cmd)) || port_mtu_cmd.status) {
+		dev_err(&pdev->dev, "Failed to set mtu, ret = %d\n",
+			port_mtu_cmd.status);
+		return -EFAULT;
+	}
+
+	return 0;
+}
+
+/**
+ * hinic_port_add_vlan - add vlan to the nic device
+ * @nic_dev: nic device
+ * @vlan_id: the vlan number to add
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_port_add_vlan(struct hinic_dev *nic_dev, u16 vlan_id)
+{
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	struct hinic_port_vlan_cmd port_vlan_cmd;
+
+	port_vlan_cmd.func_idx = HINIC_HWIF_FUNC_IDX(hwdev->hwif);
+	port_vlan_cmd.vlan_id = vlan_id;
+
+	return hinic_port_msg_cmd(hwdev, HINIC_PORT_CMD_ADD_VLAN,
+				  &port_vlan_cmd, sizeof(port_vlan_cmd),
+				  NULL, NULL);
+}
+
+/**
+ * hinic_port_del_vlan - delete vlan from the nic device
+ * @nic_dev: nic device
+ * @vlan_id: the vlan number to delete
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_port_del_vlan(struct hinic_dev *nic_dev, u16 vlan_id)
+{
+	struct hinic_hwdev *hwdev = nic_dev->hwdev;
+	struct hinic_port_vlan_cmd port_vlan_cmd;
+
+	port_vlan_cmd.func_idx = HINIC_HWIF_FUNC_IDX(hwdev->hwif);
+	port_vlan_cmd.vlan_id = vlan_id;
+
+	return hinic_port_msg_cmd(hwdev, HINIC_PORT_CMD_DEL_VLAN,
+				 &port_vlan_cmd, sizeof(port_vlan_cmd),
+				 NULL, NULL);
+}
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_port.h b/drivers/net/ethernet/huawei/hinic/hinic_port.h
new file mode 100644
index 0000000..4cafb94
--- /dev/null
+++ b/drivers/net/ethernet/huawei/hinic/hinic_port.h
@@ -0,0 +1,68 @@
+/*
+ * Huawei HiNIC PCI Express Linux driver
+ * Copyright(c) 2017 Huawei Technologies Co., Ltd
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * for more details.
+ *
+ */
+
+#ifndef HINIC_PORT_H
+#define HINIC_PORT_H
+
+#include <linux/types.h>
+#include <linux/etherdevice.h>
+
+#include "hinic_dev.h"
+
+struct hinic_port_mac_cmd {
+	u8              status;
+	u8              version;
+	u8              rsvd0[6];
+
+	u16             func_idx;
+	u16             vlan_id;
+	u16             rsvd1;
+	unsigned char   mac[ETH_ALEN];
+};
+
+struct hinic_port_mtu_cmd {
+	u8      status;
+	u8      version;
+	u8      rsvd0[6];
+
+	u16     func_idx;
+	u16     rsvd1;
+	u32     mtu;
+};
+
+struct hinic_port_vlan_cmd {
+	u8      status;
+	u8      version;
+	u8      rsvd0[6];
+
+	u16     func_idx;
+	u16     vlan_id;
+};
+
+int hinic_port_add_mac(struct hinic_dev *nic_dev, const u8 *addr,
+		       u16 vlan_id);
+
+int hinic_port_del_mac(struct hinic_dev *nic_dev, const u8 *addr,
+		       u16 vlan_id);
+
+int hinic_port_get_mac(struct hinic_dev *nic_dev, u8 *addr);
+
+int hinic_port_set_mtu(struct hinic_dev *nic_dev, int new_mtu);
+
+int hinic_port_add_vlan(struct hinic_dev *nic_dev, u16 vlan_id);
+
+int hinic_port_del_vlan(struct hinic_dev *nic_dev, u16 vlan_id);
+
+#endif
-- 
1.9.1

^ permalink raw reply related

* [PATCH V7 net-next 07/22] net-next/hinic: Add aeqs
From: Aviad Krawczyk @ 2017-08-17 19:04 UTC (permalink / raw)
  To: davem
  Cc: linux-kernel, netdev, bc.y, victor.gissin, aviad.krawczyk,
	zhaochen6, tony.qu
In-Reply-To: <cover.1502996306.git.aviad.krawczyk@huawei.com>

Handle aeq elements that are accumulated on the aeq by calling the
registered handler for the specific event.

Signed-off-by: Aviad Krawczyk <aviad.krawczyk@huawei.com>
Signed-off-by: Zhao Chen <zhaochen6@huawei.com>
---
 drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h |  49 +++
 drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.c | 464 ++++++++++++++++++++++-
 drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.h |  81 ++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_if.c  |  91 +++++
 drivers/net/ethernet/huawei/hinic/hinic_hw_if.h  |  46 +++
 5 files changed, 729 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h
index ebbf054..52eb89c 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_csr.h
@@ -65,4 +65,53 @@
 #define HINIC_CSR_API_CMD_STATUS_ADDR(idx)              \
 	(HINIC_CSR_API_CMD_BASE + 0x30 + (idx) * HINIC_CSR_API_CMD_STRIDE)
 
+/* MSI-X registers */
+#define HINIC_CSR_MSIX_CTRL_BASE                        0x2000
+#define HINIC_CSR_MSIX_CNT_BASE                         0x2004
+
+#define HINIC_CSR_MSIX_STRIDE                           0x8
+
+#define HINIC_CSR_MSIX_CTRL_ADDR(idx)                   \
+	(HINIC_CSR_MSIX_CTRL_BASE + (idx) * HINIC_CSR_MSIX_STRIDE)
+
+#define HINIC_CSR_MSIX_CNT_ADDR(idx)                    \
+	(HINIC_CSR_MSIX_CNT_BASE + (idx) * HINIC_CSR_MSIX_STRIDE)
+
+/* EQ registers */
+#define HINIC_AEQ_MTT_OFF_BASE_ADDR                     0x200
+
+#define HINIC_EQ_MTT_OFF_STRIDE                         0x40
+
+#define HINIC_CSR_AEQ_MTT_OFF(id)                       \
+	(HINIC_AEQ_MTT_OFF_BASE_ADDR + (id) * HINIC_EQ_MTT_OFF_STRIDE)
+
+#define HINIC_CSR_EQ_PAGE_OFF_STRIDE                    8
+
+#define HINIC_CSR_AEQ_HI_PHYS_ADDR_REG(q_id, pg_num)    \
+	(HINIC_CSR_AEQ_MTT_OFF(q_id) + \
+	 (pg_num) * HINIC_CSR_EQ_PAGE_OFF_STRIDE)
+
+#define HINIC_CSR_AEQ_LO_PHYS_ADDR_REG(q_id, pg_num)    \
+	(HINIC_CSR_AEQ_MTT_OFF(q_id) + \
+	 (pg_num) * HINIC_CSR_EQ_PAGE_OFF_STRIDE + 4)
+
+#define HINIC_AEQ_CTRL_0_ADDR_BASE                      0xE00
+#define HINIC_AEQ_CTRL_1_ADDR_BASE                      0xE04
+#define HINIC_AEQ_CONS_IDX_ADDR_BASE                    0xE08
+#define HINIC_AEQ_PROD_IDX_ADDR_BASE                    0xE0C
+
+#define HINIC_EQ_OFF_STRIDE                             0x80
+
+#define HINIC_CSR_AEQ_CTRL_0_ADDR(idx)                  \
+	(HINIC_AEQ_CTRL_0_ADDR_BASE + (idx) * HINIC_EQ_OFF_STRIDE)
+
+#define HINIC_CSR_AEQ_CTRL_1_ADDR(idx)                  \
+	(HINIC_AEQ_CTRL_1_ADDR_BASE + (idx) * HINIC_EQ_OFF_STRIDE)
+
+#define HINIC_CSR_AEQ_CONS_IDX_ADDR(idx)                \
+	(HINIC_AEQ_CONS_IDX_ADDR_BASE + (idx) * HINIC_EQ_OFF_STRIDE)
+
+#define HINIC_CSR_AEQ_PROD_IDX_ADDR(idx)                \
+	(HINIC_AEQ_PROD_IDX_ADDR_BASE + (idx) * HINIC_EQ_OFF_STRIDE)
+
 #endif
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.c
index a099d20..a53d5b3 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.c
@@ -13,17 +13,74 @@
  *
  */
 
+#include <linux/kernel.h>
 #include <linux/types.h>
 #include <linux/errno.h>
 #include <linux/pci.h>
 #include <linux/device.h>
 #include <linux/workqueue.h>
+#include <linux/interrupt.h>
+#include <linux/slab.h>
+#include <linux/dma-mapping.h>
+#include <linux/log2.h>
+#include <asm/byteorder.h>
+#include <asm/barrier.h>
 
+#include "hinic_hw_csr.h"
 #include "hinic_hw_if.h"
 #include "hinic_hw_eqs.h"
 
 #define HINIC_EQS_WQ_NAME                       "hinic_eqs"
 
+#define GET_EQ_NUM_PAGES(eq, pg_size)           \
+		(ALIGN((eq)->q_len * (eq)->elem_size, pg_size) / (pg_size))
+
+#define GET_EQ_NUM_ELEMS_IN_PG(eq, pg_size)     ((pg_size) / (eq)->elem_size)
+
+#define EQ_CONS_IDX_REG_ADDR(eq)        HINIC_CSR_AEQ_CONS_IDX_ADDR((eq)->q_id)
+#define EQ_PROD_IDX_REG_ADDR(eq)        HINIC_CSR_AEQ_PROD_IDX_ADDR((eq)->q_id)
+
+#define EQ_HI_PHYS_ADDR_REG(eq, pg_num) \
+		HINIC_CSR_AEQ_HI_PHYS_ADDR_REG((eq)->q_id, pg_num)
+
+#define EQ_LO_PHYS_ADDR_REG(eq, pg_num) \
+		HINIC_CSR_AEQ_LO_PHYS_ADDR_REG((eq)->q_id, pg_num)
+
+#define GET_EQ_ELEMENT(eq, idx)         \
+		((eq)->virt_addr[(idx) / (eq)->num_elem_in_pg] + \
+		 (((idx) & ((eq)->num_elem_in_pg - 1)) * (eq)->elem_size))
+
+#define GET_AEQ_ELEM(eq, idx)           ((struct hinic_aeq_elem *) \
+					GET_EQ_ELEMENT(eq, idx))
+
+#define GET_CURR_AEQ_ELEM(eq)           GET_AEQ_ELEM(eq, (eq)->cons_idx)
+
+#define PAGE_IN_4K(page_size)           ((page_size) >> 12)
+#define EQ_SET_HW_PAGE_SIZE_VAL(eq)     (ilog2(PAGE_IN_4K((eq)->page_size)))
+
+#define ELEMENT_SIZE_IN_32B(eq)         (((eq)->elem_size) >> 5)
+#define EQ_SET_HW_ELEM_SIZE_VAL(eq)     (ilog2(ELEMENT_SIZE_IN_32B(eq)))
+
+#define EQ_MAX_PAGES                    8
+
+#define aeq_to_aeqs(eq)                 \
+		container_of((eq) - (eq)->q_id, struct hinic_aeqs, aeq[0])
+
+#define work_to_aeq_work(work)          \
+		container_of(work, struct hinic_eq_work, work)
+
+#define DMA_ATTR_AEQ_DEFAULT            0
+
+enum eq_int_mode {
+	EQ_INT_MODE_ARMED,
+	EQ_INT_MODE_ALWAYS
+};
+
+enum eq_arm_state {
+	EQ_NOT_ARMED,
+	EQ_ARMED
+};
+
 /**
  * hinic_aeq_register_hw_cb - register AEQ callback for specific event
  * @aeqs: pointer to Async eqs of the chip
@@ -61,6 +118,326 @@ void hinic_aeq_unregister_hw_cb(struct hinic_aeqs *aeqs,
 	hwe_cb->hwe_handler = NULL;
 }
 
+static u8 eq_cons_idx_checksum_set(u32 val)
+{
+	u8 checksum = 0;
+	int idx;
+
+	for (idx = 0; idx < 32; idx += 4)
+		checksum ^= ((val >> idx) & 0xF);
+
+	return (checksum & 0xF);
+}
+
+/**
+ * eq_update_ci - update the HW cons idx of event queue
+ * @eq: the event queue to update the cons idx for
+ **/
+static void eq_update_ci(struct hinic_eq *eq)
+{
+	u32 val, addr = EQ_CONS_IDX_REG_ADDR(eq);
+
+	/* Read Modify Write */
+	val = hinic_hwif_read_reg(eq->hwif, addr);
+
+	val = HINIC_EQ_CI_CLEAR(val, IDX)       &
+	      HINIC_EQ_CI_CLEAR(val, WRAPPED)   &
+	      HINIC_EQ_CI_CLEAR(val, INT_ARMED) &
+	      HINIC_EQ_CI_CLEAR(val, XOR_CHKSUM);
+
+	val |= HINIC_EQ_CI_SET(eq->cons_idx, IDX)    |
+	       HINIC_EQ_CI_SET(eq->wrapped, WRAPPED) |
+	       HINIC_EQ_CI_SET(EQ_ARMED, INT_ARMED);
+
+	val |= HINIC_EQ_CI_SET(eq_cons_idx_checksum_set(val), XOR_CHKSUM);
+
+	hinic_hwif_write_reg(eq->hwif, addr, val);
+}
+
+/**
+ * aeq_irq_handler - handler for the AEQ event
+ * @eq: the Async Event Queue that received the event
+ **/
+static void aeq_irq_handler(struct hinic_eq *eq)
+{
+	struct hinic_aeqs *aeqs = aeq_to_aeqs(eq);
+	struct hinic_hwif *hwif = aeqs->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	struct hinic_aeq_elem *aeqe_curr;
+	struct hinic_hw_event_cb *hwe_cb;
+	enum hinic_aeq_type event;
+	unsigned long eqe_state;
+	u32 aeqe_desc;
+	int i, size;
+
+	for (i = 0; i < eq->q_len; i++) {
+		aeqe_curr = GET_CURR_AEQ_ELEM(eq);
+
+		/* Data in HW is in Big endian Format */
+		aeqe_desc = be32_to_cpu(aeqe_curr->desc);
+
+		/* HW toggles the wrapped bit, when it adds eq element */
+		if (HINIC_EQ_ELEM_DESC_GET(aeqe_desc, WRAPPED) == eq->wrapped)
+			break;
+
+		event = HINIC_EQ_ELEM_DESC_GET(aeqe_desc, TYPE);
+		if (event >= HINIC_MAX_AEQ_EVENTS) {
+			dev_err(&pdev->dev, "Unknown AEQ Event %d\n", event);
+			return;
+		}
+
+		if (!HINIC_EQ_ELEM_DESC_GET(aeqe_desc, SRC)) {
+			hwe_cb = &aeqs->hwe_cb[event];
+
+			size = HINIC_EQ_ELEM_DESC_GET(aeqe_desc, SIZE);
+
+			eqe_state = cmpxchg(&hwe_cb->hwe_state,
+					    HINIC_EQE_ENABLED,
+					    HINIC_EQE_ENABLED |
+					    HINIC_EQE_RUNNING);
+			if ((eqe_state == HINIC_EQE_ENABLED) &&
+			    (hwe_cb->hwe_handler))
+				hwe_cb->hwe_handler(hwe_cb->handle,
+						    aeqe_curr->data, size);
+			else
+				dev_err(&pdev->dev, "Unhandled AEQ Event %d\n",
+					event);
+
+			hwe_cb->hwe_state &= ~HINIC_EQE_RUNNING;
+		}
+
+		eq->cons_idx++;
+
+		if (eq->cons_idx == eq->q_len) {
+			eq->cons_idx = 0;
+			eq->wrapped = !eq->wrapped;
+		}
+	}
+}
+
+/**
+ * eq_irq_handler - handler for the EQ event
+ * @data: the Event Queue that received the event
+ **/
+static void eq_irq_handler(void *data)
+{
+	struct hinic_eq *eq = (struct hinic_eq *)data;
+
+	if (eq->type == HINIC_AEQ)
+		aeq_irq_handler(eq);
+
+	eq_update_ci(eq);
+}
+
+/**
+ * eq_irq_work - the work of the EQ that received the event
+ * @work: the work struct that is associated with the EQ
+ **/
+static void eq_irq_work(struct work_struct *work)
+{
+	struct hinic_eq_work *aeq_work = work_to_aeq_work(work);
+	struct hinic_eq *aeq;
+
+	aeq = (struct hinic_eq *)aeq_work->data;
+	eq_irq_handler(aeq);
+}
+
+/**
+ * aeq_interrupt - aeq interrupt handler
+ * @irq: irq number
+ * @data: the Async Event Queue that collected the event
+ **/
+static irqreturn_t aeq_interrupt(int irq, void *data)
+{
+	struct hinic_eq *aeq = (struct hinic_eq *)data;
+	struct hinic_aeqs *aeqs = aeq_to_aeqs(aeq);
+	struct hinic_eq_work *aeq_work;
+
+	/* clear resend timer cnt register */
+	hinic_msix_attr_cnt_clear(aeq->hwif, aeq->msix_entry.entry);
+
+	aeq_work = &aeq->aeq_work;
+	aeq_work->data = aeq;
+
+	queue_work(aeqs->workq, &aeq_work->work);
+
+	return IRQ_HANDLED;
+}
+
+void set_ctrl0(struct hinic_eq *eq)
+{
+	struct msix_entry *msix_entry = &eq->msix_entry;
+	struct  hinic_hwif *hwif = eq->hwif;
+	enum hinic_eq_type type = eq->type;
+	u32 addr, val, ctrl0;
+
+	if (type == HINIC_AEQ) {
+		/* RMW Ctrl0 */
+		addr = HINIC_CSR_AEQ_CTRL_0_ADDR(eq->q_id);
+
+		val = hinic_hwif_read_reg(hwif, addr);
+
+		val = HINIC_AEQ_CTRL_0_CLEAR(val, INT_IDX)      &
+		      HINIC_AEQ_CTRL_0_CLEAR(val, DMA_ATTR)     &
+		      HINIC_AEQ_CTRL_0_CLEAR(val, PCI_INTF_IDX) &
+		      HINIC_AEQ_CTRL_0_CLEAR(val, INT_MODE);
+
+		ctrl0 = HINIC_AEQ_CTRL_0_SET(msix_entry->entry, INT_IDX)     |
+			HINIC_AEQ_CTRL_0_SET(DMA_ATTR_AEQ_DEFAULT, DMA_ATTR) |
+			HINIC_AEQ_CTRL_0_SET(HINIC_HWIF_PCI_INTF(hwif),
+					     PCI_INTF_IDX)                   |
+			HINIC_AEQ_CTRL_0_SET(EQ_INT_MODE_ARMED, INT_MODE);
+
+		val |= ctrl0;
+
+		hinic_hwif_write_reg(hwif, addr, val);
+	}
+}
+
+void set_ctrl1(struct hinic_eq *eq)
+{
+	struct  hinic_hwif *hwif = eq->hwif;
+	enum hinic_eq_type type = eq->type;
+	u32 page_size_val, elem_size;
+	u32 addr, val, ctrl1;
+
+	if (type == HINIC_AEQ) {
+		/* RMW Ctrl1 */
+		addr = HINIC_CSR_AEQ_CTRL_1_ADDR(eq->q_id);
+
+		page_size_val = EQ_SET_HW_PAGE_SIZE_VAL(eq);
+		elem_size = EQ_SET_HW_ELEM_SIZE_VAL(eq);
+
+		val = hinic_hwif_read_reg(hwif, addr);
+
+		val = HINIC_AEQ_CTRL_1_CLEAR(val, LEN)          &
+		      HINIC_AEQ_CTRL_1_CLEAR(val, ELEM_SIZE)    &
+		      HINIC_AEQ_CTRL_1_CLEAR(val, PAGE_SIZE);
+
+		ctrl1 = HINIC_AEQ_CTRL_1_SET(eq->q_len, LEN)            |
+			HINIC_AEQ_CTRL_1_SET(elem_size, ELEM_SIZE)      |
+			HINIC_AEQ_CTRL_1_SET(page_size_val, PAGE_SIZE);
+
+		val |= ctrl1;
+
+		hinic_hwif_write_reg(hwif, addr, val);
+	}
+}
+
+/**
+ * set_eq_ctrls - setting eq's ctrl registers
+ * @eq: the Event Queue for setting
+ **/
+static void set_eq_ctrls(struct hinic_eq *eq)
+{
+	set_ctrl0(eq);
+	set_ctrl1(eq);
+}
+
+/**
+ * aeq_elements_init - initialize all the elements in the aeq
+ * @eq: the Async Event Queue
+ * @init_val: value to initialize the elements with it
+ **/
+static void aeq_elements_init(struct hinic_eq *eq, u32 init_val)
+{
+	struct hinic_aeq_elem *aeqe;
+	int i;
+
+	for (i = 0; i < eq->q_len; i++) {
+		aeqe = GET_AEQ_ELEM(eq, i);
+		aeqe->desc = cpu_to_be32(init_val);
+	}
+
+	wmb();  /* Write the initilzation values */
+}
+
+/**
+ * alloc_eq_pages - allocate the pages for the queue
+ * @eq: the event queue
+ *
+ * Return 0 - Success, Negative - Failure
+ **/
+static int alloc_eq_pages(struct hinic_eq *eq)
+{
+	struct hinic_hwif *hwif = eq->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	u32 init_val, addr, val;
+	size_t addr_size;
+	int err, pg;
+
+	addr_size = eq->num_pages * sizeof(*eq->dma_addr);
+	eq->dma_addr = devm_kzalloc(&pdev->dev, addr_size, GFP_KERNEL);
+	if (!eq->dma_addr)
+		return -ENOMEM;
+
+	addr_size = eq->num_pages * sizeof(*eq->virt_addr);
+	eq->virt_addr = devm_kzalloc(&pdev->dev, addr_size, GFP_KERNEL);
+	if (!eq->virt_addr) {
+		err = -ENOMEM;
+		goto err_virt_addr_alloc;
+	}
+
+	for (pg = 0; pg < eq->num_pages; pg++) {
+		eq->virt_addr[pg] = dma_zalloc_coherent(&pdev->dev,
+							eq->page_size,
+							&eq->dma_addr[pg],
+							GFP_KERNEL);
+		if (!eq->virt_addr[pg]) {
+			err = -ENOMEM;
+			goto err_dma_alloc;
+		}
+
+		addr = EQ_HI_PHYS_ADDR_REG(eq, pg);
+		val = upper_32_bits(eq->dma_addr[pg]);
+
+		hinic_hwif_write_reg(hwif, addr, val);
+
+		addr = EQ_LO_PHYS_ADDR_REG(eq, pg);
+		val = lower_32_bits(eq->dma_addr[pg]);
+
+		hinic_hwif_write_reg(hwif, addr, val);
+	}
+
+	init_val = HINIC_EQ_ELEM_DESC_SET(eq->wrapped, WRAPPED);
+
+	if (eq->type == HINIC_AEQ)
+		aeq_elements_init(eq, init_val);
+
+	return 0;
+
+err_dma_alloc:
+	while (--pg >= 0)
+		dma_free_coherent(&pdev->dev, eq->page_size,
+				  eq->virt_addr[pg],
+				  eq->dma_addr[pg]);
+
+	devm_kfree(&pdev->dev, eq->virt_addr);
+
+err_virt_addr_alloc:
+	devm_kfree(&pdev->dev, eq->dma_addr);
+	return err;
+}
+
+/**
+ * free_eq_pages - free the pages of the queue
+ * @eq: the Event Queue
+ **/
+static void free_eq_pages(struct hinic_eq *eq)
+{
+	struct hinic_hwif *hwif = eq->hwif;
+	struct pci_dev *pdev = hwif->pdev;
+	int pg;
+
+	for (pg = 0; pg < eq->num_pages; pg++)
+		dma_free_coherent(&pdev->dev, eq->page_size,
+				  eq->virt_addr[pg],
+				  eq->dma_addr[pg]);
+
+	devm_kfree(&pdev->dev, eq->virt_addr);
+	devm_kfree(&pdev->dev, eq->dma_addr);
+}
+
 /**
  * init_eq - initialize Event Queue
  * @eq: the event queue
@@ -77,8 +454,81 @@ static int init_eq(struct hinic_eq *eq, struct hinic_hwif *hwif,
 		   enum hinic_eq_type type, int q_id, u32 q_len, u32 page_size,
 		   struct msix_entry entry)
 {
-	/* should be implemented */
+	struct pci_dev *pdev = hwif->pdev;
+	int err;
+
+	eq->hwif = hwif;
+	eq->type = type;
+	eq->q_id = q_id;
+	eq->q_len = q_len;
+	eq->page_size = page_size;
+
+	/* Clear PI and CI, also clear the ARM bit */
+	hinic_hwif_write_reg(eq->hwif, EQ_CONS_IDX_REG_ADDR(eq), 0);
+	hinic_hwif_write_reg(eq->hwif, EQ_PROD_IDX_REG_ADDR(eq), 0);
+
+	eq->cons_idx = 0;
+	eq->wrapped = 0;
+
+	if (type == HINIC_AEQ) {
+		eq->elem_size = HINIC_AEQE_SIZE;
+	} else {
+		dev_err(&pdev->dev, "Invalid EQ type\n");
+		return -EINVAL;
+	}
+
+	eq->num_pages = GET_EQ_NUM_PAGES(eq, page_size);
+	eq->num_elem_in_pg = GET_EQ_NUM_ELEMS_IN_PG(eq, page_size);
+
+	eq->msix_entry = entry;
+
+	if (eq->num_elem_in_pg & (eq->num_elem_in_pg - 1)) {
+		dev_err(&pdev->dev, "num elements in eq page != power of 2\n");
+		return -EINVAL;
+	}
+
+	if (eq->num_pages > EQ_MAX_PAGES) {
+		dev_err(&pdev->dev, "too many pages for eq\n");
+		return -EINVAL;
+	}
+
+	set_eq_ctrls(eq);
+	eq_update_ci(eq);
+
+	err = alloc_eq_pages(eq);
+	if (err) {
+		dev_err(&pdev->dev, "Failed to allocate pages for eq\n");
+		return err;
+	}
+
+	if (type == HINIC_AEQ) {
+		struct hinic_eq_work *aeq_work = &eq->aeq_work;
+
+		INIT_WORK(&aeq_work->work, eq_irq_work);
+	}
+
+	/* set the attributes of the msix entry */
+	hinic_msix_attr_set(eq->hwif, eq->msix_entry.entry,
+			    HINIC_EQ_MSIX_PENDING_LIMIT_DEFAULT,
+			    HINIC_EQ_MSIX_COALESC_TIMER_DEFAULT,
+			    HINIC_EQ_MSIX_LLI_TIMER_DEFAULT,
+			    HINIC_EQ_MSIX_LLI_CREDIT_LIMIT_DEFAULT,
+			    HINIC_EQ_MSIX_RESEND_TIMER_DEFAULT);
+
+	if (type == HINIC_AEQ)
+		err = request_irq(entry.vector, aeq_interrupt, 0,
+				  "hinic_aeq", eq);
+
+	if (err) {
+		dev_err(&pdev->dev, "Failed to request irq for the EQ\n");
+		goto err_req_irq;
+	}
+
 	return 0;
+
+err_req_irq:
+	free_eq_pages(eq);
+	return err;
 }
 
 /**
@@ -87,7 +537,17 @@ static int init_eq(struct hinic_eq *eq, struct hinic_hwif *hwif,
  **/
 static void remove_eq(struct hinic_eq *eq)
 {
-	/* should be implemented */
+	struct msix_entry *entry = &eq->msix_entry;
+
+	free_irq(entry->vector, eq);
+
+	if (eq->type == HINIC_AEQ) {
+		struct hinic_eq_work *aeq_work = &eq->aeq_work;
+
+		cancel_work_sync(&aeq_work->work);
+	}
+
+	free_eq_pages(eq);
 }
 
 /**
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.h
index 1580127..7f50b2f 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_eqs.h
@@ -24,8 +24,84 @@
 
 #include "hinic_hw_if.h"
 
+#define HINIC_AEQ_CTRL_0_INT_IDX_SHIFT          0
+#define HINIC_AEQ_CTRL_0_DMA_ATTR_SHIFT         12
+#define HINIC_AEQ_CTRL_0_PCI_INTF_IDX_SHIFT     20
+#define HINIC_AEQ_CTRL_0_INT_MODE_SHIFT         31
+
+#define HINIC_AEQ_CTRL_0_INT_IDX_MASK           0x3FF
+#define HINIC_AEQ_CTRL_0_DMA_ATTR_MASK          0x3F
+#define HINIC_AEQ_CTRL_0_PCI_INTF_IDX_MASK      0x3
+#define HINIC_AEQ_CTRL_0_INT_MODE_MASK          0x1
+
+#define HINIC_AEQ_CTRL_0_SET(val, member)       \
+			(((u32)(val) & HINIC_AEQ_CTRL_0_##member##_MASK) << \
+			 HINIC_AEQ_CTRL_0_##member##_SHIFT)
+
+#define HINIC_AEQ_CTRL_0_CLEAR(val, member)     \
+			((val) & (~(HINIC_AEQ_CTRL_0_##member##_MASK \
+			 << HINIC_AEQ_CTRL_0_##member##_SHIFT)))
+
+#define HINIC_AEQ_CTRL_1_LEN_SHIFT              0
+#define HINIC_AEQ_CTRL_1_ELEM_SIZE_SHIFT        24
+#define HINIC_AEQ_CTRL_1_PAGE_SIZE_SHIFT        28
+
+#define HINIC_AEQ_CTRL_1_LEN_MASK               0x1FFFFF
+#define HINIC_AEQ_CTRL_1_ELEM_SIZE_MASK         0x3
+#define HINIC_AEQ_CTRL_1_PAGE_SIZE_MASK         0xF
+
+#define HINIC_AEQ_CTRL_1_SET(val, member)       \
+			(((u32)(val) & HINIC_AEQ_CTRL_1_##member##_MASK) << \
+			 HINIC_AEQ_CTRL_1_##member##_SHIFT)
+
+#define HINIC_AEQ_CTRL_1_CLEAR(val, member)     \
+			((val) & (~(HINIC_AEQ_CTRL_1_##member##_MASK \
+			 << HINIC_AEQ_CTRL_1_##member##_SHIFT)))
+
+#define HINIC_EQ_ELEM_DESC_TYPE_SHIFT           0
+#define HINIC_EQ_ELEM_DESC_SRC_SHIFT            7
+#define HINIC_EQ_ELEM_DESC_SIZE_SHIFT           8
+#define HINIC_EQ_ELEM_DESC_WRAPPED_SHIFT        31
+
+#define HINIC_EQ_ELEM_DESC_TYPE_MASK            0x7F
+#define HINIC_EQ_ELEM_DESC_SRC_MASK             0x1
+#define HINIC_EQ_ELEM_DESC_SIZE_MASK            0xFF
+#define HINIC_EQ_ELEM_DESC_WRAPPED_MASK         0x1
+
+#define HINIC_EQ_ELEM_DESC_SET(val, member)     \
+			(((u32)(val) & HINIC_EQ_ELEM_DESC_##member##_MASK) << \
+			 HINIC_EQ_ELEM_DESC_##member##_SHIFT)
+
+#define HINIC_EQ_ELEM_DESC_GET(val, member)     \
+			(((val) >> HINIC_EQ_ELEM_DESC_##member##_SHIFT) & \
+			 HINIC_EQ_ELEM_DESC_##member##_MASK)
+
+#define HINIC_EQ_CI_IDX_SHIFT                   0
+#define HINIC_EQ_CI_WRAPPED_SHIFT               20
+#define HINIC_EQ_CI_XOR_CHKSUM_SHIFT            24
+#define HINIC_EQ_CI_INT_ARMED_SHIFT             31
+
+#define HINIC_EQ_CI_IDX_MASK                    0xFFFFF
+#define HINIC_EQ_CI_WRAPPED_MASK                0x1
+#define HINIC_EQ_CI_XOR_CHKSUM_MASK             0xF
+#define HINIC_EQ_CI_INT_ARMED_MASK              0x1
+
+#define HINIC_EQ_CI_SET(val, member)            \
+			(((u32)(val) & HINIC_EQ_CI_##member##_MASK) << \
+			 HINIC_EQ_CI_##member##_SHIFT)
+
+#define HINIC_EQ_CI_CLEAR(val, member)          \
+			((val) & (~(HINIC_EQ_CI_##member##_MASK \
+			 << HINIC_EQ_CI_##member##_SHIFT)))
+
 #define HINIC_MAX_AEQS                  4
 
+#define HINIC_AEQE_SIZE                 64
+
+#define HINIC_AEQE_DESC_SIZE            4
+#define HINIC_AEQE_DATA_SIZE            \
+			(HINIC_AEQE_SIZE - HINIC_AEQE_DESC_SIZE)
+
 #define HINIC_DEFAULT_AEQ_LEN           64
 
 #define HINIC_EQ_PAGE_SIZE              SZ_4K
@@ -45,6 +121,11 @@ enum hinic_eqe_state {
 	HINIC_EQE_RUNNING = BIT(1),
 };
 
+struct hinic_aeq_elem {
+	u8      data[HINIC_AEQE_DATA_SIZE];
+	u32     desc;
+};
+
 struct hinic_eq_work {
 	struct work_struct      work;
 	void                    *data;
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_if.c b/drivers/net/ethernet/huawei/hinic/hinic_hw_if.c
index edf1842..2b9410c 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_if.c
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_if.c
@@ -25,6 +25,97 @@
 
 #define PCIE_ATTR_ENTRY         0
 
+#define VALID_MSIX_IDX(attr, msix_index) ((msix_index) < (attr)->num_irqs)
+
+/**
+ * hinic_msix_attr_set - set message attribute for msix entry
+ * @hwif: the HW interface of a pci function device
+ * @msix_index: msix_index
+ * @pending_limit: the maximum pending interrupt events (unit 8)
+ * @coalesc_timer: coalesc period for interrupt (unit 8 us)
+ * @lli_timer: replenishing period for low latency credit (unit 8 us)
+ * @lli_credit_limit: maximum credits for low latency msix messages (unit 8)
+ * @resend_timer: maximum wait for resending msix (unit coalesc period)
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_msix_attr_set(struct hinic_hwif *hwif, u16 msix_index,
+			u8 pending_limit, u8 coalesc_timer,
+			u8 lli_timer, u8 lli_credit_limit,
+			u8 resend_timer)
+{
+	u32 msix_ctrl, addr;
+
+	if (!VALID_MSIX_IDX(&hwif->attr, msix_index))
+		return -EINVAL;
+
+	msix_ctrl = HINIC_MSIX_ATTR_SET(pending_limit, PENDING_LIMIT)   |
+		    HINIC_MSIX_ATTR_SET(coalesc_timer, COALESC_TIMER)   |
+		    HINIC_MSIX_ATTR_SET(lli_timer, LLI_TIMER)           |
+		    HINIC_MSIX_ATTR_SET(lli_credit_limit, LLI_CREDIT)   |
+		    HINIC_MSIX_ATTR_SET(resend_timer, RESEND_TIMER);
+
+	addr = HINIC_CSR_MSIX_CTRL_ADDR(msix_index);
+
+	hinic_hwif_write_reg(hwif, addr, msix_ctrl);
+	return 0;
+}
+
+/**
+ * hinic_msix_attr_get - get message attribute of msix entry
+ * @hwif: the HW interface of a pci function device
+ * @msix_index: msix_index
+ * @pending_limit: the maximum pending interrupt events (unit 8)
+ * @coalesc_timer: coalesc period for interrupt (unit 8 us)
+ * @lli_timer: replenishing period for low latency credit (unit 8 us)
+ * @lli_credit_limit: maximum credits for low latency msix messages (unit 8)
+ * @resend_timer: maximum wait for resending msix (unit coalesc period)
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_msix_attr_get(struct hinic_hwif *hwif, u16 msix_index,
+			u8 *pending_limit, u8 *coalesc_timer,
+			u8 *lli_timer, u8 *lli_credit_limit,
+			u8 *resend_timer)
+{
+	u32 addr, val;
+
+	if (!VALID_MSIX_IDX(&hwif->attr, msix_index))
+		return -EINVAL;
+
+	addr = HINIC_CSR_MSIX_CTRL_ADDR(msix_index);
+	val  = hinic_hwif_read_reg(hwif, addr);
+
+	*pending_limit    = HINIC_MSIX_ATTR_GET(val, PENDING_LIMIT);
+	*coalesc_timer    = HINIC_MSIX_ATTR_GET(val, COALESC_TIMER);
+	*lli_timer        = HINIC_MSIX_ATTR_GET(val, LLI_TIMER);
+	*lli_credit_limit = HINIC_MSIX_ATTR_GET(val, LLI_CREDIT);
+	*resend_timer     = HINIC_MSIX_ATTR_GET(val, RESEND_TIMER);
+	return 0;
+}
+
+/**
+ * hinic_msix_attr_cnt_clear - clear message attribute counters for msix entry
+ * @hwif: the HW interface of a pci function device
+ * @msix_index: msix_index
+ *
+ * Return 0 - Success, negative - Failure
+ **/
+int hinic_msix_attr_cnt_clear(struct hinic_hwif *hwif, u16 msix_index)
+{
+	u32 msix_ctrl, addr;
+
+	if (!VALID_MSIX_IDX(&hwif->attr, msix_index))
+		return -EINVAL;
+
+	msix_ctrl = HINIC_MSIX_CNT_SET(1, RESEND_TIMER);
+
+	addr = HINIC_CSR_MSIX_CNT_ADDR(msix_index);
+
+	hinic_hwif_write_reg(hwif, addr, msix_ctrl);
+	return 0;
+}
+
 /**
  * hwif_ready - test if the HW is ready for use
  * @hwif: the HW interface of a pci function device
diff --git a/drivers/net/ethernet/huawei/hinic/hinic_hw_if.h b/drivers/net/ethernet/huawei/hinic/hinic_hw_if.h
index 98623d6..707a046 100644
--- a/drivers/net/ethernet/huawei/hinic/hinic_hw_if.h
+++ b/drivers/net/ethernet/huawei/hinic/hinic_hw_if.h
@@ -88,6 +88,34 @@
 	((val) & (~(HINIC_PPF_ELECTION_##member##_MASK          \
 	 << HINIC_PPF_ELECTION_##member##_SHIFT)))
 
+#define HINIC_MSIX_PENDING_LIMIT_SHIFT                          0
+#define HINIC_MSIX_COALESC_TIMER_SHIFT                          8
+#define HINIC_MSIX_LLI_TIMER_SHIFT                              16
+#define HINIC_MSIX_LLI_CREDIT_SHIFT                             24
+#define HINIC_MSIX_RESEND_TIMER_SHIFT                           29
+
+#define HINIC_MSIX_PENDING_LIMIT_MASK                           0xFF
+#define HINIC_MSIX_COALESC_TIMER_MASK                           0xFF
+#define HINIC_MSIX_LLI_TIMER_MASK                               0xFF
+#define HINIC_MSIX_LLI_CREDIT_MASK                              0x1F
+#define HINIC_MSIX_RESEND_TIMER_MASK                            0x7
+
+#define HINIC_MSIX_ATTR_SET(val, member)                        \
+	(((u32)(val) & HINIC_MSIX_##member##_MASK) <<           \
+	 HINIC_MSIX_##member##_SHIFT)
+
+#define HINIC_MSIX_ATTR_GET(val, member)                        \
+	(((val) >> HINIC_MSIX_##member##_SHIFT) &               \
+	 HINIC_MSIX_##member##_MASK)
+
+#define HINIC_MSIX_CNT_RESEND_TIMER_SHIFT                       29
+
+#define HINIC_MSIX_CNT_RESEND_TIMER_MASK                        0x1
+
+#define HINIC_MSIX_CNT_SET(val, member)                         \
+	(((u32)(val) & HINIC_MSIX_CNT_##member##_MASK) <<       \
+	 HINIC_MSIX_CNT_##member##_SHIFT)
+
 #define HINIC_HWIF_NUM_AEQS(hwif)       ((hwif)->attr.num_aeqs)
 #define HINIC_HWIF_NUM_CEQS(hwif)       ((hwif)->attr.num_ceqs)
 #define HINIC_HWIF_NUM_IRQS(hwif)       ((hwif)->attr.num_irqs)
@@ -105,6 +133,12 @@
 #define HINIC_PCIE_AT_DISABLE           0
 #define HINIC_PCIE_PH_DISABLE           0
 
+#define HINIC_EQ_MSIX_PENDING_LIMIT_DEFAULT     0       /* Disabled */
+#define HINIC_EQ_MSIX_COALESC_TIMER_DEFAULT     0xFF    /* max */
+#define HINIC_EQ_MSIX_LLI_TIMER_DEFAULT         0       /* Disabled */
+#define HINIC_EQ_MSIX_LLI_CREDIT_LIMIT_DEFAULT  0       /* Disabled */
+#define HINIC_EQ_MSIX_RESEND_TIMER_DEFAULT      7       /* max */
+
 enum hinic_pcie_nosnoop {
 	HINIC_PCIE_SNOOP        = 0,
 	HINIC_PCIE_NO_SNOOP     = 1,
@@ -166,6 +200,18 @@ static inline void hinic_hwif_write_reg(struct hinic_hwif *hwif, u32 reg,
 	writel(cpu_to_be32(val), hwif->cfg_regs_bar + reg);
 }
 
+int hinic_msix_attr_set(struct hinic_hwif *hwif, u16 msix_index,
+			u8 pending_limit, u8 coalesc_timer,
+			u8 lli_timer_cfg, u8 lli_credit_limit,
+			u8 resend_timer);
+
+int hinic_msix_attr_get(struct hinic_hwif *hwif, u16 msix_index,
+			u8 *pending_limit, u8 *coalesc_timer_cfg,
+			u8 *lli_timer, u8 *lli_credit_limit,
+			u8 *resend_timer);
+
+int hinic_msix_attr_cnt_clear(struct hinic_hwif *hwif, u16 msix_index);
+
 int hinic_init_hwif(struct hinic_hwif *hwif, struct pci_dev *pdev);
 
 void hinic_free_hwif(struct hinic_hwif *hwif);
-- 
1.9.1

^ permalink raw reply related


This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox