* [PATCH net-next v3 3/5] ethtool: (uapi) Add ETHTOOL_PHY_DOWNSHIFT to PHY tunables
From: Allan W. Nielsen @ 2016-11-15 10:20 UTC (permalink / raw)
To: netdev; +Cc: andrew, raju.lakkaraju, allan.nielsen, Raju Lakkaraju
In-Reply-To: <1479205204-27768-1-git-send-email-allan.nielsen@microsemi.com>
From: Raju Lakkaraju <Raju.Lakkaraju@microsemi.com>
For operation in cabling environments that are incompatible with
1000BASE-T, PHY device may provide an automatic link speed downshift
operation. When enabled, the device automatically changes its 1000BASE-T
auto-negotiation to the next slower speed after a configured number of
failed attempts at 1000BASE-T. This feature is useful in setting up in
networks using older cable installations that include only pairs A and B,
and not pairs C and D.
Signed-off-by: Raju Lakkaraju <Raju.Lakkaraju@microsemi.com>
Signed-off-by: Allan W. Nielsen <allan.nielsen@microsemi.com>
---
include/uapi/linux/ethtool.h | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/include/uapi/linux/ethtool.h b/include/uapi/linux/ethtool.h
index 42f696f..f0db778 100644
--- a/include/uapi/linux/ethtool.h
+++ b/include/uapi/linux/ethtool.h
@@ -248,9 +248,12 @@ struct ethtool_tunable {
void *data[0];
};
+#define DOWNSHIFT_DEV_DEFAULT_COUNT 0xff
+#define DOWNSHIFT_DEV_DISABLE 0
+
enum phy_tunable_id {
ETHTOOL_PHY_ID_UNSPEC,
-
+ ETHTOOL_PHY_DOWNSHIFT,
/*
* Add your fresh new phy tunable attribute above and remember to update
* phy_tunable_strings[] in net/core/ethtool.c
--
2.7.3
^ permalink raw reply related
* [PATCH net-next v3 1/5] ethtool: (uapi) Add ETHTOOL_PHY_GTUNABLE and ETHTOOL_PHY_STUNABLE
From: Allan W. Nielsen @ 2016-11-15 10:20 UTC (permalink / raw)
To: netdev; +Cc: andrew, raju.lakkaraju, allan.nielsen, Raju Lakkaraju
In-Reply-To: <1479205204-27768-1-git-send-email-allan.nielsen@microsemi.com>
From: Raju Lakkaraju <Raju.Lakkaraju@microsemi.com>
Defines a generic API to get/set phy tunables. The API is using the
existing ethtool_tunable/tunable_type_id types which is already being used
for mac level tunables.
Signed-off-by: Raju Lakkaraju <Raju.Lakkaraju@microsemi.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Allan W. Nielsen <allan.nielsen@microsemi.com>
---
include/uapi/linux/ethtool.h | 15 ++++++++++++++-
1 file changed, 14 insertions(+), 1 deletion(-)
diff --git a/include/uapi/linux/ethtool.h b/include/uapi/linux/ethtool.h
index 8e54723..42f696f 100644
--- a/include/uapi/linux/ethtool.h
+++ b/include/uapi/linux/ethtool.h
@@ -248,6 +248,16 @@ struct ethtool_tunable {
void *data[0];
};
+enum phy_tunable_id {
+ ETHTOOL_PHY_ID_UNSPEC,
+
+ /*
+ * Add your fresh new phy tunable attribute above and remember to update
+ * phy_tunable_strings[] in net/core/ethtool.c
+ */
+ __ETHTOOL_PHY_TUNABLE_COUNT,
+};
+
/**
* struct ethtool_regs - hardware register dump
* @cmd: Command number = %ETHTOOL_GREGS
@@ -548,6 +558,7 @@ struct ethtool_pauseparam {
* @ETH_SS_FEATURES: Device feature names
* @ETH_SS_RSS_HASH_FUNCS: RSS hush function names
* @ETH_SS_PHY_STATS: Statistic names, for use with %ETHTOOL_GPHYSTATS
+ * @ETH_SS_PHY_TUNABLES: PHY tunable names
*/
enum ethtool_stringset {
ETH_SS_TEST = 0,
@@ -558,6 +569,7 @@ enum ethtool_stringset {
ETH_SS_RSS_HASH_FUNCS,
ETH_SS_TUNABLES,
ETH_SS_PHY_STATS,
+ ETH_SS_PHY_TUNABLES,
};
/**
@@ -1313,7 +1325,8 @@ struct ethtool_per_queue_op {
#define ETHTOOL_GLINKSETTINGS 0x0000004c /* Get ethtool_link_settings */
#define ETHTOOL_SLINKSETTINGS 0x0000004d /* Set ethtool_link_settings */
-
+#define ETHTOOL_PHY_GTUNABLE 0x0000004e /* Get PHY tunable configuration */
+#define ETHTOOL_PHY_STUNABLE 0x0000004f /* Set PHY tunable configuration */
/* compatibility with older code */
#define SPARC_ETH_GSET ETHTOOL_GSET
--
2.7.3
^ permalink raw reply related
* [PATCH net-next v3 0/5] Adding PHY-Tunables and downshift support
From: Allan W. Nielsen @ 2016-11-15 10:19 UTC (permalink / raw)
To: netdev; +Cc: andrew, raju.lakkaraju, allan.nielsen
Hi All,
This follow fixes the comments provided on the v2 series.
Following is the list of changes.
- Spelling in "net: phy: Add downshift get/set support in Microsemi PHYs driver"
Ethtool patches will follow shortly.
Old cover letters included below.
Please review.
Best regards
Allan and Raju
Raju Lakkaraju (5):
ethtool: (uapi) Add ETHTOOL_PHY_GTUNABLE and ETHTOOL_PHY_STUNABLE
ethtool: Implements ETHTOOL_PHY_GTUNABLE/ETHTOOL_PHY_STUNABLE
ethtool: (uapi) Add ETHTOOL_PHY_DOWNSHIFT to PHY tunables
ethtool: Core impl for ETHTOOL_PHY_DOWNSHIFT tunable
net: phy: Add downshift get/set support in Microsemi PHYs driver
drivers/net/phy/mscc.c | 100 +++++++++++++++++++++++++++++++++++++++++++
include/linux/phy.h | 7 +++
include/uapi/linux/ethtool.h | 18 +++++++-
net/core/ethtool.c | 93 ++++++++++++++++++++++++++++++++++++++++
4 files changed, 217 insertions(+), 1 deletion(-)
> From 5ca4dac36f45e097b93ae4cfdf35a8922e5d975e Mon Sep 17 00:00:00 2001
> prom: "Allan W. Nielsen" <allan.nielsen@microsemi.com>
> Date: Mon, 14 Nov 2016 10:04:07 +0100
> Subject: [PATCH net-next v2 0/5] Adding PHY-Tunables and downshift support
>
> Hi,
>
> This is a follow-up on the patch series posted at Fri, 4 Nov 2016 10:55:25
> +0100 (old cover letter included below).
>
> The following is changed/added/addressed:
> - Support for __ethtool_get_strings()/phy_tunables is added
> - Using DOWNSHIFT_DEV_DISABLE define instead of '0' when disabling
> - If downshifting clould not be eanbled with the requested count, then a
> trace message with a hist is printed.
> - Using ERANGE as requested by Andrew.
> - Andrew suggested to use ENOSUPP when a tunable is not implemented We kept
> the EINVAL because ENOSUPP does not exists, and the existing MAC-level
> tunable is also using EINVAL in the same scenario.
> - Andrew suggested that the driver should accept a count of 1 and "just"
> round it up to 2 (the nearest allowed value). We discussed it a bit (Raju
> and I) and preferred to return EINVAL when a configuration value could
> not be accommodated. People which wants "help" to choose a supported
> value should be using the DOWNSHIFT_DEV_DEFAULT_COUNT value.
> - (ethtool) Syntax is changed from "--set-phy-tunable downshift on|off|%d"
> to "--set-phy-tunable [downshift on|off [count N]]" - as requested by
> Andrew.
>
> Ethtool patches will follow shortly.
>
> Old cover letter:
> > From 1a318266822f5d74c58b5219ebbdf5a6a5f567dc Mon Sep 17 00:00:00 2001
> > From: "Allan W. Nielsen" <allan.nielsen@microsemi.com>
> > Date: Fri, 4 Nov 2016 10:55:25 +0100
> > Subject: [PATCH net-next 0/5] Adding PHY-Tunables and downshift support
> >
> > Hi All,
> >
> > This series add support for PHY tunables, and uses this facility to configure
> > downshifting. The downshifting mechanism is implemented for MSCC phys.
> >
> > This series tries to address the comments provided back in mid October when this
> > feature was posted along with fast-link-failure. Fast-link-failure has been
> > separated out, but we would like to pick continue on that if/when we agree on
> > how the phy-tunables and downshifting should be done.
> >
> > The proposed generic interface is similar to ETHTOOL_GTUNABLE/ETHTOOL_STUNABLE,
> > it uses the same type (ethtool_tunable/tunable_type_id) but a new enum
> > (phy_tunable_id) is added to reflect the PHY tunable.
> >
> > The implementation just call the newly added function pointers in
> > get_tunable/set_tunable phy_device structure.
> >
> > To configure downshifting, the ethtool_tunable structure is used. 'id' must be
> > set to 'ETHTOOL_PHY_DOWNSHIFT', 'type_id' must be set to 'ETHTOOL_TUNABLE_U8'
> > and 'data' value configure the amount of downshift re-tries.
> >
> > If configured to DOWNSHIFT_DEV_DISABLE, then downshift is disabled
> > If configured to DOWNSHIFT_DEV_DEFAULT_COUNT, then it is up to the device to
> > choose a device-specific re-try count.
> >
> > Patches to implement this in ethtool will follow in a few minutes.
--
2.7.3
^ permalink raw reply
* [PATCH net] rtnetlink: fix rtnl message size computation for XDP
From: Sabrina Dubroca @ 2016-11-15 10:16 UTC (permalink / raw)
To: netdev; +Cc: Brenden Blanco, Sabrina Dubroca
rtnl_xdp_size() only considers the size of the actual payload attribute,
and misses the space taken by the attribute used for nesting (IFLA_XDP).
Fixes: d1fdd9138682 ("rtnl: add option for setting link xdp prog")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
---
net/core/rtnetlink.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index 5aeb61aadd32..5b4c3ba507d0 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -901,7 +901,8 @@ static size_t rtnl_port_size(const struct net_device *dev,
static size_t rtnl_xdp_size(const struct net_device *dev)
{
- size_t xdp_size = nla_total_size(1); /* XDP_ATTACHED */
+ size_t xdp_size = nla_total_size(0) + /* nest IFLA_XDP */
+ nla_total_size(1); /* XDP_ATTACHED */
if (!dev->netdev_ops->ndo_xdp)
return 0;
--
2.10.2
^ permalink raw reply related
* Re: [PATCH net-next v2] ipv6: sr: fix IPv6 initialization failure without lwtunnels
From: David Lebrun @ 2016-11-15 10:17 UTC (permalink / raw)
To: Roopa Prabhu; +Cc: David Miller, netdev, lorenzo
In-Reply-To: <5829C89B.7010405@cumulusnetworks.com>
[-- Attachment #1: Type: text/plain, Size: 351 bytes --]
On 11/14/2016 03:22 PM, Roopa Prabhu wrote:
> I prefer option b). most LWTUNNEL encaps are done this way.
>
> seg6 and seg6_iptunnel is new segment routing code and can be under
> CONFIG_IPV6_SEG6 which depends on CONFIG_LWTUNNEL and CONFIG_IPV6.
> CONFIG_IPV6_SEG6_HMAC could then depend on CONFIG_IPV6_SEG6
Will do that, thanks
David
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 163 bytes --]
^ permalink raw reply
* [PATCH 4.9.0-rc5] AR9300 calibration problems with antenna selected
From: Krzysztof Hałasa @ 2016-11-15 9:56 UTC (permalink / raw)
To: QCA ath9k Development, Kalle Valo, linux-wireless, ath9k-devel,
netdev, linux-kernel
Hi,
I recently tried to select a single antenna on AR9300 and it works for
30 seconds only. The subsequent calibration makes the RX signal level to
drop from the usual -30/-40 dBm to -70/-80 dBm, and the transmission
practically stops.
With the attached patch it works, though selecting the antenna doesn't
seem to have any visible effect, at least with "iw wlanX station dump"
(perhaps it works for TX).
I'm using ad-hoc mode:
rmmod ath9k
modprobe ath9k
iw dev wlan0 set type ibss
iw phy phyX set antenna 2
ip link set up dev wlan0
iw dev wlan0 set bitrates legacy-2.4 18
iw dev wlan0 ibss join nameXXX freqYYY
ip addr add ZZZ broadcast + dev wlan0
The card in question is Mikrotik (Routerboard) R11e-2HPnD mPCIe adapter:
AR9580 Wireless Network Adapter (rev 01), ID 168c:0033, subsystem
19b6:d016.
ieee80211 phy0: Atheros AR9300 Rev:4 mem=0xc0f40000, irq=334
https://routerboard.com/R11e-2HPnD
Linux 4.9.0-rc5.
Is there a better way?
Signed-off-by: Krzysztof Halasa <khalasa@piap.pl>
diff --git a/drivers/net/wireless/ath/ath9k/main.c b/drivers/net/wireless/ath/ath9k/main.c
index e9f32b5..7f17e5d 100644
--- a/drivers/net/wireless/ath/ath9k/main.c
+++ b/drivers/net/wireless/ath/ath9k/main.c
@@ -2245,7 +2245,7 @@ static int ath9k_set_antenna(struct ieee80211_hw *hw, u32 tx_ant, u32 rx_ant)
return 0;
/* AR9100 runs into calibration issues if not all rx chains are enabled */
- if (AR_SREV_9100(ah))
+ if (AR_SREV_9100(ah) || AR_SREV_9300(ah))
ah->rxchainmask = 0x7;
else
ah->rxchainmask = fill_chainmask(ah->caps.rx_chainmask, rx_ant);
--
Krzysztof Halasa
Industrial Research Institute for Automation and Measurements PIAP
Al. Jerozolimskie 202, 02-486 Warsaw, Poland
^ permalink raw reply related
* [PATCH net] rtnetlink: fix rtnl_vfinfo_size
From: Sabrina Dubroca @ 2016-11-15 9:39 UTC (permalink / raw)
To: netdev
Cc: Eran Ben Elisha, Hadar Hen Zion, Or Gerlitz, Sucheta Chakraborty,
Sabrina Dubroca
The size reported by rtnl_vfinfo_size doesn't match the space used by
rtnl_fill_vfinfo.
rtnl_vfinfo_size currently doesn't account for the nest attributes
used by statistics (added in commit 3b766cd83232), nor for struct
ifla_vf_tx_rate (since commit ed616689a3d9, which added ifla_vf_rate
to the dump without removing ifla_vf_tx_rate, but replaced
ifla_vf_tx_rate with ifla_vf_rate in the size computation).
Fixes: 3b766cd83232 ("net/core: Add reading VF statistics through the PF netdevice")
Fixes: ed616689a3d9 ("net-next:v4: Add support to configure SR-IOV VF minimum and maximum Tx rate through ip tool")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
---
net/core/rtnetlink.c | 12 +++++++-----
1 file changed, 7 insertions(+), 5 deletions(-)
diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index db313ec7af32..96f4bf274e30 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -840,18 +840,20 @@ static inline int rtnl_vfinfo_size(const struct net_device *dev,
if (dev->dev.parent && dev_is_pci(dev->dev.parent) &&
(ext_filter_mask & RTEXT_FILTER_VF)) {
int num_vfs = dev_num_vf(dev->dev.parent);
- size_t size = nla_total_size(sizeof(struct nlattr));
- size += nla_total_size(num_vfs * sizeof(struct nlattr));
+ size_t size = nla_total_size(0);
size += num_vfs *
- (nla_total_size(sizeof(struct ifla_vf_mac)) +
- nla_total_size(MAX_VLAN_LIST_LEN *
- sizeof(struct nlattr)) +
+ (nla_total_size(0) +
+ nla_total_size(sizeof(struct ifla_vf_mac)) +
+ nla_total_size(sizeof(struct ifla_vf_vlan)) +
+ nla_total_size(0) + /* nest IFLA_VF_VLAN_LIST */
nla_total_size(MAX_VLAN_LIST_LEN *
sizeof(struct ifla_vf_vlan_info)) +
nla_total_size(sizeof(struct ifla_vf_spoofchk)) +
+ nla_total_size(sizeof(struct ifla_vf_tx_rate)) +
nla_total_size(sizeof(struct ifla_vf_rate)) +
nla_total_size(sizeof(struct ifla_vf_link_state)) +
nla_total_size(sizeof(struct ifla_vf_rss_query_en)) +
+ nla_total_size(0) + /* nest IFLA_VF_STATS */
/* IFLA_VF_STATS_RX_PACKETS */
nla_total_size_64bit(sizeof(__u64)) +
/* IFLA_VF_STATS_TX_PACKETS */
--
2.10.2
^ permalink raw reply related
* Re: [patch net-next v2 0/8] Add support for offloading packet-sampling
From: Jiri Pirko @ 2016-11-15 9:27 UTC (permalink / raw)
To: netdev
Cc: davem, yotamg, idosch, eladr, nogahf, ogerlitz, jhs,
geert+renesas, stephen, xiyou.wangcong, linux, roopa,
john.fastabend, simon.horman
In-Reply-To: <1479135638-3580-1-git-send-email-jiri@resnulli.us>
Mon, Nov 14, 2016 at 04:00:30PM CET, jiri@resnulli.us wrote:
>From: Jiri Pirko <jiri@mellanox.com>
>
>Add the sample tc action, which allows to sample packet matching
>a classifier. The sample action peeks randomly packets, duplicates them,
>truncates them and adds informative metadata on the packet, for example,
>the input interface and the original packet length. The sampled packets
>are marked to allow matching them and redirecting them to a specific
>collector device.
>
>The sampled packets metadata is packed using ife encapsulation. To do
>that, this patch-set extracts ife logics from the tc_ife action into an
>independent ife module, and uses that functionality to pack the metadata.
>To include all the needed metadata, this patch-set introduces some new
>IFE_META tlv types.
>
>In addition, Add the support for offloading the macthall-sample tc command
>in the Mellanox mlxsw driver, for ingress qdiscs.
>
>Userspace examples for that code can be found in:
> - https://github.com/yotamgi/host-sflow: sflow client that uses the sample
> with combination of tap device to sample packets and send to a
> centralized sflow collector.
> - https://github.com/yotamgi/libife: a library for manipulating ife
> packets in userspace. The library is used in the host-sflow program to
> parse the sampled packets.
Dave, please drop this patchset for now. We just discovered couple of
more changes are needed.
Thanks
^ permalink raw reply
* Re: linux-next: BUG: unable to handle kernel NULL pointer dereference in __sk_mem_raise_allocated()
From: Paolo Abeni @ 2016-11-15 9:26 UTC (permalink / raw)
To: Andrei Vagin; +Cc: Linux Kernel Network Developers
In-Reply-To: <CANaxB-yCk8hhP68L4Q2nFOJht8sqgXGGQO2AftpHs0u1xyGG5A@mail.gmail.com>
Hi,
On Mon, 2016-11-14 at 15:24 -0800, Andrei Vagin wrote:
> Our test system detected a kernel oops. Looks like a problem in the
> "udp: refactor memory accounting" series.
My fault: I missed udplite in my tests.
Thank you for reporting.
I'm fine with Eric's patch, setting both .memory_allocated
and .sysctl_mem.
We could also remove .backlog_rcv, but it's not strictly needed.
Paolo
^ permalink raw reply
* [PATCH nf-next 1/2] ipvs: Use IS_ERR_OR_NULL(svc) instead of IS_ERR(svc) || svc == NULL
From: Simon Horman @ 2016-11-15 9:01 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Gao Feng, Simon Horman
In-Reply-To: <1479200503-25842-1-git-send-email-horms@verge.net.au>
From: Gao Feng <fgao@ikuai8.com>
This minor refactoring does not change the logic of function
ip_vs_genl_dump_dests.
Signed-off-by: Gao Feng <fgao@ikuai8.com>
Acked-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_ctl.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/net/netfilter/ipvs/ip_vs_ctl.c b/net/netfilter/ipvs/ip_vs_ctl.c
index 6b85ded4f91d..217e0105b5e0 100644
--- a/net/netfilter/ipvs/ip_vs_ctl.c
+++ b/net/netfilter/ipvs/ip_vs_ctl.c
@@ -3260,7 +3260,7 @@ static int ip_vs_genl_dump_dests(struct sk_buff *skb,
svc = ip_vs_genl_find_service(ipvs, attrs[IPVS_CMD_ATTR_SERVICE]);
- if (IS_ERR(svc) || svc == NULL)
+ if (IS_ERR_OR_NULL(svc))
goto out_err;
/* Dump the destinations */
--
2.7.0.rc3.207.g0ac5344
^ permalink raw reply related
* [PATCH nf-next 2/2] ipvs: Decrement ttl
From: Simon Horman @ 2016-11-15 9:01 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Dwip Banerjee, Simon Horman
In-Reply-To: <1479200503-25842-1-git-send-email-horms@verge.net.au>
From: Dwip Banerjee <dwip@linux.vnet.ibm.com>
We decrement the IP ttl in all the modes in order to prevent infinite
route loops. The changes were done based on Julian Anastasov's
suggestions in a prior thread.
The ttl based check/discard and the actual decrement are done in
__ip_vs_get_out_rt() and in __ip_vs_get_out_rt_v6(), for the IPv6
case. decrement_ttl() implements the actual functionality for the
two cases.
Signed-off-by: Dwip Banerjee <dwip@linux.vnet.ibm.com>
Acked-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
---
net/netfilter/ipvs/ip_vs_xmit.c | 54 +++++++++++++++++++++++++++++++++++++++++
1 file changed, 54 insertions(+)
diff --git a/net/netfilter/ipvs/ip_vs_xmit.c b/net/netfilter/ipvs/ip_vs_xmit.c
index 01d3d894de46..4e1a98fcc8c3 100644
--- a/net/netfilter/ipvs/ip_vs_xmit.c
+++ b/net/netfilter/ipvs/ip_vs_xmit.c
@@ -254,6 +254,54 @@ static inline bool ensure_mtu_is_adequate(struct netns_ipvs *ipvs, int skb_af,
return true;
}
+static inline bool decrement_ttl(struct netns_ipvs *ipvs,
+ int skb_af,
+ struct sk_buff *skb)
+{
+ struct net *net = ipvs->net;
+
+#ifdef CONFIG_IP_VS_IPV6
+ if (skb_af == AF_INET6) {
+ struct dst_entry *dst = skb_dst(skb);
+
+ /* check and decrement ttl */
+ if (ipv6_hdr(skb)->hop_limit <= 1) {
+ /* Force OUTPUT device used as source address */
+ skb->dev = dst->dev;
+ icmpv6_send(skb, ICMPV6_TIME_EXCEED,
+ ICMPV6_EXC_HOPLIMIT, 0);
+ __IP6_INC_STATS(net, ip6_dst_idev(dst),
+ IPSTATS_MIB_INHDRERRORS);
+
+ return false;
+ }
+
+ /* don't propagate ttl change to cloned packets */
+ if (!skb_make_writable(skb, sizeof(struct ipv6hdr)))
+ return false;
+
+ ipv6_hdr(skb)->hop_limit--;
+ } else
+#endif
+ {
+ if (ip_hdr(skb)->ttl <= 1) {
+ /* Tell the sender its packet died... */
+ __IP_INC_STATS(net, IPSTATS_MIB_INHDRERRORS);
+ icmp_send(skb, ICMP_TIME_EXCEEDED, ICMP_EXC_TTL, 0);
+ return false;
+ }
+
+ /* don't propagate ttl change to cloned packets */
+ if (!skb_make_writable(skb, sizeof(struct iphdr)))
+ return false;
+
+ /* Decrease ttl */
+ ip_decrease_ttl(ip_hdr(skb));
+ }
+
+ return true;
+}
+
/* Get route to destination or remote server */
static int
__ip_vs_get_out_rt(struct netns_ipvs *ipvs, int skb_af, struct sk_buff *skb,
@@ -326,6 +374,9 @@ __ip_vs_get_out_rt(struct netns_ipvs *ipvs, int skb_af, struct sk_buff *skb,
return local;
}
+ if (!decrement_ttl(ipvs, skb_af, skb))
+ goto err_put;
+
if (likely(!(rt_mode & IP_VS_RT_MODE_TUNNEL))) {
mtu = dst_mtu(&rt->dst);
} else {
@@ -473,6 +524,9 @@ __ip_vs_get_out_rt_v6(struct netns_ipvs *ipvs, int skb_af, struct sk_buff *skb,
return local;
}
+ if (!decrement_ttl(ipvs, skb_af, skb))
+ goto err_put;
+
/* MTU checking */
if (likely(!(rt_mode & IP_VS_RT_MODE_TUNNEL)))
mtu = dst_mtu(&rt->dst);
--
2.7.0.rc3.207.g0ac5344
^ permalink raw reply related
* [GIT PULL nf-next 0/2] IPVS Updates for v4.10
From: Simon Horman @ 2016-11-15 9:01 UTC (permalink / raw)
To: Pablo Neira Ayuso
Cc: lvs-devel, netdev, netfilter-devel, Wensong Zhang,
Julian Anastasov, Simon Horman
Hi Pablo,
please consider these enhancements to the IPVS for v4.10.
* Decrement the IP ttl in all the modes in order to prevent infinite
route loops. Thanks to Dwip Banerjee.
* Use IS_ERR_OR_NULL macro. Clean-up from Gao Feng.
The following changes since commit 7d384846b9987f7b611357adf3cdfecfdcf0c402:
Merge git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf-next (2016-11-13 22:41:25 -0500)
are available in the git repository at:
https://git.kernel.org/pub/scm/linux/kernel/git/horms/ipvs-next.git tags/ipvs-for-v4.10
for you to fetch changes up to 8d8e20e2d7bba8c50e64e0eca1cb83956f468e49:
ipvs: Decrement ttl (2016-11-15 09:49:20 +0100)
----------------------------------------------------------------
Dwip Banerjee (1):
ipvs: Decrement ttl
Gao Feng (1):
ipvs: Use IS_ERR_OR_NULL(svc) instead of IS_ERR(svc) || svc == NULL
net/netfilter/ipvs/ip_vs_ctl.c | 2 +-
net/netfilter/ipvs/ip_vs_xmit.c | 54 +++++++++++++++++++++++++++++++++++++++++
2 files changed, 55 insertions(+), 1 deletion(-)
^ permalink raw reply
* Re: linux-next: BUG: unable to handle kernel NULL pointer dereference in __sk_mem_raise_allocated()
From: Paolo Abeni @ 2016-11-15 8:57 UTC (permalink / raw)
To: Eric Dumazet; +Cc: Andrei Vagin, Linux Kernel Network Developers
In-Reply-To: <1479169886.8455.111.camel@edumazet-glaptop3.roam.corp.google.com>
On Mon, 2016-11-14 at 16:31 -0800, Eric Dumazet wrote:
> On Mon, 2016-11-14 at 15:58 -0800, Andrei Vagin wrote:
>
> >
> > Yes, you are right. It works if we set .memory_allocated and .sysctl_mem.
>
> Now the question would be :
>
> Are we okay if UDP and UDPlite share the same limits ?
>
> I would vote for yes, because these default limits are huge anyway
> (The 50% reduction done in
> https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=b66e91ccbc34ebd5a2f90f9e1bc1597e2924a500
> only impacted TCP )
Thank you for jumping on this so early!
I'm sorry for the udplite left over: my fault.
I agree with sharing the limits between UDP and UDPlite. I think that
the current ones are so high we can apply also a similar reduction to
UDP (and SCTP, too)
Cheers,
Paolo
^ permalink raw reply
* [PATCH net-next v8 7/9] arch/powerpc: Enable FSL_PAMU
From: Madalin Bucur @ 2016-11-15 8:41 UTC (permalink / raw)
To: netdev
Cc: linuxppc-dev, linux-kernel, davem, oss, ppc, joe, pebolle,
joakim.tjernlund
In-Reply-To: <1479199269-9748-1-git-send-email-madalin.bucur@nxp.com>
Signed-off-by: Madalin Bucur <madalin.bucur@nxp.com>
---
arch/powerpc/configs/dpaa.config | 1 +
1 file changed, 1 insertion(+)
diff --git a/arch/powerpc/configs/dpaa.config b/arch/powerpc/configs/dpaa.config
index efa99c0..f124ee1 100644
--- a/arch/powerpc/configs/dpaa.config
+++ b/arch/powerpc/configs/dpaa.config
@@ -1 +1,2 @@
CONFIG_FSL_DPAA=y
+CONFIG_FSL_PAMU=y
--
2.1.0
^ permalink raw reply related
* [PATCH net-next v8 5/9] dpaa_eth: add sysfs exports
From: Madalin Bucur @ 2016-11-15 8:41 UTC (permalink / raw)
To: netdev
Cc: linuxppc-dev, linux-kernel, davem, oss, ppc, joe, pebolle,
joakim.tjernlund
In-Reply-To: <1479199269-9748-1-git-send-email-madalin.bucur@nxp.com>
Export Frame Queue and Buffer Pool IDs through sysfs.
Signed-off-by: Madalin Bucur <madalin.bucur@nxp.com>
---
drivers/net/ethernet/freescale/dpaa/Makefile | 2 +-
drivers/net/ethernet/freescale/dpaa/dpaa_eth.c | 4 +
drivers/net/ethernet/freescale/dpaa/dpaa_eth.h | 4 +
.../net/ethernet/freescale/dpaa/dpaa_eth_sysfs.c | 165 +++++++++++++++++++++
4 files changed, 174 insertions(+), 1 deletion(-)
create mode 100644 drivers/net/ethernet/freescale/dpaa/dpaa_eth_sysfs.c
diff --git a/drivers/net/ethernet/freescale/dpaa/Makefile b/drivers/net/ethernet/freescale/dpaa/Makefile
index 43a4cfd..bfb03d4 100644
--- a/drivers/net/ethernet/freescale/dpaa/Makefile
+++ b/drivers/net/ethernet/freescale/dpaa/Makefile
@@ -8,4 +8,4 @@ ccflags-y += -I$(FMAN)
obj-$(CONFIG_FSL_DPAA_ETH) += fsl_dpa.o
-fsl_dpa-objs += dpaa_eth.o dpaa_ethtool.o
+fsl_dpa-objs += dpaa_eth.o dpaa_ethtool.o dpaa_eth_sysfs.o
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
index fcb9cac..07ec2fe 100644
--- a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
@@ -2619,6 +2619,8 @@ static int dpaa_eth_probe(struct platform_device *pdev)
if (err < 0)
goto netdev_init_failed;
+ dpaa_eth_sysfs_init(&net_dev->dev);
+
netif_info(priv, probe, net_dev, "Probed interface %s\n",
net_dev->name);
@@ -2664,6 +2666,8 @@ static int dpaa_remove(struct platform_device *pdev)
priv = netdev_priv(net_dev);
+ dpaa_eth_sysfs_remove(dev);
+
dev_set_drvdata(dev, NULL);
unregister_netdev(net_dev);
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
index 711fb06..44323e2 100644
--- a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
@@ -177,4 +177,8 @@ struct dpaa_priv {
/* from dpaa_ethtool.c */
extern const struct ethtool_ops dpaa_ethtool_ops;
+
+/* from dpaa_eth_sysfs.c */
+void dpaa_eth_sysfs_remove(struct device *dev);
+void dpaa_eth_sysfs_init(struct device *dev);
#endif /* __DPAA_H */
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth_sysfs.c b/drivers/net/ethernet/freescale/dpaa/dpaa_eth_sysfs.c
new file mode 100644
index 0000000..ec75d1c
--- /dev/null
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth_sysfs.c
@@ -0,0 +1,165 @@
+/* Copyright 2008-2016 Freescale Semiconductor Inc.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * * Neither the name of Freescale Semiconductor nor the
+ * names of its contributors may be used to endorse or promote products
+ * derived from this software without specific prior written permission.
+ *
+ *
+ * ALTERNATIVELY, this software may be distributed under the terms of the
+ * GNU General Public License ("GPL") as published by the Free Software
+ * Foundation, either version 2 of that License or (at your option) any
+ * later version.
+ *
+ * THIS SOFTWARE IS PROVIDED BY Freescale Semiconductor ``AS IS'' AND ANY
+ * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ * DISCLAIMED. IN NO EVENT SHALL Freescale Semiconductor BE LIABLE FOR ANY
+ * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+ * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/io.h>
+#include <linux/of_net.h>
+#include "dpaa_eth.h"
+#include "mac.h"
+
+static ssize_t dpaa_eth_show_addr(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct dpaa_priv *priv = netdev_priv(to_net_dev(dev));
+ struct mac_device *mac_dev = priv->mac_dev;
+
+ if (mac_dev)
+ return sprintf(buf, "%llx",
+ (unsigned long long)mac_dev->res->start);
+ else
+ return sprintf(buf, "none");
+}
+
+static ssize_t dpaa_eth_show_fqids(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct dpaa_priv *priv = netdev_priv(to_net_dev(dev));
+ struct dpaa_fq *prev = NULL;
+ char *prevstr = NULL;
+ struct dpaa_fq *tmp;
+ struct dpaa_fq *fq;
+ u32 first_fqid = 0;
+ u32 last_fqid = 0;
+ ssize_t bytes = 0;
+ char *str;
+ int i = 0;
+
+ list_for_each_entry_safe(fq, tmp, &priv->dpaa_fq_list, list) {
+ switch (fq->fq_type) {
+ case FQ_TYPE_RX_DEFAULT:
+ str = "Rx default";
+ break;
+ case FQ_TYPE_RX_ERROR:
+ str = "Rx error";
+ break;
+ case FQ_TYPE_TX_CONFIRM:
+ str = "Tx default confirmation";
+ break;
+ case FQ_TYPE_TX_CONF_MQ:
+ str = "Tx confirmation (mq)";
+ break;
+ case FQ_TYPE_TX_ERROR:
+ str = "Tx error";
+ break;
+ case FQ_TYPE_TX:
+ str = "Tx";
+ break;
+ default:
+ str = "Unknown";
+ }
+
+ if (prev && (abs(fq->fqid - prev->fqid) != 1 ||
+ str != prevstr)) {
+ if (last_fqid == first_fqid)
+ bytes += sprintf(buf + bytes,
+ "%s: %d\n", prevstr, prev->fqid);
+ else
+ bytes += sprintf(buf + bytes,
+ "%s: %d - %d\n", prevstr,
+ first_fqid, last_fqid);
+ }
+
+ if (prev && abs(fq->fqid - prev->fqid) == 1 &&
+ str == prevstr) {
+ last_fqid = fq->fqid;
+ } else {
+ first_fqid = fq->fqid;
+ last_fqid = fq->fqid;
+ }
+
+ prev = fq;
+ prevstr = str;
+ i++;
+ }
+
+ if (prev) {
+ if (last_fqid == first_fqid)
+ bytes += sprintf(buf + bytes, "%s: %d\n", prevstr,
+ prev->fqid);
+ else
+ bytes += sprintf(buf + bytes, "%s: %d - %d\n", prevstr,
+ first_fqid, last_fqid);
+ }
+
+ return bytes;
+}
+
+static ssize_t dpaa_eth_show_bpids(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct dpaa_priv *priv = netdev_priv(to_net_dev(dev));
+ ssize_t bytes = 0;
+ int i = 0;
+
+ for (i = 0; i < DPAA_BPS_NUM; i++)
+ bytes += snprintf(buf + bytes, PAGE_SIZE - bytes, "%u\n",
+ priv->dpaa_bps[i]->bpid);
+
+ return bytes;
+}
+
+static struct device_attribute dpaa_eth_attrs[] = {
+ __ATTR(device_addr, 0444, dpaa_eth_show_addr, NULL),
+ __ATTR(fqids, 0444, dpaa_eth_show_fqids, NULL),
+ __ATTR(bpids, 0444, dpaa_eth_show_bpids, NULL),
+};
+
+void dpaa_eth_sysfs_init(struct device *dev)
+{
+ int i;
+
+ for (i = 0; i < ARRAY_SIZE(dpaa_eth_attrs); i++)
+ if (device_create_file(dev, &dpaa_eth_attrs[i])) {
+ dev_err(dev, "Error creating sysfs file\n");
+ while (i > 0)
+ device_remove_file(dev, &dpaa_eth_attrs[--i]);
+ return;
+ }
+}
+
+void dpaa_eth_sysfs_remove(struct device *dev)
+{
+ int i;
+
+ for (i = 0; i < ARRAY_SIZE(dpaa_eth_attrs); i++)
+ device_remove_file(dev, &dpaa_eth_attrs[i]);
+}
--
2.1.0
^ permalink raw reply related
* [PATCH net-next v8 9/9] arch/powerpc: Enable dpaa_eth
From: Madalin Bucur @ 2016-11-15 8:41 UTC (permalink / raw)
To: netdev
Cc: linuxppc-dev, linux-kernel, davem, oss, ppc, joe, pebolle,
joakim.tjernlund
In-Reply-To: <1479199269-9748-1-git-send-email-madalin.bucur@nxp.com>
Signed-off-by: Madalin Bucur <madalin.bucur@nxp.com>
---
arch/powerpc/configs/dpaa.config | 1 +
1 file changed, 1 insertion(+)
diff --git a/arch/powerpc/configs/dpaa.config b/arch/powerpc/configs/dpaa.config
index 9ad9bc0..2fe76f5 100644
--- a/arch/powerpc/configs/dpaa.config
+++ b/arch/powerpc/configs/dpaa.config
@@ -1,3 +1,4 @@
CONFIG_FSL_DPAA=y
CONFIG_FSL_PAMU=y
CONFIG_FSL_FMAN=y
+CONFIG_FSL_DPAA_ETH=y
--
2.1.0
^ permalink raw reply related
* [PATCH net-next v8 8/9] arch/powerpc: Enable FSL_FMAN
From: Madalin Bucur @ 2016-11-15 8:41 UTC (permalink / raw)
To: netdev
Cc: linuxppc-dev, linux-kernel, davem, oss, ppc, joe, pebolle,
joakim.tjernlund
In-Reply-To: <1479199269-9748-1-git-send-email-madalin.bucur@nxp.com>
Signed-off-by: Madalin Bucur <madalin.bucur@nxp.com>
---
arch/powerpc/configs/dpaa.config | 1 +
1 file changed, 1 insertion(+)
diff --git a/arch/powerpc/configs/dpaa.config b/arch/powerpc/configs/dpaa.config
index f124ee1..9ad9bc0 100644
--- a/arch/powerpc/configs/dpaa.config
+++ b/arch/powerpc/configs/dpaa.config
@@ -1,2 +1,3 @@
CONFIG_FSL_DPAA=y
CONFIG_FSL_PAMU=y
+CONFIG_FSL_FMAN=y
--
2.1.0
^ permalink raw reply related
* [PATCH net-next v8 6/9] dpaa_eth: add trace points
From: Madalin Bucur @ 2016-11-15 8:41 UTC (permalink / raw)
To: netdev
Cc: linuxppc-dev, linux-kernel, davem, oss, ppc, joe, pebolle,
joakim.tjernlund
In-Reply-To: <1479199269-9748-1-git-send-email-madalin.bucur@nxp.com>
Add trace points on the hot processing path.
Signed-off-by: Ruxandra Ioana Radulescu <ruxandra.radulescu@nxp.com>
---
drivers/net/ethernet/freescale/dpaa/Makefile | 1 +
drivers/net/ethernet/freescale/dpaa/dpaa_eth.c | 15 +++
drivers/net/ethernet/freescale/dpaa/dpaa_eth.h | 1 +
.../net/ethernet/freescale/dpaa/dpaa_eth_trace.h | 141 +++++++++++++++++++++
4 files changed, 158 insertions(+)
create mode 100644 drivers/net/ethernet/freescale/dpaa/dpaa_eth_trace.h
diff --git a/drivers/net/ethernet/freescale/dpaa/Makefile b/drivers/net/ethernet/freescale/dpaa/Makefile
index bfb03d4..7db50bc 100644
--- a/drivers/net/ethernet/freescale/dpaa/Makefile
+++ b/drivers/net/ethernet/freescale/dpaa/Makefile
@@ -9,3 +9,4 @@ ccflags-y += -I$(FMAN)
obj-$(CONFIG_FSL_DPAA_ETH) += fsl_dpa.o
fsl_dpa-objs += dpaa_eth.o dpaa_ethtool.o dpaa_eth_sysfs.o
+CFLAGS_dpaa_eth.o := -I$(src)
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
index 07ec2fe..3c48a84 100644
--- a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
@@ -59,6 +59,12 @@
#include "mac.h"
#include "dpaa_eth.h"
+/* CREATE_TRACE_POINTS only needs to be defined once. Other dpaa files
+ * using trace events only need to #include <trace/events/sched.h>
+ */
+#define CREATE_TRACE_POINTS
+#include "dpaa_eth_trace.h"
+
static int debug = -1;
module_param(debug, int, 0444);
MODULE_PARM_DESC(debug, "Module/Driver verbosity level (0=none,...,16=all)");
@@ -1863,6 +1869,9 @@ static inline int dpaa_xmit(struct dpaa_priv *priv,
if (fd->bpid == FSL_DPAA_BPID_INV)
fd->cmd |= qman_fq_fqid(priv->conf_fqs[queue]);
+ /* Trace this Tx fd */
+ trace_dpaa_tx_fd(priv->net_dev, egress_fq, fd);
+
for (i = 0; i < DPAA_ENQUEUE_RETRIES; i++) {
err = qman_enqueue(egress_fq, fd);
if (err != -EBUSY)
@@ -2097,6 +2106,9 @@ static enum qman_cb_dqrr_result rx_default_dqrr(struct qman_portal *portal,
if (!dpaa_bp)
return qman_cb_dqrr_consume;
+ /* Trace the Rx fd */
+ trace_dpaa_rx_fd(net_dev, fq, &dq->fd);
+
percpu_priv = this_cpu_ptr(priv->percpu_priv);
percpu_stats = &percpu_priv->stats;
@@ -2194,6 +2206,9 @@ static enum qman_cb_dqrr_result conf_dflt_dqrr(struct qman_portal *portal,
net_dev = ((struct dpaa_fq *)fq)->net_dev;
priv = netdev_priv(net_dev);
+ /* Trace the fd */
+ trace_dpaa_tx_conf_fd(net_dev, fq, &dq->fd);
+
percpu_priv = this_cpu_ptr(priv->percpu_priv);
if (dpaa_eth_napi_schedule(percpu_priv, portal))
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
index 44323e2..1f9aebf 100644
--- a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
@@ -37,6 +37,7 @@
#include "fman.h"
#include "mac.h"
+#include "dpaa_eth_trace.h"
#define DPAA_ETH_TXQ_NUM NR_CPUS
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth_trace.h b/drivers/net/ethernet/freescale/dpaa/dpaa_eth_trace.h
new file mode 100644
index 0000000..409c1dc
--- /dev/null
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth_trace.h
@@ -0,0 +1,141 @@
+/* Copyright 2013-2015 Freescale Semiconductor Inc.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * * Neither the name of Freescale Semiconductor nor the
+ * names of its contributors may be used to endorse or promote products
+ * derived from this software without specific prior written permission.
+ *
+ *
+ * ALTERNATIVELY, this software may be distributed under the terms of the
+ * GNU General Public License ("GPL") as published by the Free Software
+ * Foundation, either version 2 of that License or (at your option) any
+ * later version.
+ *
+ * THIS SOFTWARE IS PROVIDED BY Freescale Semiconductor ``AS IS'' AND ANY
+ * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ * DISCLAIMED. IN NO EVENT SHALL Freescale Semiconductor BE LIABLE FOR ANY
+ * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+ * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM dpaa_eth
+
+#if !defined(_DPAA_ETH_TRACE_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _DPAA_ETH_TRACE_H
+
+#include <linux/skbuff.h>
+#include <linux/netdevice.h>
+#include "dpaa_eth.h"
+#include <linux/tracepoint.h>
+
+#define fd_format_name(format) { qm_fd_##format, #format }
+#define fd_format_list \
+ fd_format_name(contig), \
+ fd_format_name(sg)
+
+/* This is used to declare a class of events.
+ * individual events of this type will be defined below.
+ */
+
+/* Store details about a frame descriptor and the FQ on which it was
+ * transmitted/received.
+ */
+DECLARE_EVENT_CLASS(dpaa_eth_fd,
+ /* Trace function prototype */
+ TP_PROTO(struct net_device *netdev,
+ struct qman_fq *fq,
+ const struct qm_fd *fd),
+
+ /* Repeat argument list here */
+ TP_ARGS(netdev, fq, fd),
+
+ /* A structure containing the relevant information we want to record.
+ * Declare name and type for each normal element, name, type and size
+ * for arrays. Use __string for variable length strings.
+ */
+ TP_STRUCT__entry(
+ __field(u32, fqid)
+ __field(u64, fd_addr)
+ __field(u8, fd_format)
+ __field(u16, fd_offset)
+ __field(u32, fd_length)
+ __field(u32, fd_status)
+ __string(name, netdev->name)
+ ),
+
+ /* The function that assigns values to the above declared fields */
+ TP_fast_assign(
+ __entry->fqid = fq->fqid;
+ __entry->fd_addr = qm_fd_addr_get64(fd);
+ __entry->fd_format = qm_fd_get_format(fd);
+ __entry->fd_offset = qm_fd_get_offset(fd);
+ __entry->fd_length = qm_fd_get_length(fd);
+ __entry->fd_status = fd->status;
+ __assign_str(name, netdev->name);
+ ),
+
+ /* This is what gets printed when the trace event is triggered */
+ TP_printk("[%s] fqid=%d, fd: addr=0x%llx, format=%s, off=%u, len=%u, status=0x%08x",
+ __get_str(name), __entry->fqid, __entry->fd_addr,
+ __print_symbolic(__entry->fd_format, fd_format_list),
+ __entry->fd_offset, __entry->fd_length, __entry->fd_status)
+);
+
+/* Now declare events of the above type. Format is:
+ * DEFINE_EVENT(class, name, proto, args), with proto and args same as for class
+ */
+
+/* Tx (egress) fd */
+DEFINE_EVENT(dpaa_eth_fd, dpaa_tx_fd,
+
+ TP_PROTO(struct net_device *netdev,
+ struct qman_fq *fq,
+ const struct qm_fd *fd),
+
+ TP_ARGS(netdev, fq, fd)
+);
+
+/* Rx fd */
+DEFINE_EVENT(dpaa_eth_fd, dpaa_rx_fd,
+
+ TP_PROTO(struct net_device *netdev,
+ struct qman_fq *fq,
+ const struct qm_fd *fd),
+
+ TP_ARGS(netdev, fq, fd)
+);
+
+/* Tx confirmation fd */
+DEFINE_EVENT(dpaa_eth_fd, dpaa_tx_conf_fd,
+
+ TP_PROTO(struct net_device *netdev,
+ struct qman_fq *fq,
+ const struct qm_fd *fd),
+
+ TP_ARGS(netdev, fq, fd)
+);
+
+/* If only one event of a certain type needs to be declared, use TRACE_EVENT().
+ * The syntax is the same as for DECLARE_EVENT_CLASS().
+ */
+
+#endif /* _DPAA_ETH_TRACE_H */
+
+/* This must be outside ifdef _DPAA_ETH_TRACE_H */
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH .
+#undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_FILE dpaa_eth_trace
+#include <trace/define_trace.h>
--
2.1.0
^ permalink raw reply related
* [PATCH net-next v8 4/9] dpaa_eth: add ethtool statistics
From: Madalin Bucur @ 2016-11-15 8:41 UTC (permalink / raw)
To: netdev
Cc: linuxppc-dev, linux-kernel, davem, oss, ppc, joe, pebolle,
joakim.tjernlund
In-Reply-To: <1479199269-9748-1-git-send-email-madalin.bucur@nxp.com>
Add a series of counters to be exported through ethtool:
- add detailed counters for reception errors;
- add detailed counters for QMan enqueue reject events;
- count the number of fragmented skbs received from the stack;
- count all frames received on the Tx confirmation path;
- add congestion group statistics;
- count the number of interrupts for each CPU.
Signed-off-by: Ioana Ciornei <ioana.ciornei@nxp.com>
Signed-off-by: Madalin Bucur <madalin.bucur@nxp.com>
---
drivers/net/ethernet/freescale/dpaa/dpaa_eth.c | 54 +++++-
drivers/net/ethernet/freescale/dpaa/dpaa_eth.h | 33 ++++
drivers/net/ethernet/freescale/dpaa/dpaa_ethtool.c | 199 +++++++++++++++++++++
3 files changed, 284 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
index 0e7f1c7..fcb9cac 100644
--- a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
@@ -700,10 +700,15 @@ static void dpaa_eth_cgscn(struct qman_portal *qm, struct qman_cgr *cgr,
struct dpaa_priv *priv = (struct dpaa_priv *)container_of(cgr,
struct dpaa_priv, cgr_data.cgr);
- if (congested)
+ if (congested) {
+ priv->cgr_data.congestion_start_jiffies = jiffies;
netif_tx_stop_all_queues(priv->net_dev);
- else
+ priv->cgr_data.cgr_congested_count++;
+ } else {
+ priv->cgr_data.congested_jiffies +=
+ (jiffies - priv->cgr_data.congestion_start_jiffies);
netif_tx_wake_all_queues(priv->net_dev);
+ }
}
static int dpaa_eth_cgr_init(struct dpaa_priv *priv)
@@ -1217,6 +1222,37 @@ static void dpaa_fd_release(const struct net_device *net_dev,
dpaa_bman_release(dpaa_bp, &bmb, 1);
}
+static void count_ern(struct dpaa_percpu_priv *percpu_priv,
+ const union qm_mr_entry *msg)
+{
+ switch (msg->ern.rc & QM_MR_RC_MASK) {
+ case QM_MR_RC_CGR_TAILDROP:
+ percpu_priv->ern_cnt.cg_tdrop++;
+ break;
+ case QM_MR_RC_WRED:
+ percpu_priv->ern_cnt.wred++;
+ break;
+ case QM_MR_RC_ERROR:
+ percpu_priv->ern_cnt.err_cond++;
+ break;
+ case QM_MR_RC_ORPWINDOW_EARLY:
+ percpu_priv->ern_cnt.early_window++;
+ break;
+ case QM_MR_RC_ORPWINDOW_LATE:
+ percpu_priv->ern_cnt.late_window++;
+ break;
+ case QM_MR_RC_FQ_TAILDROP:
+ percpu_priv->ern_cnt.fq_tdrop++;
+ break;
+ case QM_MR_RC_ORPWINDOW_RETIRED:
+ percpu_priv->ern_cnt.fq_retired++;
+ break;
+ case QM_MR_RC_ORP_ZERO:
+ percpu_priv->ern_cnt.orp_zero++;
+ break;
+ }
+}
+
/* Turn on HW checksum computation for this outgoing frame.
* If the current protocol is not something we support in this regard
* (or if the stack has already computed the SW checksum), we do nothing.
@@ -1882,6 +1918,7 @@ static int dpaa_start_xmit(struct sk_buff *skb, struct net_device *net_dev)
likely(skb_shinfo(skb)->nr_frags < DPAA_SGT_MAX_ENTRIES)) {
/* Just create a S/G fd based on the skb */
err = skb_to_sg_fd(priv, skb, &fd);
+ percpu_priv->tx_frag_skbuffs++;
} else {
/* If the egress skb contains more fragments than we support
* we have no choice but to linearize it ourselves.
@@ -1918,6 +1955,15 @@ static void dpaa_rx_error(struct net_device *net_dev,
percpu_priv->stats.rx_errors++;
+ if (fd->status & FM_FD_ERR_DMA)
+ percpu_priv->rx_errors.dme++;
+ if (fd->status & FM_FD_ERR_PHYSICAL)
+ percpu_priv->rx_errors.fpe++;
+ if (fd->status & FM_FD_ERR_SIZE)
+ percpu_priv->rx_errors.fse++;
+ if (fd->status & FM_FD_ERR_PRS_HDR_ERR)
+ percpu_priv->rx_errors.phe++;
+
dpaa_fd_release(net_dev, fd);
}
@@ -1973,6 +2019,8 @@ static void dpaa_tx_conf(struct net_device *net_dev,
percpu_priv->stats.tx_errors++;
}
+ percpu_priv->tx_confirm++;
+
skb = dpaa_cleanup_tx_fd(priv, fd);
consume_skb(skb);
@@ -1987,6 +2035,7 @@ static inline int dpaa_eth_napi_schedule(struct dpaa_percpu_priv *percpu_priv,
percpu_priv->np.p = portal;
napi_schedule(&percpu_priv->np.napi);
+ percpu_priv->in_interrupt++;
return 1;
}
return 0;
@@ -2171,6 +2220,7 @@ static void egress_ern(struct qman_portal *portal,
percpu_priv->stats.tx_dropped++;
percpu_priv->stats.tx_fifo_errors++;
+ count_ern(percpu_priv, msg);
skb = dpaa_cleanup_tx_fd(priv, fd);
dev_kfree_skb_any(skb);
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
index d6ab335..711fb06 100644
--- a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
@@ -95,6 +95,25 @@ struct dpaa_bp {
atomic_t refs;
};
+struct dpaa_rx_errors {
+ u64 dme; /* DMA Error */
+ u64 fpe; /* Frame Physical Error */
+ u64 fse; /* Frame Size Error */
+ u64 phe; /* Header Error */
+};
+
+/* Counters for QMan ERN frames - one counter per rejection code */
+struct dpaa_ern_cnt {
+ u64 cg_tdrop; /* Congestion group taildrop */
+ u64 wred; /* WRED congestion */
+ u64 err_cond; /* Error condition */
+ u64 early_window; /* Order restoration, frame too early */
+ u64 late_window; /* Order restoration, frame too late */
+ u64 fq_tdrop; /* FQ taildrop */
+ u64 fq_retired; /* FQ is retired */
+ u64 orp_zero; /* ORP disabled */
+};
+
struct dpaa_napi_portal {
struct napi_struct napi;
struct qman_portal *p;
@@ -104,7 +123,13 @@ struct dpaa_napi_portal {
struct dpaa_percpu_priv {
struct net_device *net_dev;
struct dpaa_napi_portal np;
+ u64 in_interrupt;
+ u64 tx_confirm;
+ /* fragmented (non-linear) skbuffs received from the stack */
+ u64 tx_frag_skbuffs;
struct rtnl_link_stats64 stats;
+ struct dpaa_rx_errors rx_errors;
+ struct dpaa_ern_cnt ern_cnt;
};
struct dpaa_buffer_layout {
@@ -133,6 +158,14 @@ struct dpaa_priv {
* (and the same) congestion group.
*/
struct qman_cgr cgr;
+ /* If congested, when it began. Used for performance stats. */
+ u32 congestion_start_jiffies;
+ /* Number of jiffies the Tx port was congested. */
+ u32 congested_jiffies;
+ /* Counter for the number of times the CGR
+ * entered congestion state
+ */
+ u32 cgr_congested_count;
} cgr_data;
/* Use a per-port CGR for ingress traffic. */
bool use_ingress_cgr;
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_ethtool.c b/drivers/net/ethernet/freescale/dpaa/dpaa_ethtool.c
index 3580a62..27e7044 100644
--- a/drivers/net/ethernet/freescale/dpaa/dpaa_ethtool.c
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_ethtool.c
@@ -36,6 +36,42 @@
#include "dpaa_eth.h"
#include "mac.h"
+static const char dpaa_stats_percpu[][ETH_GSTRING_LEN] = {
+ "interrupts",
+ "rx packets",
+ "tx packets",
+ "tx confirm",
+ "tx S/G",
+ "tx error",
+ "rx error",
+};
+
+static char dpaa_stats_global[][ETH_GSTRING_LEN] = {
+ /* dpa rx errors */
+ "rx dma error",
+ "rx frame physical error",
+ "rx frame size error",
+ "rx header error",
+
+ /* demultiplexing errors */
+ "qman cg_tdrop",
+ "qman wred",
+ "qman error cond",
+ "qman early window",
+ "qman late window",
+ "qman fq tdrop",
+ "qman fq retired",
+ "qman orp disabled",
+
+ /* congestion related stats */
+ "congestion time (ms)",
+ "entered congestion",
+ "congested (0/1)"
+};
+
+#define DPAA_STATS_PERCPU_LEN ARRAY_SIZE(dpaa_stats_percpu)
+#define DPAA_STATS_GLOBAL_LEN ARRAY_SIZE(dpaa_stats_global)
+
static int dpaa_get_settings(struct net_device *net_dev,
struct ethtool_cmd *et_cmd)
{
@@ -205,6 +241,166 @@ static int dpaa_set_pauseparam(struct net_device *net_dev,
return err;
}
+static int dpaa_get_sset_count(struct net_device *net_dev, int type)
+{
+ unsigned int total_stats, num_stats;
+
+ num_stats = num_online_cpus() + 1;
+ total_stats = num_stats * (DPAA_STATS_PERCPU_LEN + DPAA_BPS_NUM) +
+ DPAA_STATS_GLOBAL_LEN;
+
+ switch (type) {
+ case ETH_SS_STATS:
+ return total_stats;
+ default:
+ return -EOPNOTSUPP;
+ }
+}
+
+static void copy_stats(struct dpaa_percpu_priv *percpu_priv, int num_cpus,
+ int crr_cpu, u64 *bp_count, u64 *data)
+{
+ int num_values = num_cpus + 1;
+ int crr = 0, j;
+
+ /* update current CPU's stats and also add them to the total values */
+ data[crr * num_values + crr_cpu] = percpu_priv->in_interrupt;
+ data[crr++ * num_values + num_cpus] += percpu_priv->in_interrupt;
+
+ data[crr * num_values + crr_cpu] = percpu_priv->stats.rx_packets;
+ data[crr++ * num_values + num_cpus] += percpu_priv->stats.rx_packets;
+
+ data[crr * num_values + crr_cpu] = percpu_priv->stats.tx_packets;
+ data[crr++ * num_values + num_cpus] += percpu_priv->stats.tx_packets;
+
+ data[crr * num_values + crr_cpu] = percpu_priv->tx_confirm;
+ data[crr++ * num_values + num_cpus] += percpu_priv->tx_confirm;
+
+ data[crr * num_values + crr_cpu] = percpu_priv->tx_frag_skbuffs;
+ data[crr++ * num_values + num_cpus] += percpu_priv->tx_frag_skbuffs;
+
+ data[crr * num_values + crr_cpu] = percpu_priv->stats.tx_errors;
+ data[crr++ * num_values + num_cpus] += percpu_priv->stats.tx_errors;
+
+ data[crr * num_values + crr_cpu] = percpu_priv->stats.rx_errors;
+ data[crr++ * num_values + num_cpus] += percpu_priv->stats.rx_errors;
+
+ for (j = 0; j < DPAA_BPS_NUM; j++) {
+ data[crr * num_values + crr_cpu] = bp_count[j];
+ data[crr++ * num_values + num_cpus] += bp_count[j];
+ }
+}
+
+static void dpaa_get_ethtool_stats(struct net_device *net_dev,
+ struct ethtool_stats *stats, u64 *data)
+{
+ u64 bp_count[DPAA_BPS_NUM], cg_time, cg_num;
+ struct dpaa_percpu_priv *percpu_priv;
+ struct dpaa_rx_errors rx_errors;
+ unsigned int num_cpus, offset;
+ struct dpaa_ern_cnt ern_cnt;
+ struct dpaa_bp *dpaa_bp;
+ struct dpaa_priv *priv;
+ int total_stats, i, j;
+ bool cg_status;
+
+ total_stats = dpaa_get_sset_count(net_dev, ETH_SS_STATS);
+ priv = netdev_priv(net_dev);
+ num_cpus = num_online_cpus();
+
+ memset(&bp_count, 0, sizeof(bp_count));
+ memset(&rx_errors, 0, sizeof(struct dpaa_rx_errors));
+ memset(&ern_cnt, 0, sizeof(struct dpaa_ern_cnt));
+ memset(data, 0, total_stats * sizeof(u64));
+
+ for_each_online_cpu(i) {
+ percpu_priv = per_cpu_ptr(priv->percpu_priv, i);
+ for (j = 0; j < DPAA_BPS_NUM; j++) {
+ dpaa_bp = priv->dpaa_bps[j];
+ if (!dpaa_bp->percpu_count)
+ continue;
+ bp_count[j] = *(per_cpu_ptr(dpaa_bp->percpu_count, i));
+ }
+ rx_errors.dme += percpu_priv->rx_errors.dme;
+ rx_errors.fpe += percpu_priv->rx_errors.fpe;
+ rx_errors.fse += percpu_priv->rx_errors.fse;
+ rx_errors.phe += percpu_priv->rx_errors.phe;
+
+ ern_cnt.cg_tdrop += percpu_priv->ern_cnt.cg_tdrop;
+ ern_cnt.wred += percpu_priv->ern_cnt.wred;
+ ern_cnt.err_cond += percpu_priv->ern_cnt.err_cond;
+ ern_cnt.early_window += percpu_priv->ern_cnt.early_window;
+ ern_cnt.late_window += percpu_priv->ern_cnt.late_window;
+ ern_cnt.fq_tdrop += percpu_priv->ern_cnt.fq_tdrop;
+ ern_cnt.fq_retired += percpu_priv->ern_cnt.fq_retired;
+ ern_cnt.orp_zero += percpu_priv->ern_cnt.orp_zero;
+
+ copy_stats(percpu_priv, num_cpus, i, bp_count, data);
+ }
+
+ offset = (num_cpus + 1) * (DPAA_STATS_PERCPU_LEN + DPAA_BPS_NUM);
+ memcpy(data + offset, &rx_errors, sizeof(struct dpaa_rx_errors));
+
+ offset += sizeof(struct dpaa_rx_errors) / sizeof(u64);
+ memcpy(data + offset, &ern_cnt, sizeof(struct dpaa_ern_cnt));
+
+ /* gather congestion related counters */
+ cg_num = 0;
+ cg_status = 0;
+ cg_time = jiffies_to_msecs(priv->cgr_data.congested_jiffies);
+ if (qman_query_cgr_congested(&priv->cgr_data.cgr, &cg_status) == 0) {
+ cg_num = priv->cgr_data.cgr_congested_count;
+
+ /* reset congestion stats (like QMan API does */
+ priv->cgr_data.congested_jiffies = 0;
+ priv->cgr_data.cgr_congested_count = 0;
+ }
+
+ offset += sizeof(struct dpaa_ern_cnt) / sizeof(u64);
+ data[offset++] = cg_time;
+ data[offset++] = cg_num;
+ data[offset++] = cg_status;
+}
+
+static void dpaa_get_strings(struct net_device *net_dev, u32 stringset,
+ u8 *data)
+{
+ unsigned int i, j, num_cpus, size;
+ char string_cpu[ETH_GSTRING_LEN];
+ u8 *strings;
+
+ memset(string_cpu, 0, sizeof(string_cpu));
+ strings = data;
+ num_cpus = num_online_cpus();
+ size = DPAA_STATS_GLOBAL_LEN * ETH_GSTRING_LEN;
+
+ for (i = 0; i < DPAA_STATS_PERCPU_LEN; i++) {
+ for (j = 0; j < num_cpus; j++) {
+ snprintf(string_cpu, ETH_GSTRING_LEN, "%s [CPU %d]",
+ dpaa_stats_percpu[i], j);
+ memcpy(strings, string_cpu, ETH_GSTRING_LEN);
+ strings += ETH_GSTRING_LEN;
+ }
+ snprintf(string_cpu, ETH_GSTRING_LEN, "%s [TOTAL]",
+ dpaa_stats_percpu[i]);
+ memcpy(strings, string_cpu, ETH_GSTRING_LEN);
+ strings += ETH_GSTRING_LEN;
+ }
+ for (i = 0; i < DPAA_BPS_NUM; i++) {
+ for (j = 0; j < num_cpus; j++) {
+ snprintf(string_cpu, ETH_GSTRING_LEN,
+ "bpool %c [CPU %d]", 'a' + i, j);
+ memcpy(strings, string_cpu, ETH_GSTRING_LEN);
+ strings += ETH_GSTRING_LEN;
+ }
+ snprintf(string_cpu, ETH_GSTRING_LEN, "bpool %c [TOTAL]",
+ 'a' + i);
+ memcpy(strings, string_cpu, ETH_GSTRING_LEN);
+ strings += ETH_GSTRING_LEN;
+ }
+ memcpy(strings, dpaa_stats_global, size);
+}
+
const struct ethtool_ops dpaa_ethtool_ops = {
.get_settings = dpaa_get_settings,
.set_settings = dpaa_set_settings,
@@ -215,4 +411,7 @@ const struct ethtool_ops dpaa_ethtool_ops = {
.get_pauseparam = dpaa_get_pauseparam,
.set_pauseparam = dpaa_set_pauseparam,
.get_link = ethtool_op_get_link,
+ .get_sset_count = dpaa_get_sset_count,
+ .get_ethtool_stats = dpaa_get_ethtool_stats,
+ .get_strings = dpaa_get_strings,
};
--
2.1.0
^ permalink raw reply related
* [PATCH net-next v8 3/9] dpaa_eth: add ethtool functionality
From: Madalin Bucur @ 2016-11-15 8:41 UTC (permalink / raw)
To: netdev
Cc: linuxppc-dev, linux-kernel, davem, oss, ppc, joe, pebolle,
joakim.tjernlund
In-Reply-To: <1479199269-9748-1-git-send-email-madalin.bucur@nxp.com>
Add support for basic ethtool operations.
Signed-off-by: Madalin Bucur <madalin.bucur@nxp.com>
---
drivers/net/ethernet/freescale/dpaa/Makefile | 2 +-
drivers/net/ethernet/freescale/dpaa/dpaa_eth.c | 2 +
drivers/net/ethernet/freescale/dpaa/dpaa_eth.h | 3 +
drivers/net/ethernet/freescale/dpaa/dpaa_ethtool.c | 218 +++++++++++++++++++++
4 files changed, 224 insertions(+), 1 deletion(-)
create mode 100644 drivers/net/ethernet/freescale/dpaa/dpaa_ethtool.c
diff --git a/drivers/net/ethernet/freescale/dpaa/Makefile b/drivers/net/ethernet/freescale/dpaa/Makefile
index fc76029..43a4cfd 100644
--- a/drivers/net/ethernet/freescale/dpaa/Makefile
+++ b/drivers/net/ethernet/freescale/dpaa/Makefile
@@ -8,4 +8,4 @@ ccflags-y += -I$(FMAN)
obj-$(CONFIG_FSL_DPAA_ETH) += fsl_dpa.o
-fsl_dpa-objs += dpaa_eth.o
+fsl_dpa-objs += dpaa_eth.o dpaa_ethtool.o
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
index 8b9b0720f..0e7f1c7 100644
--- a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
@@ -240,6 +240,8 @@ static int dpaa_netdev_init(struct net_device *net_dev,
memcpy(net_dev->perm_addr, mac_addr, net_dev->addr_len);
memcpy(net_dev->dev_addr, mac_addr, net_dev->addr_len);
+ net_dev->ethtool_ops = &dpaa_ethtool_ops;
+
net_dev->needed_headroom = priv->tx_headroom;
net_dev->watchdog_timeo = msecs_to_jiffies(tx_timeout);
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
index fe98e08..d6ab335 100644
--- a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
@@ -141,4 +141,7 @@ struct dpaa_priv {
struct dpaa_buffer_layout buf_layout[2];
u16 rx_headroom;
};
+
+/* from dpaa_ethtool.c */
+extern const struct ethtool_ops dpaa_ethtool_ops;
#endif /* __DPAA_H */
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_ethtool.c b/drivers/net/ethernet/freescale/dpaa/dpaa_ethtool.c
new file mode 100644
index 0000000..3580a62
--- /dev/null
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_ethtool.c
@@ -0,0 +1,218 @@
+/* Copyright 2008-2016 Freescale Semiconductor, Inc.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * * Neither the name of Freescale Semiconductor nor the
+ * names of its contributors may be used to endorse or promote products
+ * derived from this software without specific prior written permission.
+ *
+ *
+ * ALTERNATIVELY, this software may be distributed under the terms of the
+ * GNU General Public License ("GPL") as published by the Free Software
+ * Foundation, either version 2 of that License or (at your option) any
+ * later version.
+ *
+ * THIS SOFTWARE IS PROVIDED BY Freescale Semiconductor ``AS IS'' AND ANY
+ * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ * DISCLAIMED. IN NO EVENT SHALL Freescale Semiconductor BE LIABLE FOR ANY
+ * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+ * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/string.h>
+
+#include "dpaa_eth.h"
+#include "mac.h"
+
+static int dpaa_get_settings(struct net_device *net_dev,
+ struct ethtool_cmd *et_cmd)
+{
+ int err;
+
+ if (!net_dev->phydev) {
+ netdev_dbg(net_dev, "phy device not initialized\n");
+ return 0;
+ }
+
+ err = phy_ethtool_gset(net_dev->phydev, et_cmd);
+
+ return err;
+}
+
+static int dpaa_set_settings(struct net_device *net_dev,
+ struct ethtool_cmd *et_cmd)
+{
+ int err;
+
+ if (!net_dev->phydev) {
+ netdev_err(net_dev, "phy device not initialized\n");
+ return -ENODEV;
+ }
+
+ err = phy_ethtool_sset(net_dev->phydev, et_cmd);
+ if (err < 0)
+ netdev_err(net_dev, "phy_ethtool_sset() = %d\n", err);
+
+ return err;
+}
+
+static void dpaa_get_drvinfo(struct net_device *net_dev,
+ struct ethtool_drvinfo *drvinfo)
+{
+ int len;
+
+ strlcpy(drvinfo->driver, KBUILD_MODNAME,
+ sizeof(drvinfo->driver));
+ len = snprintf(drvinfo->version, sizeof(drvinfo->version),
+ "%X", 0);
+ len = snprintf(drvinfo->fw_version, sizeof(drvinfo->fw_version),
+ "%X", 0);
+
+ if (len >= sizeof(drvinfo->fw_version)) {
+ /* Truncated output */
+ netdev_notice(net_dev, "snprintf() = %d\n", len);
+ }
+ strlcpy(drvinfo->bus_info, dev_name(net_dev->dev.parent->parent),
+ sizeof(drvinfo->bus_info));
+}
+
+static u32 dpaa_get_msglevel(struct net_device *net_dev)
+{
+ return ((struct dpaa_priv *)netdev_priv(net_dev))->msg_enable;
+}
+
+static void dpaa_set_msglevel(struct net_device *net_dev,
+ u32 msg_enable)
+{
+ ((struct dpaa_priv *)netdev_priv(net_dev))->msg_enable = msg_enable;
+}
+
+static int dpaa_nway_reset(struct net_device *net_dev)
+{
+ int err;
+
+ if (!net_dev->phydev) {
+ netdev_err(net_dev, "phy device not initialized\n");
+ return -ENODEV;
+ }
+
+ err = 0;
+ if (net_dev->phydev->autoneg) {
+ err = phy_start_aneg(net_dev->phydev);
+ if (err < 0)
+ netdev_err(net_dev, "phy_start_aneg() = %d\n",
+ err);
+ }
+
+ return err;
+}
+
+static void dpaa_get_pauseparam(struct net_device *net_dev,
+ struct ethtool_pauseparam *epause)
+{
+ struct mac_device *mac_dev;
+ struct dpaa_priv *priv;
+
+ priv = netdev_priv(net_dev);
+ mac_dev = priv->mac_dev;
+
+ if (!net_dev->phydev) {
+ netdev_err(net_dev, "phy device not initialized\n");
+ return;
+ }
+
+ epause->autoneg = mac_dev->autoneg_pause;
+ epause->rx_pause = mac_dev->rx_pause_active;
+ epause->tx_pause = mac_dev->tx_pause_active;
+}
+
+static int dpaa_set_pauseparam(struct net_device *net_dev,
+ struct ethtool_pauseparam *epause)
+{
+ struct mac_device *mac_dev;
+ struct phy_device *phydev;
+ bool rx_pause, tx_pause;
+ struct dpaa_priv *priv;
+ u32 newadv, oldadv;
+ int err;
+
+ priv = netdev_priv(net_dev);
+ mac_dev = priv->mac_dev;
+
+ phydev = net_dev->phydev;
+ if (!phydev) {
+ netdev_err(net_dev, "phy device not initialized\n");
+ return -ENODEV;
+ }
+
+ if (!(phydev->supported & SUPPORTED_Pause) ||
+ (!(phydev->supported & SUPPORTED_Asym_Pause) &&
+ (epause->rx_pause != epause->tx_pause)))
+ return -EINVAL;
+
+ /* The MAC should know how to handle PAUSE frame autonegotiation before
+ * adjust_link is triggered by a forced renegotiation of sym/asym PAUSE
+ * settings.
+ */
+ mac_dev->autoneg_pause = !!epause->autoneg;
+ mac_dev->rx_pause_req = !!epause->rx_pause;
+ mac_dev->tx_pause_req = !!epause->tx_pause;
+
+ /* Determine the sym/asym advertised PAUSE capabilities from the desired
+ * rx/tx pause settings.
+ */
+ newadv = 0;
+ if (epause->rx_pause)
+ newadv = ADVERTISED_Pause | ADVERTISED_Asym_Pause;
+ if (epause->tx_pause)
+ newadv |= ADVERTISED_Asym_Pause;
+
+ oldadv = phydev->advertising &
+ (ADVERTISED_Pause | ADVERTISED_Asym_Pause);
+
+ /* If there are differences between the old and the new advertised
+ * values, restart PHY autonegotiation and advertise the new values.
+ */
+ if (oldadv != newadv) {
+ phydev->advertising &= ~(ADVERTISED_Pause
+ | ADVERTISED_Asym_Pause);
+ phydev->advertising |= newadv;
+ if (phydev->autoneg) {
+ err = phy_start_aneg(phydev);
+ if (err < 0)
+ netdev_err(net_dev, "phy_start_aneg() = %d\n",
+ err);
+ }
+ }
+
+ fman_get_pause_cfg(mac_dev, &rx_pause, &tx_pause);
+ err = fman_set_mac_active_pause(mac_dev, rx_pause, tx_pause);
+ if (err < 0)
+ netdev_err(net_dev, "set_mac_active_pause() = %d\n", err);
+
+ return err;
+}
+
+const struct ethtool_ops dpaa_ethtool_ops = {
+ .get_settings = dpaa_get_settings,
+ .set_settings = dpaa_set_settings,
+ .get_drvinfo = dpaa_get_drvinfo,
+ .get_msglevel = dpaa_get_msglevel,
+ .set_msglevel = dpaa_set_msglevel,
+ .nway_reset = dpaa_nway_reset,
+ .get_pauseparam = dpaa_get_pauseparam,
+ .set_pauseparam = dpaa_set_pauseparam,
+ .get_link = ethtool_op_get_link,
+};
--
2.1.0
^ permalink raw reply related
* [PATCH net-next v8 2/9] dpaa_eth: add support for DPAA Ethernet
From: Madalin Bucur @ 2016-11-15 8:41 UTC (permalink / raw)
To: netdev
Cc: linuxppc-dev, linux-kernel, davem, oss, ppc, joe, pebolle,
joakim.tjernlund
In-Reply-To: <1479199269-9748-1-git-send-email-madalin.bucur@nxp.com>
This introduces the Freescale Data Path Acceleration Architecture
(DPAA) Ethernet driver (dpaa_eth) that builds upon the DPAA QMan,
BMan, PAMU and FMan drivers to deliver Ethernet connectivity on
the Freescale DPAA QorIQ platforms.
Signed-off-by: Madalin Bucur <madalin.bucur@nxp.com>
---
drivers/net/ethernet/freescale/Kconfig | 2 +
drivers/net/ethernet/freescale/Makefile | 1 +
drivers/net/ethernet/freescale/dpaa/Kconfig | 10 +
drivers/net/ethernet/freescale/dpaa/Makefile | 11 +
drivers/net/ethernet/freescale/dpaa/dpaa_eth.c | 2682 ++++++++++++++++++++++++
drivers/net/ethernet/freescale/dpaa/dpaa_eth.h | 144 ++
6 files changed, 2850 insertions(+)
create mode 100644 drivers/net/ethernet/freescale/dpaa/Kconfig
create mode 100644 drivers/net/ethernet/freescale/dpaa/Makefile
create mode 100644 drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
create mode 100644 drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
diff --git a/drivers/net/ethernet/freescale/Kconfig b/drivers/net/ethernet/freescale/Kconfig
index d1ca45f..aa3f615 100644
--- a/drivers/net/ethernet/freescale/Kconfig
+++ b/drivers/net/ethernet/freescale/Kconfig
@@ -93,4 +93,6 @@ config GIANFAR
and MPC86xx family of chips, the eTSEC on LS1021A and the FEC
on the 8540.
+source "drivers/net/ethernet/freescale/dpaa/Kconfig"
+
endif # NET_VENDOR_FREESCALE
diff --git a/drivers/net/ethernet/freescale/Makefile b/drivers/net/ethernet/freescale/Makefile
index cbe21dc..4a13115 100644
--- a/drivers/net/ethernet/freescale/Makefile
+++ b/drivers/net/ethernet/freescale/Makefile
@@ -22,3 +22,4 @@ obj-$(CONFIG_UCC_GETH) += ucc_geth_driver.o
ucc_geth_driver-objs := ucc_geth.o ucc_geth_ethtool.o
obj-$(CONFIG_FSL_FMAN) += fman/
+obj-$(CONFIG_FSL_DPAA_ETH) += dpaa/
diff --git a/drivers/net/ethernet/freescale/dpaa/Kconfig b/drivers/net/ethernet/freescale/dpaa/Kconfig
new file mode 100644
index 0000000..f3a3454
--- /dev/null
+++ b/drivers/net/ethernet/freescale/dpaa/Kconfig
@@ -0,0 +1,10 @@
+menuconfig FSL_DPAA_ETH
+ tristate "DPAA Ethernet"
+ depends on FSL_SOC && FSL_DPAA && FSL_FMAN
+ select PHYLIB
+ select FSL_FMAN_MAC
+ ---help---
+ Data Path Acceleration Architecture Ethernet driver,
+ supporting the Freescale QorIQ chips.
+ Depends on Freescale Buffer Manager and Queue Manager
+ driver and Frame Manager Driver.
diff --git a/drivers/net/ethernet/freescale/dpaa/Makefile b/drivers/net/ethernet/freescale/dpaa/Makefile
new file mode 100644
index 0000000..fc76029
--- /dev/null
+++ b/drivers/net/ethernet/freescale/dpaa/Makefile
@@ -0,0 +1,11 @@
+#
+# Makefile for the Freescale DPAA Ethernet controllers
+#
+
+# Include FMan headers
+FMAN = $(srctree)/drivers/net/ethernet/freescale/fman
+ccflags-y += -I$(FMAN)
+
+obj-$(CONFIG_FSL_DPAA_ETH) += fsl_dpa.o
+
+fsl_dpa-objs += dpaa_eth.o
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
new file mode 100644
index 0000000..8b9b0720f
--- /dev/null
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
@@ -0,0 +1,2682 @@
+/* Copyright 2008 - 2016 Freescale Semiconductor Inc.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * * Neither the name of Freescale Semiconductor nor the
+ * names of its contributors may be used to endorse or promote products
+ * derived from this software without specific prior written permission.
+ *
+ * ALTERNATIVELY, this software may be distributed under the terms of the
+ * GNU General Public License ("GPL") as published by the Free Software
+ * Foundation, either version 2 of that License or (at your option) any
+ * later version.
+ *
+ * THIS SOFTWARE IS PROVIDED BY Freescale Semiconductor ``AS IS'' AND ANY
+ * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ * DISCLAIMED. IN NO EVENT SHALL Freescale Semiconductor BE LIABLE FOR ANY
+ * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+ * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/of_platform.h>
+#include <linux/of_mdio.h>
+#include <linux/of_net.h>
+#include <linux/io.h>
+#include <linux/if_arp.h>
+#include <linux/if_vlan.h>
+#include <linux/icmp.h>
+#include <linux/ip.h>
+#include <linux/ipv6.h>
+#include <linux/udp.h>
+#include <linux/tcp.h>
+#include <linux/net.h>
+#include <linux/skbuff.h>
+#include <linux/etherdevice.h>
+#include <linux/if_ether.h>
+#include <linux/highmem.h>
+#include <linux/percpu.h>
+#include <linux/dma-mapping.h>
+#include <linux/sort.h>
+#include <soc/fsl/bman.h>
+#include <soc/fsl/qman.h>
+
+#include "fman.h"
+#include "fman_port.h"
+#include "mac.h"
+#include "dpaa_eth.h"
+
+static int debug = -1;
+module_param(debug, int, 0444);
+MODULE_PARM_DESC(debug, "Module/Driver verbosity level (0=none,...,16=all)");
+
+static u16 tx_timeout = 1000;
+module_param(tx_timeout, ushort, 0444);
+MODULE_PARM_DESC(tx_timeout, "The Tx timeout in ms");
+
+#define FM_FD_STAT_RX_ERRORS \
+ (FM_FD_ERR_DMA | FM_FD_ERR_PHYSICAL | \
+ FM_FD_ERR_SIZE | FM_FD_ERR_CLS_DISCARD | \
+ FM_FD_ERR_EXTRACTION | FM_FD_ERR_NO_SCHEME | \
+ FM_FD_ERR_PRS_TIMEOUT | FM_FD_ERR_PRS_ILL_INSTRUCT | \
+ FM_FD_ERR_PRS_HDR_ERR)
+
+#define FM_FD_STAT_TX_ERRORS \
+ (FM_FD_ERR_UNSUPPORTED_FORMAT | \
+ FM_FD_ERR_LENGTH | FM_FD_ERR_DMA)
+
+#define DPAA_MSG_DEFAULT (NETIF_MSG_DRV | NETIF_MSG_PROBE | \
+ NETIF_MSG_LINK | NETIF_MSG_IFUP | \
+ NETIF_MSG_IFDOWN)
+
+#define DPAA_INGRESS_CS_THRESHOLD 0x10000000
+/* Ingress congestion threshold on FMan ports
+ * The size in bytes of the ingress tail-drop threshold on FMan ports.
+ * Traffic piling up above this value will be rejected by QMan and discarded
+ * by FMan.
+ */
+
+/* Size in bytes of the FQ taildrop threshold */
+#define DPAA_FQ_TD 0x200000
+
+#define DPAA_CS_THRESHOLD_1G 0x06000000
+/* Egress congestion threshold on 1G ports, range 0x1000 .. 0x10000000
+ * The size in bytes of the egress Congestion State notification threshold on
+ * 1G ports. The 1G dTSECs can quite easily be flooded by cores doing Tx in a
+ * tight loop (e.g. by sending UDP datagrams at "while(1) speed"),
+ * and the larger the frame size, the more acute the problem.
+ * So we have to find a balance between these factors:
+ * - avoiding the device staying congested for a prolonged time (risking
+ * the netdev watchdog to fire - see also the tx_timeout module param);
+ * - affecting performance of protocols such as TCP, which otherwise
+ * behave well under the congestion notification mechanism;
+ * - preventing the Tx cores from tightly-looping (as if the congestion
+ * threshold was too low to be effective);
+ * - running out of memory if the CS threshold is set too high.
+ */
+
+#define DPAA_CS_THRESHOLD_10G 0x10000000
+/* The size in bytes of the egress Congestion State notification threshold on
+ * 10G ports, range 0x1000 .. 0x10000000
+ */
+
+/* Largest value that the FQD's OAL field can hold */
+#define FSL_QMAN_MAX_OAL 127
+
+/* Default alignment for start of data in an Rx FD */
+#define DPAA_FD_DATA_ALIGNMENT 16
+
+/* Values for the L3R field of the FM Parse Results
+ */
+/* L3 Type field: First IP Present IPv4 */
+#define FM_L3_PARSE_RESULT_IPV4 0x8000
+/* L3 Type field: First IP Present IPv6 */
+#define FM_L3_PARSE_RESULT_IPV6 0x4000
+/* Values for the L4R field of the FM Parse Results */
+/* L4 Type field: UDP */
+#define FM_L4_PARSE_RESULT_UDP 0x40
+/* L4 Type field: TCP */
+#define FM_L4_PARSE_RESULT_TCP 0x20
+
+#define DPAA_SGT_MAX_ENTRIES 16 /* maximum number of entries in SG Table */
+#define DPAA_BUFF_RELEASE_MAX 8 /* maximum number of buffers released at once */
+
+#define FSL_DPAA_BPID_INV 0xff
+#define FSL_DPAA_ETH_MAX_BUF_COUNT 128
+#define FSL_DPAA_ETH_REFILL_THRESHOLD 80
+
+#define DPAA_TX_PRIV_DATA_SIZE 16
+#define DPAA_PARSE_RESULTS_SIZE sizeof(struct fman_prs_result)
+#define DPAA_TIME_STAMP_SIZE 8
+#define DPAA_HASH_RESULTS_SIZE 8
+#define DPAA_RX_PRIV_DATA_SIZE (u16)(DPAA_TX_PRIV_DATA_SIZE + \
+ dpaa_rx_extra_headroom)
+
+#define DPAA_ETH_RX_QUEUES 128
+
+#define DPAA_ENQUEUE_RETRIES 100000
+
+enum port_type {RX, TX};
+
+struct fm_port_fqs {
+ struct dpaa_fq *tx_defq;
+ struct dpaa_fq *tx_errq;
+ struct dpaa_fq *rx_defq;
+ struct dpaa_fq *rx_errq;
+};
+
+/* All the dpa bps in use at any moment */
+static struct dpaa_bp *dpaa_bp_array[BM_MAX_NUM_OF_POOLS];
+
+/* The raw buffer size must be cacheline aligned */
+#define DPAA_BP_RAW_SIZE 4096
+/* When using more than one buffer pool, the raw sizes are as follows:
+ * 1 bp: 4KB
+ * 2 bp: 2KB, 4KB
+ * 3 bp: 1KB, 2KB, 4KB
+ * 4 bp: 1KB, 2KB, 4KB, 8KB
+ */
+static inline size_t bpool_buffer_raw_size(u8 index, u8 cnt)
+{
+ size_t res = DPAA_BP_RAW_SIZE / 4;
+ u8 i;
+
+ for (i = (cnt < 3) ? cnt : 3; i < 3 + index; i++)
+ res *= 2;
+ return res;
+}
+
+/* FMan-DMA requires 16-byte alignment for Rx buffers, but SKB_DATA_ALIGN is
+ * even stronger (SMP_CACHE_BYTES-aligned), so we just get away with that,
+ * via SKB_WITH_OVERHEAD(). We can't rely on netdev_alloc_frag() giving us
+ * half-page-aligned buffers, so we reserve some more space for start-of-buffer
+ * alignment.
+ */
+#define dpaa_bp_size(raw_size) SKB_WITH_OVERHEAD((raw_size) - SMP_CACHE_BYTES)
+
+static int dpaa_max_frm;
+
+static int dpaa_rx_extra_headroom;
+
+#define dpaa_get_max_mtu() \
+ (dpaa_max_frm - (VLAN_ETH_HLEN + ETH_FCS_LEN))
+
+static int dpaa_netdev_init(struct net_device *net_dev,
+ const struct net_device_ops *dpaa_ops,
+ u16 tx_timeout)
+{
+ struct dpaa_priv *priv = netdev_priv(net_dev);
+ struct device *dev = net_dev->dev.parent;
+ struct dpaa_percpu_priv *percpu_priv;
+ const u8 *mac_addr;
+ int i, err;
+
+ /* Although we access another CPU's private data here
+ * we do it at initialization so it is safe
+ */
+ for_each_possible_cpu(i) {
+ percpu_priv = per_cpu_ptr(priv->percpu_priv, i);
+ percpu_priv->net_dev = net_dev;
+ }
+
+ net_dev->netdev_ops = dpaa_ops;
+ mac_addr = priv->mac_dev->addr;
+
+ net_dev->mem_start = priv->mac_dev->res->start;
+ net_dev->mem_end = priv->mac_dev->res->end;
+
+ net_dev->min_mtu = ETH_MIN_MTU;
+ net_dev->max_mtu = dpaa_get_max_mtu();
+
+ net_dev->hw_features |= (NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
+ NETIF_F_LLTX);
+
+ net_dev->hw_features |= NETIF_F_SG | NETIF_F_HIGHDMA;
+ /* The kernels enables GSO automatically, if we declare NETIF_F_SG.
+ * For conformity, we'll still declare GSO explicitly.
+ */
+ net_dev->features |= NETIF_F_GSO;
+
+ net_dev->priv_flags |= IFF_LIVE_ADDR_CHANGE;
+ /* we do not want shared skbs on TX */
+ net_dev->priv_flags &= ~IFF_TX_SKB_SHARING;
+
+ net_dev->features |= net_dev->hw_features;
+ net_dev->vlan_features = net_dev->features;
+
+ memcpy(net_dev->perm_addr, mac_addr, net_dev->addr_len);
+ memcpy(net_dev->dev_addr, mac_addr, net_dev->addr_len);
+
+ net_dev->needed_headroom = priv->tx_headroom;
+ net_dev->watchdog_timeo = msecs_to_jiffies(tx_timeout);
+
+ /* start without the RUNNING flag, phylib controls it later */
+ netif_carrier_off(net_dev);
+
+ err = register_netdev(net_dev);
+ if (err < 0) {
+ dev_err(dev, "register_netdev() = %d\n", err);
+ return err;
+ }
+
+ return 0;
+}
+
+static int dpaa_stop(struct net_device *net_dev)
+{
+ struct mac_device *mac_dev;
+ struct dpaa_priv *priv;
+ int i, err, error;
+
+ priv = netdev_priv(net_dev);
+ mac_dev = priv->mac_dev;
+
+ netif_tx_stop_all_queues(net_dev);
+ /* Allow the Fman (Tx) port to process in-flight frames before we
+ * try switching it off.
+ */
+ usleep_range(5000, 10000);
+
+ err = mac_dev->stop(mac_dev);
+ if (err < 0)
+ netif_err(priv, ifdown, net_dev, "mac_dev->stop() = %d\n",
+ err);
+
+ for (i = 0; i < ARRAY_SIZE(mac_dev->port); i++) {
+ error = fman_port_disable(mac_dev->port[i]);
+ if (error)
+ err = error;
+ }
+
+ if (net_dev->phydev)
+ phy_disconnect(net_dev->phydev);
+ net_dev->phydev = NULL;
+
+ return err;
+}
+
+static void dpaa_tx_timeout(struct net_device *net_dev)
+{
+ struct dpaa_percpu_priv *percpu_priv;
+ const struct dpaa_priv *priv;
+
+ priv = netdev_priv(net_dev);
+ percpu_priv = this_cpu_ptr(priv->percpu_priv);
+
+ netif_crit(priv, timer, net_dev, "Transmit timeout latency: %u ms\n",
+ jiffies_to_msecs(jiffies - dev_trans_start(net_dev)));
+
+ percpu_priv->stats.tx_errors++;
+}
+
+/* Calculates the statistics for the given device by adding the statistics
+ * collected by each CPU.
+ */
+static struct rtnl_link_stats64 *dpaa_get_stats64(struct net_device *net_dev,
+ struct rtnl_link_stats64 *s)
+{
+ int numstats = sizeof(struct rtnl_link_stats64) / sizeof(u64);
+ struct dpaa_priv *priv = netdev_priv(net_dev);
+ struct dpaa_percpu_priv *percpu_priv;
+ u64 *netstats = (u64 *)s;
+ u64 *cpustats;
+ int i, j;
+
+ for_each_possible_cpu(i) {
+ percpu_priv = per_cpu_ptr(priv->percpu_priv, i);
+
+ cpustats = (u64 *)&percpu_priv->stats;
+
+ /* add stats from all CPUs */
+ for (j = 0; j < numstats; j++)
+ netstats[j] += cpustats[j];
+ }
+
+ return s;
+}
+
+static struct mac_device *dpaa_mac_dev_get(struct platform_device *pdev)
+{
+ struct platform_device *of_dev;
+ struct dpaa_eth_data *eth_data;
+ struct device *dpaa_dev, *dev;
+ struct device_node *mac_node;
+ struct mac_device *mac_dev;
+
+ dpaa_dev = &pdev->dev;
+ eth_data = dpaa_dev->platform_data;
+ if (!eth_data)
+ return ERR_PTR(-ENODEV);
+
+ mac_node = eth_data->mac_node;
+
+ of_dev = of_find_device_by_node(mac_node);
+ if (!of_dev) {
+ dev_err(dpaa_dev, "of_find_device_by_node(%s) failed\n",
+ mac_node->full_name);
+ of_node_put(mac_node);
+ return ERR_PTR(-EINVAL);
+ }
+ of_node_put(mac_node);
+
+ dev = &of_dev->dev;
+
+ mac_dev = dev_get_drvdata(dev);
+ if (!mac_dev) {
+ dev_err(dpaa_dev, "dev_get_drvdata(%s) failed\n",
+ dev_name(dev));
+ return ERR_PTR(-EINVAL);
+ }
+
+ return mac_dev;
+}
+
+static int dpaa_set_mac_address(struct net_device *net_dev, void *addr)
+{
+ const struct dpaa_priv *priv;
+ struct mac_device *mac_dev;
+ struct sockaddr old_addr;
+ int err;
+
+ priv = netdev_priv(net_dev);
+
+ memcpy(old_addr.sa_data, net_dev->dev_addr, ETH_ALEN);
+
+ err = eth_mac_addr(net_dev, addr);
+ if (err < 0) {
+ netif_err(priv, drv, net_dev, "eth_mac_addr() = %d\n", err);
+ return err;
+ }
+
+ mac_dev = priv->mac_dev;
+
+ err = mac_dev->change_addr(mac_dev->fman_mac,
+ (enet_addr_t *)net_dev->dev_addr);
+ if (err < 0) {
+ netif_err(priv, drv, net_dev, "mac_dev->change_addr() = %d\n",
+ err);
+ /* reverting to previous address */
+ eth_mac_addr(net_dev, &old_addr);
+
+ return err;
+ }
+
+ return 0;
+}
+
+static void dpaa_set_rx_mode(struct net_device *net_dev)
+{
+ const struct dpaa_priv *priv;
+ int err;
+
+ priv = netdev_priv(net_dev);
+
+ if (!!(net_dev->flags & IFF_PROMISC) != priv->mac_dev->promisc) {
+ priv->mac_dev->promisc = !priv->mac_dev->promisc;
+ err = priv->mac_dev->set_promisc(priv->mac_dev->fman_mac,
+ priv->mac_dev->promisc);
+ if (err < 0)
+ netif_err(priv, drv, net_dev,
+ "mac_dev->set_promisc() = %d\n",
+ err);
+ }
+
+ err = priv->mac_dev->set_multi(net_dev, priv->mac_dev);
+ if (err < 0)
+ netif_err(priv, drv, net_dev, "mac_dev->set_multi() = %d\n",
+ err);
+}
+
+static struct dpaa_bp *dpaa_bpid2pool(int bpid)
+{
+ if (WARN_ON(bpid < 0 || bpid >= BM_MAX_NUM_OF_POOLS))
+ return NULL;
+
+ return dpaa_bp_array[bpid];
+}
+
+/* checks if this bpool is already allocated */
+static bool dpaa_bpid2pool_use(int bpid)
+{
+ if (dpaa_bpid2pool(bpid)) {
+ atomic_inc(&dpaa_bp_array[bpid]->refs);
+ return true;
+ }
+
+ return false;
+}
+
+/* called only once per bpid by dpaa_bp_alloc_pool() */
+static void dpaa_bpid2pool_map(int bpid, struct dpaa_bp *dpaa_bp)
+{
+ dpaa_bp_array[bpid] = dpaa_bp;
+ atomic_set(&dpaa_bp->refs, 1);
+}
+
+static int dpaa_bp_alloc_pool(struct dpaa_bp *dpaa_bp)
+{
+ int err;
+
+ if (dpaa_bp->size == 0 || dpaa_bp->config_count == 0) {
+ pr_err("%s: Buffer pool is not properly initialized! Missing size or initial number of buffers\n",
+ __func__);
+ return -EINVAL;
+ }
+
+ /* If the pool is already specified, we only create one per bpid */
+ if (dpaa_bp->bpid != FSL_DPAA_BPID_INV &&
+ dpaa_bpid2pool_use(dpaa_bp->bpid))
+ return 0;
+
+ if (dpaa_bp->bpid == FSL_DPAA_BPID_INV) {
+ dpaa_bp->pool = bman_new_pool();
+ if (!dpaa_bp->pool) {
+ pr_err("%s: bman_new_pool() failed\n",
+ __func__);
+ return -ENODEV;
+ }
+
+ dpaa_bp->bpid = (u8)bman_get_bpid(dpaa_bp->pool);
+ }
+
+ if (dpaa_bp->seed_cb) {
+ err = dpaa_bp->seed_cb(dpaa_bp);
+ if (err)
+ goto pool_seed_failed;
+ }
+
+ dpaa_bpid2pool_map(dpaa_bp->bpid, dpaa_bp);
+
+ return 0;
+
+pool_seed_failed:
+ pr_err("%s: pool seeding failed\n", __func__);
+ bman_free_pool(dpaa_bp->pool);
+
+ return err;
+}
+
+/* remove and free all the buffers from the given buffer pool */
+static void dpaa_bp_drain(struct dpaa_bp *bp)
+{
+ u8 num = 8;
+ int ret;
+
+ do {
+ struct bm_buffer bmb[8];
+ int i;
+
+ ret = bman_acquire(bp->pool, bmb, num);
+ if (ret < 0) {
+ if (num == 8) {
+ /* we have less than 8 buffers left;
+ * drain them one by one
+ */
+ num = 1;
+ ret = 1;
+ continue;
+ } else {
+ /* Pool is fully drained */
+ break;
+ }
+ }
+
+ if (bp->free_buf_cb)
+ for (i = 0; i < num; i++)
+ bp->free_buf_cb(bp, &bmb[i]);
+ } while (ret > 0);
+}
+
+static void dpaa_bp_free(struct dpaa_bp *dpaa_bp)
+{
+ struct dpaa_bp *bp = dpaa_bpid2pool(dpaa_bp->bpid);
+
+ /* the mapping between bpid and dpaa_bp is done very late in the
+ * allocation procedure; if something failed before the mapping, the bp
+ * was not configured, therefore we don't need the below instructions
+ */
+ if (!bp)
+ return;
+
+ if (!atomic_dec_and_test(&bp->refs))
+ return;
+
+ if (bp->free_buf_cb)
+ dpaa_bp_drain(bp);
+
+ dpaa_bp_array[bp->bpid] = NULL;
+ bman_free_pool(bp->pool);
+}
+
+static void dpaa_bps_free(struct dpaa_priv *priv)
+{
+ int i;
+
+ for (i = 0; i < DPAA_BPS_NUM; i++)
+ dpaa_bp_free(priv->dpaa_bps[i]);
+}
+
+/* Use multiple WQs for FQ assignment:
+ * - Tx Confirmation queues go to WQ1.
+ * - Rx Error and Tx Error queues go to WQ2 (giving them a better chance
+ * to be scheduled, in case there are many more FQs in WQ3).
+ * - Rx Default and Tx queues go to WQ3 (no differentiation between
+ * Rx and Tx traffic).
+ * This ensures that Tx-confirmed buffers are timely released. In particular,
+ * it avoids congestion on the Tx Confirm FQs, which can pile up PFDRs if they
+ * are greatly outnumbered by other FQs in the system, while
+ * dequeue scheduling is round-robin.
+ */
+static inline void dpaa_assign_wq(struct dpaa_fq *fq)
+{
+ switch (fq->fq_type) {
+ case FQ_TYPE_TX_CONFIRM:
+ case FQ_TYPE_TX_CONF_MQ:
+ fq->wq = 1;
+ break;
+ case FQ_TYPE_RX_ERROR:
+ case FQ_TYPE_TX_ERROR:
+ fq->wq = 2;
+ break;
+ case FQ_TYPE_RX_DEFAULT:
+ case FQ_TYPE_TX:
+ fq->wq = 3;
+ break;
+ default:
+ WARN(1, "Invalid FQ type %d for FQID %d!\n",
+ fq->fq_type, fq->fqid);
+ }
+}
+
+static struct dpaa_fq *dpaa_fq_alloc(struct device *dev,
+ u32 start, u32 count,
+ struct list_head *list,
+ enum dpaa_fq_type fq_type)
+{
+ struct dpaa_fq *dpaa_fq;
+ int i;
+
+ dpaa_fq = devm_kzalloc(dev, sizeof(*dpaa_fq) * count,
+ GFP_KERNEL);
+ if (!dpaa_fq)
+ return NULL;
+
+ for (i = 0; i < count; i++) {
+ dpaa_fq[i].fq_type = fq_type;
+ dpaa_fq[i].fqid = start ? start + i : 0;
+ list_add_tail(&dpaa_fq[i].list, list);
+ }
+
+ for (i = 0; i < count; i++)
+ dpaa_assign_wq(dpaa_fq + i);
+
+ return dpaa_fq;
+}
+
+static int dpaa_alloc_all_fqs(struct device *dev, struct list_head *list,
+ struct fm_port_fqs *port_fqs)
+{
+ struct dpaa_fq *dpaa_fq;
+
+ dpaa_fq = dpaa_fq_alloc(dev, 0, 1, list, FQ_TYPE_RX_ERROR);
+ if (!dpaa_fq)
+ goto fq_alloc_failed;
+
+ port_fqs->rx_errq = &dpaa_fq[0];
+
+ dpaa_fq = dpaa_fq_alloc(dev, 0, 1, list, FQ_TYPE_RX_DEFAULT);
+ if (!dpaa_fq)
+ goto fq_alloc_failed;
+
+ port_fqs->rx_defq = &dpaa_fq[0];
+
+ if (!dpaa_fq_alloc(dev, 0, DPAA_ETH_TXQ_NUM, list, FQ_TYPE_TX_CONF_MQ))
+ goto fq_alloc_failed;
+
+ dpaa_fq = dpaa_fq_alloc(dev, 0, 1, list, FQ_TYPE_TX_ERROR);
+ if (!dpaa_fq)
+ goto fq_alloc_failed;
+
+ port_fqs->tx_errq = &dpaa_fq[0];
+
+ dpaa_fq = dpaa_fq_alloc(dev, 0, 1, list, FQ_TYPE_TX_CONFIRM);
+ if (!dpaa_fq)
+ goto fq_alloc_failed;
+
+ port_fqs->tx_defq = &dpaa_fq[0];
+
+ if (!dpaa_fq_alloc(dev, 0, DPAA_ETH_TXQ_NUM, list, FQ_TYPE_TX))
+ goto fq_alloc_failed;
+
+ return 0;
+
+fq_alloc_failed:
+ dev_err(dev, "dpaa_fq_alloc() failed\n");
+ return -ENOMEM;
+}
+
+static u32 rx_pool_channel;
+static DEFINE_SPINLOCK(rx_pool_channel_init);
+
+static int dpaa_get_channel(void)
+{
+ spin_lock(&rx_pool_channel_init);
+ if (!rx_pool_channel) {
+ u32 pool;
+ int ret;
+
+ ret = qman_alloc_pool(&pool);
+
+ if (!ret)
+ rx_pool_channel = pool;
+ }
+ spin_unlock(&rx_pool_channel_init);
+ if (!rx_pool_channel)
+ return -ENOMEM;
+ return rx_pool_channel;
+}
+
+static void dpaa_release_channel(void)
+{
+ qman_release_pool(rx_pool_channel);
+}
+
+static void dpaa_eth_add_channel(u16 channel)
+{
+ u32 pool = QM_SDQCR_CHANNELS_POOL_CONV(channel);
+ const cpumask_t *cpus = qman_affine_cpus();
+ struct qman_portal *portal;
+ int cpu;
+
+ for_each_cpu(cpu, cpus) {
+ portal = qman_get_affine_portal(cpu);
+ qman_p_static_dequeue_add(portal, pool);
+ }
+}
+
+/* Congestion group state change notification callback.
+ * Stops the device's egress queues while they are congested and
+ * wakes them upon exiting congested state.
+ * Also updates some CGR-related stats.
+ */
+static void dpaa_eth_cgscn(struct qman_portal *qm, struct qman_cgr *cgr,
+ int congested)
+{
+ struct dpaa_priv *priv = (struct dpaa_priv *)container_of(cgr,
+ struct dpaa_priv, cgr_data.cgr);
+
+ if (congested)
+ netif_tx_stop_all_queues(priv->net_dev);
+ else
+ netif_tx_wake_all_queues(priv->net_dev);
+}
+
+static int dpaa_eth_cgr_init(struct dpaa_priv *priv)
+{
+ struct qm_mcc_initcgr initcgr;
+ u32 cs_th;
+ int err;
+
+ err = qman_alloc_cgrid(&priv->cgr_data.cgr.cgrid);
+ if (err < 0) {
+ if (netif_msg_drv(priv))
+ pr_err("%s: Error %d allocating CGR ID\n",
+ __func__, err);
+ goto out_error;
+ }
+ priv->cgr_data.cgr.cb = dpaa_eth_cgscn;
+
+ /* Enable Congestion State Change Notifications and CS taildrop */
+ initcgr.we_mask = QM_CGR_WE_CSCN_EN | QM_CGR_WE_CS_THRES;
+ initcgr.cgr.cscn_en = QM_CGR_EN;
+
+ /* Set different thresholds based on the MAC speed.
+ * This may turn suboptimal if the MAC is reconfigured at a speed
+ * lower than its max, e.g. if a dTSEC later negotiates a 100Mbps link.
+ * In such cases, we ought to reconfigure the threshold, too.
+ */
+ if (priv->mac_dev->if_support & SUPPORTED_10000baseT_Full)
+ cs_th = DPAA_CS_THRESHOLD_10G;
+ else
+ cs_th = DPAA_CS_THRESHOLD_1G;
+ qm_cgr_cs_thres_set64(&initcgr.cgr.cs_thres, cs_th, 1);
+
+ initcgr.we_mask |= QM_CGR_WE_CSTD_EN;
+ initcgr.cgr.cstd_en = QM_CGR_EN;
+
+ err = qman_create_cgr(&priv->cgr_data.cgr, QMAN_CGR_FLAG_USE_INIT,
+ &initcgr);
+ if (err < 0) {
+ if (netif_msg_drv(priv))
+ pr_err("%s: Error %d creating CGR with ID %d\n",
+ __func__, err, priv->cgr_data.cgr.cgrid);
+ qman_release_cgrid(priv->cgr_data.cgr.cgrid);
+ goto out_error;
+ }
+ if (netif_msg_drv(priv))
+ pr_debug("Created CGR %d for netdev with hwaddr %pM on QMan channel %d\n",
+ priv->cgr_data.cgr.cgrid, priv->mac_dev->addr,
+ priv->cgr_data.cgr.chan);
+
+out_error:
+ return err;
+}
+
+static inline void dpaa_setup_ingress(const struct dpaa_priv *priv,
+ struct dpaa_fq *fq,
+ const struct qman_fq *template)
+{
+ fq->fq_base = *template;
+ fq->net_dev = priv->net_dev;
+
+ fq->flags = QMAN_FQ_FLAG_NO_ENQUEUE;
+ fq->channel = priv->channel;
+}
+
+static inline void dpaa_setup_egress(const struct dpaa_priv *priv,
+ struct dpaa_fq *fq,
+ struct fman_port *port,
+ const struct qman_fq *template)
+{
+ fq->fq_base = *template;
+ fq->net_dev = priv->net_dev;
+
+ if (port) {
+ fq->flags = QMAN_FQ_FLAG_TO_DCPORTAL;
+ fq->channel = (u16)fman_port_get_qman_channel_id(port);
+ } else {
+ fq->flags = QMAN_FQ_FLAG_NO_MODIFY;
+ }
+}
+
+static void dpaa_fq_setup(struct dpaa_priv *priv,
+ const struct dpaa_fq_cbs *fq_cbs,
+ struct fman_port *tx_port)
+{
+ int egress_cnt = 0, conf_cnt = 0, num_portals = 0, cpu;
+ const cpumask_t *affine_cpus = qman_affine_cpus();
+ u16 portals[NR_CPUS];
+ struct dpaa_fq *fq;
+
+ for_each_cpu(cpu, affine_cpus)
+ portals[num_portals++] = qman_affine_channel(cpu);
+ if (num_portals == 0)
+ dev_err(priv->net_dev->dev.parent,
+ "No Qman software (affine) channels found");
+
+ /* Initialize each FQ in the list */
+ list_for_each_entry(fq, &priv->dpaa_fq_list, list) {
+ switch (fq->fq_type) {
+ case FQ_TYPE_RX_DEFAULT:
+ dpaa_setup_ingress(priv, fq, &fq_cbs->rx_defq);
+ break;
+ case FQ_TYPE_RX_ERROR:
+ dpaa_setup_ingress(priv, fq, &fq_cbs->rx_errq);
+ break;
+ case FQ_TYPE_TX:
+ dpaa_setup_egress(priv, fq, tx_port,
+ &fq_cbs->egress_ern);
+ /* If we have more Tx queues than the number of cores,
+ * just ignore the extra ones.
+ */
+ if (egress_cnt < DPAA_ETH_TXQ_NUM)
+ priv->egress_fqs[egress_cnt++] = &fq->fq_base;
+ break;
+ case FQ_TYPE_TX_CONF_MQ:
+ priv->conf_fqs[conf_cnt++] = &fq->fq_base;
+ /* fall through */
+ case FQ_TYPE_TX_CONFIRM:
+ dpaa_setup_ingress(priv, fq, &fq_cbs->tx_defq);
+ break;
+ case FQ_TYPE_TX_ERROR:
+ dpaa_setup_ingress(priv, fq, &fq_cbs->tx_errq);
+ break;
+ default:
+ dev_warn(priv->net_dev->dev.parent,
+ "Unknown FQ type detected!\n");
+ break;
+ }
+ }
+
+ /* Make sure all CPUs receive a corresponding Tx queue. */
+ while (egress_cnt < DPAA_ETH_TXQ_NUM) {
+ list_for_each_entry(fq, &priv->dpaa_fq_list, list) {
+ if (fq->fq_type != FQ_TYPE_TX)
+ continue;
+ priv->egress_fqs[egress_cnt++] = &fq->fq_base;
+ if (egress_cnt == DPAA_ETH_TXQ_NUM)
+ break;
+ }
+ }
+}
+
+static inline int dpaa_tx_fq_to_id(const struct dpaa_priv *priv,
+ struct qman_fq *tx_fq)
+{
+ int i;
+
+ for (i = 0; i < DPAA_ETH_TXQ_NUM; i++)
+ if (priv->egress_fqs[i] == tx_fq)
+ return i;
+
+ return -EINVAL;
+}
+
+static int dpaa_fq_init(struct dpaa_fq *dpaa_fq, bool td_enable)
+{
+ const struct dpaa_priv *priv;
+ struct qman_fq *confq = NULL;
+ struct qm_mcc_initfq initfq;
+ struct device *dev;
+ struct qman_fq *fq;
+ int queue_id;
+ int err;
+
+ priv = netdev_priv(dpaa_fq->net_dev);
+ dev = dpaa_fq->net_dev->dev.parent;
+
+ if (dpaa_fq->fqid == 0)
+ dpaa_fq->flags |= QMAN_FQ_FLAG_DYNAMIC_FQID;
+
+ dpaa_fq->init = !(dpaa_fq->flags & QMAN_FQ_FLAG_NO_MODIFY);
+
+ err = qman_create_fq(dpaa_fq->fqid, dpaa_fq->flags, &dpaa_fq->fq_base);
+ if (err) {
+ dev_err(dev, "qman_create_fq() failed\n");
+ return err;
+ }
+ fq = &dpaa_fq->fq_base;
+
+ if (dpaa_fq->init) {
+ memset(&initfq, 0, sizeof(initfq));
+
+ initfq.we_mask = QM_INITFQ_WE_FQCTRL;
+ /* Note: we may get to keep an empty FQ in cache */
+ initfq.fqd.fq_ctrl = QM_FQCTRL_PREFERINCACHE;
+
+ /* Try to reduce the number of portal interrupts for
+ * Tx Confirmation FQs.
+ */
+ if (dpaa_fq->fq_type == FQ_TYPE_TX_CONFIRM)
+ initfq.fqd.fq_ctrl |= QM_FQCTRL_HOLDACTIVE;
+
+ /* FQ placement */
+ initfq.we_mask |= QM_INITFQ_WE_DESTWQ;
+
+ qm_fqd_set_destwq(&initfq.fqd, dpaa_fq->channel, dpaa_fq->wq);
+
+ /* Put all egress queues in a congestion group of their own.
+ * Sensu stricto, the Tx confirmation queues are Rx FQs,
+ * rather than Tx - but they nonetheless account for the
+ * memory footprint on behalf of egress traffic. We therefore
+ * place them in the netdev's CGR, along with the Tx FQs.
+ */
+ if (dpaa_fq->fq_type == FQ_TYPE_TX ||
+ dpaa_fq->fq_type == FQ_TYPE_TX_CONFIRM ||
+ dpaa_fq->fq_type == FQ_TYPE_TX_CONF_MQ) {
+ initfq.we_mask |= QM_INITFQ_WE_CGID;
+ initfq.fqd.fq_ctrl |= QM_FQCTRL_CGE;
+ initfq.fqd.cgid = (u8)priv->cgr_data.cgr.cgrid;
+ /* Set a fixed overhead accounting, in an attempt to
+ * reduce the impact of fixed-size skb shells and the
+ * driver's needed headroom on system memory. This is
+ * especially the case when the egress traffic is
+ * composed of small datagrams.
+ * Unfortunately, QMan's OAL value is capped to an
+ * insufficient value, but even that is better than
+ * no overhead accounting at all.
+ */
+ initfq.we_mask |= QM_INITFQ_WE_OAC;
+ qm_fqd_set_oac(&initfq.fqd, QM_OAC_CG);
+ qm_fqd_set_oal(&initfq.fqd,
+ min(sizeof(struct sk_buff) +
+ priv->tx_headroom,
+ (size_t)FSL_QMAN_MAX_OAL));
+ }
+
+ if (td_enable) {
+ initfq.we_mask |= QM_INITFQ_WE_TDTHRESH;
+ qm_fqd_set_taildrop(&initfq.fqd, DPAA_FQ_TD, 1);
+ initfq.fqd.fq_ctrl = QM_FQCTRL_TDE;
+ }
+
+ if (dpaa_fq->fq_type == FQ_TYPE_TX) {
+ queue_id = dpaa_tx_fq_to_id(priv, &dpaa_fq->fq_base);
+ if (queue_id >= 0)
+ confq = priv->conf_fqs[queue_id];
+ if (confq) {
+ initfq.we_mask |= QM_INITFQ_WE_CONTEXTA;
+ /* ContextA: OVOM=1(use contextA2 bits instead of ICAD)
+ * A2V=1 (contextA A2 field is valid)
+ * A0V=1 (contextA A0 field is valid)
+ * B0V=1 (contextB field is valid)
+ * ContextA A2: EBD=1 (deallocate buffers inside FMan)
+ * ContextB B0(ASPID): 0 (absolute Virtual Storage ID)
+ */
+ initfq.fqd.context_a.hi = 0x1e000000;
+ initfq.fqd.context_a.lo = 0x80000000;
+ }
+ }
+
+ /* Put all the ingress queues in our "ingress CGR". */
+ if (priv->use_ingress_cgr &&
+ (dpaa_fq->fq_type == FQ_TYPE_RX_DEFAULT ||
+ dpaa_fq->fq_type == FQ_TYPE_RX_ERROR)) {
+ initfq.we_mask |= QM_INITFQ_WE_CGID;
+ initfq.fqd.fq_ctrl |= QM_FQCTRL_CGE;
+ initfq.fqd.cgid = (u8)priv->ingress_cgr.cgrid;
+ /* Set a fixed overhead accounting, just like for the
+ * egress CGR.
+ */
+ initfq.we_mask |= QM_INITFQ_WE_OAC;
+ qm_fqd_set_oac(&initfq.fqd, QM_OAC_CG);
+ qm_fqd_set_oal(&initfq.fqd,
+ min(sizeof(struct sk_buff) +
+ priv->tx_headroom,
+ (size_t)FSL_QMAN_MAX_OAL));
+ }
+
+ /* Initialization common to all ingress queues */
+ if (dpaa_fq->flags & QMAN_FQ_FLAG_NO_ENQUEUE) {
+ initfq.we_mask |= QM_INITFQ_WE_CONTEXTA;
+ initfq.fqd.fq_ctrl |=
+ QM_FQCTRL_HOLDACTIVE;
+ initfq.fqd.context_a.stashing.exclusive =
+ QM_STASHING_EXCL_DATA | QM_STASHING_EXCL_CTX |
+ QM_STASHING_EXCL_ANNOTATION;
+ qm_fqd_set_stashing(&initfq.fqd, 1, 2,
+ DIV_ROUND_UP(sizeof(struct qman_fq),
+ 64));
+ }
+
+ err = qman_init_fq(fq, QMAN_INITFQ_FLAG_SCHED, &initfq);
+ if (err < 0) {
+ dev_err(dev, "qman_init_fq(%u) = %d\n",
+ qman_fq_fqid(fq), err);
+ qman_destroy_fq(fq);
+ return err;
+ }
+ }
+
+ dpaa_fq->fqid = qman_fq_fqid(fq);
+
+ return 0;
+}
+
+static int dpaa_fq_free_entry(struct device *dev, struct qman_fq *fq)
+{
+ const struct dpaa_priv *priv;
+ struct dpaa_fq *dpaa_fq;
+ int err, error;
+
+ err = 0;
+
+ dpaa_fq = container_of(fq, struct dpaa_fq, fq_base);
+ priv = netdev_priv(dpaa_fq->net_dev);
+
+ if (dpaa_fq->init) {
+ err = qman_retire_fq(fq, NULL);
+ if (err < 0 && netif_msg_drv(priv))
+ dev_err(dev, "qman_retire_fq(%u) = %d\n",
+ qman_fq_fqid(fq), err);
+
+ error = qman_oos_fq(fq);
+ if (error < 0 && netif_msg_drv(priv)) {
+ dev_err(dev, "qman_oos_fq(%u) = %d\n",
+ qman_fq_fqid(fq), error);
+ if (err >= 0)
+ err = error;
+ }
+ }
+
+ qman_destroy_fq(fq);
+ list_del(&dpaa_fq->list);
+
+ return err;
+}
+
+static int dpaa_fq_free(struct device *dev, struct list_head *list)
+{
+ struct dpaa_fq *dpaa_fq, *tmp;
+ int err, error;
+
+ err = 0;
+ list_for_each_entry_safe(dpaa_fq, tmp, list, list) {
+ error = dpaa_fq_free_entry(dev, (struct qman_fq *)dpaa_fq);
+ if (error < 0 && err >= 0)
+ err = error;
+ }
+
+ return err;
+}
+
+static void dpaa_eth_init_tx_port(struct fman_port *port, struct dpaa_fq *errq,
+ struct dpaa_fq *defq,
+ struct dpaa_buffer_layout *buf_layout)
+{
+ struct fman_buffer_prefix_content buf_prefix_content;
+ struct fman_port_params params;
+ int err;
+
+ memset(¶ms, 0, sizeof(params));
+ memset(&buf_prefix_content, 0, sizeof(buf_prefix_content));
+
+ buf_prefix_content.priv_data_size = buf_layout->priv_data_size;
+ buf_prefix_content.pass_prs_result = true;
+ buf_prefix_content.pass_hash_result = true;
+ buf_prefix_content.pass_time_stamp = false;
+ buf_prefix_content.data_align = DPAA_FD_DATA_ALIGNMENT;
+
+ params.specific_params.non_rx_params.err_fqid = errq->fqid;
+ params.specific_params.non_rx_params.dflt_fqid = defq->fqid;
+
+ err = fman_port_config(port, ¶ms);
+ if (err)
+ pr_err("%s: fman_port_config failed\n", __func__);
+
+ err = fman_port_cfg_buf_prefix_content(port, &buf_prefix_content);
+ if (err)
+ pr_err("%s: fman_port_cfg_buf_prefix_content failed\n",
+ __func__);
+
+ err = fman_port_init(port);
+ if (err)
+ pr_err("%s: fm_port_init failed\n", __func__);
+}
+
+static void dpaa_eth_init_rx_port(struct fman_port *port, struct dpaa_bp **bps,
+ size_t count, struct dpaa_fq *errq,
+ struct dpaa_fq *defq,
+ struct dpaa_buffer_layout *buf_layout)
+{
+ struct fman_buffer_prefix_content buf_prefix_content;
+ struct fman_port_rx_params *rx_p;
+ struct fman_port_params params;
+ int i, err;
+
+ memset(¶ms, 0, sizeof(params));
+ memset(&buf_prefix_content, 0, sizeof(buf_prefix_content));
+
+ buf_prefix_content.priv_data_size = buf_layout->priv_data_size;
+ buf_prefix_content.pass_prs_result = true;
+ buf_prefix_content.pass_hash_result = true;
+ buf_prefix_content.pass_time_stamp = false;
+ buf_prefix_content.data_align = DPAA_FD_DATA_ALIGNMENT;
+
+ rx_p = ¶ms.specific_params.rx_params;
+ rx_p->err_fqid = errq->fqid;
+ rx_p->dflt_fqid = defq->fqid;
+
+ count = min(ARRAY_SIZE(rx_p->ext_buf_pools.ext_buf_pool), count);
+ rx_p->ext_buf_pools.num_of_pools_used = (u8)count;
+ for (i = 0; i < count; i++) {
+ rx_p->ext_buf_pools.ext_buf_pool[i].id = bps[i]->bpid;
+ rx_p->ext_buf_pools.ext_buf_pool[i].size = (u16)bps[i]->size;
+ }
+
+ err = fman_port_config(port, ¶ms);
+ if (err)
+ pr_err("%s: fman_port_config failed\n", __func__);
+
+ err = fman_port_cfg_buf_prefix_content(port, &buf_prefix_content);
+ if (err)
+ pr_err("%s: fman_port_cfg_buf_prefix_content failed\n",
+ __func__);
+
+ err = fman_port_init(port);
+ if (err)
+ pr_err("%s: fm_port_init failed\n", __func__);
+}
+
+static void dpaa_eth_init_ports(struct mac_device *mac_dev,
+ struct dpaa_bp **bps, size_t count,
+ struct fm_port_fqs *port_fqs,
+ struct dpaa_buffer_layout *buf_layout,
+ struct device *dev)
+{
+ struct fman_port *rxport = mac_dev->port[RX];
+ struct fman_port *txport = mac_dev->port[TX];
+
+ dpaa_eth_init_tx_port(txport, port_fqs->tx_errq,
+ port_fqs->tx_defq, &buf_layout[TX]);
+ dpaa_eth_init_rx_port(rxport, bps, count, port_fqs->rx_errq,
+ port_fqs->rx_defq, &buf_layout[RX]);
+}
+
+static int dpaa_bman_release(const struct dpaa_bp *dpaa_bp,
+ struct bm_buffer *bmb, int cnt)
+{
+ int err;
+
+ err = bman_release(dpaa_bp->pool, bmb, cnt);
+ /* Should never occur, address anyway to avoid leaking the buffers */
+ if (unlikely(WARN_ON(err)) && dpaa_bp->free_buf_cb)
+ while (cnt-- > 0)
+ dpaa_bp->free_buf_cb(dpaa_bp, &bmb[cnt]);
+
+ return cnt;
+}
+
+static void dpaa_release_sgt_members(struct qm_sg_entry *sgt)
+{
+ struct bm_buffer bmb[DPAA_BUFF_RELEASE_MAX];
+ struct dpaa_bp *dpaa_bp;
+ int i = 0, j;
+
+ memset(bmb, 0, sizeof(bmb));
+
+ do {
+ dpaa_bp = dpaa_bpid2pool(sgt[i].bpid);
+ if (!dpaa_bp)
+ return;
+
+ j = 0;
+ do {
+ WARN_ON(qm_sg_entry_is_ext(&sgt[i]));
+
+ bm_buffer_set64(&bmb[j], qm_sg_entry_get64(&sgt[i]));
+
+ j++; i++;
+ } while (j < ARRAY_SIZE(bmb) &&
+ !qm_sg_entry_is_final(&sgt[i - 1]) &&
+ sgt[i - 1].bpid == sgt[i].bpid);
+
+ dpaa_bman_release(dpaa_bp, bmb, j);
+ } while (!qm_sg_entry_is_final(&sgt[i - 1]));
+}
+
+static void dpaa_fd_release(const struct net_device *net_dev,
+ const struct qm_fd *fd)
+{
+ struct qm_sg_entry *sgt;
+ struct dpaa_bp *dpaa_bp;
+ struct bm_buffer bmb;
+ dma_addr_t addr;
+ void *vaddr;
+
+ bmb.data = 0;
+ bm_buffer_set64(&bmb, qm_fd_addr(fd));
+
+ dpaa_bp = dpaa_bpid2pool(fd->bpid);
+ if (!dpaa_bp)
+ return;
+
+ if (qm_fd_get_format(fd) == qm_fd_sg) {
+ vaddr = phys_to_virt(qm_fd_addr(fd));
+ sgt = vaddr + qm_fd_get_offset(fd);
+
+ dma_unmap_single(dpaa_bp->dev, qm_fd_addr(fd), dpaa_bp->size,
+ DMA_FROM_DEVICE);
+
+ dpaa_release_sgt_members(sgt);
+
+ addr = dma_map_single(dpaa_bp->dev, vaddr, dpaa_bp->size,
+ DMA_FROM_DEVICE);
+ if (dma_mapping_error(dpaa_bp->dev, addr)) {
+ dev_err(dpaa_bp->dev, "DMA mapping failed");
+ return;
+ }
+ bm_buffer_set64(&bmb, addr);
+ }
+
+ dpaa_bman_release(dpaa_bp, &bmb, 1);
+}
+
+/* Turn on HW checksum computation for this outgoing frame.
+ * If the current protocol is not something we support in this regard
+ * (or if the stack has already computed the SW checksum), we do nothing.
+ *
+ * Returns 0 if all goes well (or HW csum doesn't apply), and a negative value
+ * otherwise.
+ *
+ * Note that this function may modify the fd->cmd field and the skb data buffer
+ * (the Parse Results area).
+ */
+static int dpaa_enable_tx_csum(struct dpaa_priv *priv,
+ struct sk_buff *skb,
+ struct qm_fd *fd,
+ char *parse_results)
+{
+ struct fman_prs_result *parse_result;
+ u16 ethertype = ntohs(skb->protocol);
+ struct ipv6hdr *ipv6h = NULL;
+ struct iphdr *iph;
+ int retval = 0;
+ u8 l4_proto;
+
+ if (skb->ip_summed != CHECKSUM_PARTIAL)
+ return 0;
+
+ /* Note: L3 csum seems to be already computed in sw, but we can't choose
+ * L4 alone from the FM configuration anyway.
+ */
+
+ /* Fill in some fields of the Parse Results array, so the FMan
+ * can find them as if they came from the FMan Parser.
+ */
+ parse_result = (struct fman_prs_result *)parse_results;
+
+ /* If we're dealing with VLAN, get the real Ethernet type */
+ if (ethertype == ETH_P_8021Q) {
+ /* We can't always assume the MAC header is set correctly
+ * by the stack, so reset to beginning of skb->data
+ */
+ skb_reset_mac_header(skb);
+ ethertype = ntohs(vlan_eth_hdr(skb)->h_vlan_encapsulated_proto);
+ }
+
+ /* Fill in the relevant L3 parse result fields
+ * and read the L4 protocol type
+ */
+ switch (ethertype) {
+ case ETH_P_IP:
+ parse_result->l3r = cpu_to_be16(FM_L3_PARSE_RESULT_IPV4);
+ iph = ip_hdr(skb);
+ WARN_ON(!iph);
+ l4_proto = iph->protocol;
+ break;
+ case ETH_P_IPV6:
+ parse_result->l3r = cpu_to_be16(FM_L3_PARSE_RESULT_IPV6);
+ ipv6h = ipv6_hdr(skb);
+ WARN_ON(!ipv6h);
+ l4_proto = ipv6h->nexthdr;
+ break;
+ default:
+ /* We shouldn't even be here */
+ if (net_ratelimit())
+ netif_alert(priv, tx_err, priv->net_dev,
+ "Can't compute HW csum for L3 proto 0x%x\n",
+ ntohs(skb->protocol));
+ retval = -EIO;
+ goto return_error;
+ }
+
+ /* Fill in the relevant L4 parse result fields */
+ switch (l4_proto) {
+ case IPPROTO_UDP:
+ parse_result->l4r = FM_L4_PARSE_RESULT_UDP;
+ break;
+ case IPPROTO_TCP:
+ parse_result->l4r = FM_L4_PARSE_RESULT_TCP;
+ break;
+ default:
+ if (net_ratelimit())
+ netif_alert(priv, tx_err, priv->net_dev,
+ "Can't compute HW csum for L4 proto 0x%x\n",
+ l4_proto);
+ retval = -EIO;
+ goto return_error;
+ }
+
+ /* At index 0 is IPOffset_1 as defined in the Parse Results */
+ parse_result->ip_off[0] = (u8)skb_network_offset(skb);
+ parse_result->l4_off = (u8)skb_transport_offset(skb);
+
+ /* Enable L3 (and L4, if TCP or UDP) HW checksum. */
+ fd->cmd |= FM_FD_CMD_RPD | FM_FD_CMD_DTC;
+
+ /* On P1023 and similar platforms fd->cmd interpretation could
+ * be disabled by setting CONTEXT_A bit ICMD; currently this bit
+ * is not set so we do not need to check; in the future, if/when
+ * using context_a we need to check this bit
+ */
+
+return_error:
+ return retval;
+}
+
+static int dpaa_bp_add_8_bufs(const struct dpaa_bp *dpaa_bp)
+{
+ struct device *dev = dpaa_bp->dev;
+ struct bm_buffer bmb[8];
+ dma_addr_t addr;
+ void *new_buf;
+ u8 i;
+
+ for (i = 0; i < 8; i++) {
+ new_buf = netdev_alloc_frag(dpaa_bp->raw_size);
+ if (unlikely(!new_buf)) {
+ dev_err(dev, "netdev_alloc_frag() failed, size %zu\n",
+ dpaa_bp->raw_size);
+ goto release_previous_buffs;
+ }
+ new_buf = PTR_ALIGN(new_buf, SMP_CACHE_BYTES);
+
+ addr = dma_map_single(dev, new_buf,
+ dpaa_bp->size, DMA_FROM_DEVICE);
+ if (unlikely(dma_mapping_error(dev, addr))) {
+ dev_err(dpaa_bp->dev, "DMA map failed");
+ goto release_previous_buffs;
+ }
+
+ bmb[i].data = 0;
+ bm_buffer_set64(&bmb[i], addr);
+ }
+
+release_bufs:
+ return dpaa_bman_release(dpaa_bp, bmb, i);
+
+release_previous_buffs:
+ WARN_ONCE(1, "dpaa_eth: failed to add buffers on Rx\n");
+
+ bm_buffer_set64(&bmb[i], 0);
+ /* Avoid releasing a completely null buffer; bman_release() requires
+ * at least one buffer.
+ */
+ if (likely(i))
+ goto release_bufs;
+
+ return 0;
+}
+
+static int dpaa_bp_seed(struct dpaa_bp *dpaa_bp)
+{
+ int i;
+
+ /* Give each CPU an allotment of "config_count" buffers */
+ for_each_possible_cpu(i) {
+ int *count_ptr = per_cpu_ptr(dpaa_bp->percpu_count, i);
+ int j;
+
+ /* Although we access another CPU's counters here
+ * we do it at boot time so it is safe
+ */
+ for (j = 0; j < dpaa_bp->config_count; j += 8)
+ *count_ptr += dpaa_bp_add_8_bufs(dpaa_bp);
+ }
+ return 0;
+}
+
+/* Add buffers/(pages) for Rx processing whenever bpool count falls below
+ * REFILL_THRESHOLD.
+ */
+static int dpaa_eth_refill_bpool(struct dpaa_bp *dpaa_bp, int *countptr)
+{
+ int count = *countptr;
+ int new_bufs;
+
+ if (unlikely(count < FSL_DPAA_ETH_REFILL_THRESHOLD)) {
+ do {
+ new_bufs = dpaa_bp_add_8_bufs(dpaa_bp);
+ if (unlikely(!new_bufs)) {
+ /* Avoid looping forever if we've temporarily
+ * run out of memory. We'll try again at the
+ * next NAPI cycle.
+ */
+ break;
+ }
+ count += new_bufs;
+ } while (count < FSL_DPAA_ETH_MAX_BUF_COUNT);
+
+ *countptr = count;
+ if (unlikely(count < FSL_DPAA_ETH_MAX_BUF_COUNT))
+ return -ENOMEM;
+ }
+
+ return 0;
+}
+
+static int dpaa_eth_refill_bpools(struct dpaa_priv *priv)
+{
+ struct dpaa_bp *dpaa_bp;
+ int *countptr;
+ int res, i;
+
+ for (i = 0; i < DPAA_BPS_NUM; i++) {
+ dpaa_bp = priv->dpaa_bps[i];
+ if (!dpaa_bp)
+ return -EINVAL;
+ countptr = this_cpu_ptr(dpaa_bp->percpu_count);
+ res = dpaa_eth_refill_bpool(dpaa_bp, countptr);
+ if (res)
+ return res;
+ }
+ return 0;
+}
+
+/* Cleanup function for outgoing frame descriptors that were built on Tx path,
+ * either contiguous frames or scatter/gather ones.
+ * Skb freeing is not handled here.
+ *
+ * This function may be called on error paths in the Tx function, so guard
+ * against cases when not all fd relevant fields were filled in.
+ *
+ * Return the skb backpointer, since for S/G frames the buffer containing it
+ * gets freed here.
+ */
+static struct sk_buff *dpaa_cleanup_tx_fd(const struct dpaa_priv *priv,
+ const struct qm_fd *fd)
+{
+ const enum dma_data_direction dma_dir = DMA_TO_DEVICE;
+ struct device *dev = priv->net_dev->dev.parent;
+ dma_addr_t addr = qm_fd_addr(fd);
+ const struct qm_sg_entry *sgt;
+ struct sk_buff **skbh, *skb;
+ int nr_frags, i;
+
+ skbh = (struct sk_buff **)phys_to_virt(addr);
+ skb = *skbh;
+
+ if (unlikely(qm_fd_get_format(fd) == qm_fd_sg)) {
+ nr_frags = skb_shinfo(skb)->nr_frags;
+ dma_unmap_single(dev, addr, qm_fd_get_offset(fd) +
+ sizeof(struct qm_sg_entry) * (1 + nr_frags),
+ dma_dir);
+
+ /* The sgt buffer has been allocated with netdev_alloc_frag(),
+ * it's from lowmem.
+ */
+ sgt = phys_to_virt(addr + qm_fd_get_offset(fd));
+
+ /* sgt[0] is from lowmem, was dma_map_single()-ed */
+ dma_unmap_single(dev, qm_sg_addr(&sgt[0]),
+ qm_sg_entry_get_len(&sgt[0]), dma_dir);
+
+ /* remaining pages were mapped with skb_frag_dma_map() */
+ for (i = 1; i < nr_frags; i++) {
+ WARN_ON(qm_sg_entry_is_ext(&sgt[i]));
+
+ dma_unmap_page(dev, qm_sg_addr(&sgt[i]),
+ qm_sg_entry_get_len(&sgt[i]), dma_dir);
+ }
+
+ /* Free the page frag that we allocated on Tx */
+ skb_free_frag(phys_to_virt(addr));
+ } else {
+ dma_unmap_single(dev, addr,
+ skb_tail_pointer(skb) - (u8 *)skbh, dma_dir);
+ }
+
+ return skb;
+}
+
+/* Build a linear skb around the received buffer.
+ * We are guaranteed there is enough room at the end of the data buffer to
+ * accommodate the shared info area of the skb.
+ */
+static struct sk_buff *contig_fd_to_skb(const struct dpaa_priv *priv,
+ const struct qm_fd *fd)
+{
+ ssize_t fd_off = qm_fd_get_offset(fd);
+ dma_addr_t addr = qm_fd_addr(fd);
+ struct dpaa_bp *dpaa_bp;
+ struct sk_buff *skb;
+ void *vaddr;
+
+ vaddr = phys_to_virt(addr);
+ WARN_ON(!IS_ALIGNED((unsigned long)vaddr, SMP_CACHE_BYTES));
+
+ dpaa_bp = dpaa_bpid2pool(fd->bpid);
+ if (!dpaa_bp)
+ goto free_buffer;
+
+ skb = build_skb(vaddr, dpaa_bp->size +
+ SKB_DATA_ALIGN(sizeof(struct skb_shared_info)));
+ if (unlikely(!skb)) {
+ WARN_ONCE(1, "Build skb failure on Rx\n");
+ goto free_buffer;
+ }
+ WARN_ON(fd_off != priv->rx_headroom);
+ skb_reserve(skb, fd_off);
+ skb_put(skb, qm_fd_get_length(fd));
+
+ skb->ip_summed = CHECKSUM_NONE;
+
+ return skb;
+
+free_buffer:
+ skb_free_frag(vaddr);
+ return NULL;
+}
+
+/* Build an skb with the data of the first S/G entry in the linear portion and
+ * the rest of the frame as skb fragments.
+ *
+ * The page fragment holding the S/G Table is recycled here.
+ */
+static struct sk_buff *sg_fd_to_skb(const struct dpaa_priv *priv,
+ const struct qm_fd *fd)
+{
+ ssize_t fd_off = qm_fd_get_offset(fd);
+ dma_addr_t addr = qm_fd_addr(fd);
+ const struct qm_sg_entry *sgt;
+ struct page *page, *head_page;
+ struct dpaa_bp *dpaa_bp;
+ void *vaddr, *sg_vaddr;
+ int frag_off, frag_len;
+ struct sk_buff *skb;
+ dma_addr_t sg_addr;
+ int page_offset;
+ unsigned int sz;
+ int *count_ptr;
+ int i;
+
+ vaddr = phys_to_virt(addr);
+ WARN_ON(!IS_ALIGNED((unsigned long)vaddr, SMP_CACHE_BYTES));
+
+ /* Iterate through the SGT entries and add data buffers to the skb */
+ sgt = vaddr + fd_off;
+ for (i = 0; i < DPAA_SGT_MAX_ENTRIES; i++) {
+ /* Extension bit is not supported */
+ WARN_ON(qm_sg_entry_is_ext(&sgt[i]));
+
+ sg_addr = qm_sg_addr(&sgt[i]);
+ sg_vaddr = phys_to_virt(sg_addr);
+ WARN_ON(!IS_ALIGNED((unsigned long)sg_vaddr,
+ SMP_CACHE_BYTES));
+
+ /* We may use multiple Rx pools */
+ dpaa_bp = dpaa_bpid2pool(sgt[i].bpid);
+ if (!dpaa_bp)
+ goto free_buffers;
+
+ count_ptr = this_cpu_ptr(dpaa_bp->percpu_count);
+ dma_unmap_single(dpaa_bp->dev, sg_addr, dpaa_bp->size,
+ DMA_FROM_DEVICE);
+ if (i == 0) {
+ sz = dpaa_bp->size +
+ SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
+ skb = build_skb(sg_vaddr, sz);
+ if (WARN_ON(unlikely(!skb)))
+ goto free_buffers;
+
+ skb->ip_summed = CHECKSUM_NONE;
+
+ /* Make sure forwarded skbs will have enough space
+ * on Tx, if extra headers are added.
+ */
+ WARN_ON(fd_off != priv->rx_headroom);
+ skb_reserve(skb, fd_off);
+ skb_put(skb, qm_sg_entry_get_len(&sgt[i]));
+ } else {
+ /* Not the first S/G entry; all data from buffer will
+ * be added in an skb fragment; fragment index is offset
+ * by one since first S/G entry was incorporated in the
+ * linear part of the skb.
+ *
+ * Caution: 'page' may be a tail page.
+ */
+ page = virt_to_page(sg_vaddr);
+ head_page = virt_to_head_page(sg_vaddr);
+
+ /* Compute offset in (possibly tail) page */
+ page_offset = ((unsigned long)sg_vaddr &
+ (PAGE_SIZE - 1)) +
+ (page_address(page) - page_address(head_page));
+ /* page_offset only refers to the beginning of sgt[i];
+ * but the buffer itself may have an internal offset.
+ */
+ frag_off = qm_sg_entry_get_off(&sgt[i]) + page_offset;
+ frag_len = qm_sg_entry_get_len(&sgt[i]);
+ /* skb_add_rx_frag() does no checking on the page; if
+ * we pass it a tail page, we'll end up with
+ * bad page accounting and eventually with segafults.
+ */
+ skb_add_rx_frag(skb, i - 1, head_page, frag_off,
+ frag_len, dpaa_bp->size);
+ }
+ /* Update the pool count for the current {cpu x bpool} */
+ (*count_ptr)--;
+
+ if (qm_sg_entry_is_final(&sgt[i]))
+ break;
+ }
+ WARN_ONCE(i == DPAA_SGT_MAX_ENTRIES, "No final bit on SGT\n");
+
+ /* free the SG table buffer */
+ skb_free_frag(vaddr);
+
+ return skb;
+
+free_buffers:
+ /* compensate sw bpool counter changes */
+ for (i--; i > 0; i--) {
+ dpaa_bp = dpaa_bpid2pool(sgt[i].bpid);
+ if (dpaa_bp) {
+ count_ptr = this_cpu_ptr(dpaa_bp->percpu_count);
+ (*count_ptr)++;
+ }
+ }
+ /* free all the SG entries */
+ for (i = 0; i < DPAA_SGT_MAX_ENTRIES ; i++) {
+ sg_addr = qm_sg_addr(&sgt[i]);
+ sg_vaddr = phys_to_virt(sg_addr);
+ skb_free_frag(sg_vaddr);
+ dpaa_bp = dpaa_bpid2pool(sgt[i].bpid);
+ if (dpaa_bp) {
+ count_ptr = this_cpu_ptr(dpaa_bp->percpu_count);
+ (*count_ptr)--;
+ }
+
+ if (qm_sg_entry_is_final(&sgt[i]))
+ break;
+ }
+ /* free the SGT fragment */
+ skb_free_frag(vaddr);
+
+ return NULL;
+}
+
+static int skb_to_contig_fd(struct dpaa_priv *priv,
+ struct sk_buff *skb, struct qm_fd *fd,
+ int *offset)
+{
+ struct net_device *net_dev = priv->net_dev;
+ struct device *dev = net_dev->dev.parent;
+ enum dma_data_direction dma_dir;
+ unsigned char *buffer_start;
+ struct sk_buff **skbh;
+ dma_addr_t addr;
+ int err;
+
+ /* We are guaranteed to have at least tx_headroom bytes
+ * available, so just use that for offset.
+ */
+ fd->bpid = FSL_DPAA_BPID_INV;
+ buffer_start = skb->data - priv->tx_headroom;
+ dma_dir = DMA_TO_DEVICE;
+
+ skbh = (struct sk_buff **)buffer_start;
+ *skbh = skb;
+
+ /* Enable L3/L4 hardware checksum computation.
+ *
+ * We must do this before dma_map_single(DMA_TO_DEVICE), because we may
+ * need to write into the skb.
+ */
+ err = dpaa_enable_tx_csum(priv, skb, fd,
+ ((char *)skbh) + DPAA_TX_PRIV_DATA_SIZE);
+ if (unlikely(err < 0)) {
+ if (net_ratelimit())
+ netif_err(priv, tx_err, net_dev, "HW csum error: %d\n",
+ err);
+ return err;
+ }
+
+ /* Fill in the rest of the FD fields */
+ qm_fd_set_contig(fd, priv->tx_headroom, skb->len);
+ fd->cmd |= FM_FD_CMD_FCO;
+
+ /* Map the entire buffer size that may be seen by FMan, but no more */
+ addr = dma_map_single(dev, skbh,
+ skb_tail_pointer(skb) - buffer_start, dma_dir);
+ if (unlikely(dma_mapping_error(dev, addr))) {
+ if (net_ratelimit())
+ netif_err(priv, tx_err, net_dev, "dma_map_single() failed\n");
+ return -EINVAL;
+ }
+ qm_fd_addr_set64(fd, addr);
+
+ return 0;
+}
+
+static int skb_to_sg_fd(struct dpaa_priv *priv,
+ struct sk_buff *skb, struct qm_fd *fd)
+{
+ const enum dma_data_direction dma_dir = DMA_TO_DEVICE;
+ const int nr_frags = skb_shinfo(skb)->nr_frags;
+ struct net_device *net_dev = priv->net_dev;
+ struct device *dev = net_dev->dev.parent;
+ struct qm_sg_entry *sgt;
+ struct sk_buff **skbh;
+ int i, j, err, sz;
+ void *buffer_start;
+ skb_frag_t *frag;
+ dma_addr_t addr;
+ size_t frag_len;
+ void *sgt_buf;
+
+ /* get a page frag to store the SGTable */
+ sz = SKB_DATA_ALIGN(priv->tx_headroom +
+ sizeof(struct qm_sg_entry) * (1 + nr_frags));
+ sgt_buf = netdev_alloc_frag(sz);
+ if (unlikely(!sgt_buf)) {
+ netdev_err(net_dev, "netdev_alloc_frag() failed for size %d\n",
+ sz);
+ return -ENOMEM;
+ }
+
+ /* Enable L3/L4 hardware checksum computation.
+ *
+ * We must do this before dma_map_single(DMA_TO_DEVICE), because we may
+ * need to write into the skb.
+ */
+ err = dpaa_enable_tx_csum(priv, skb, fd,
+ sgt_buf + DPAA_TX_PRIV_DATA_SIZE);
+ if (unlikely(err < 0)) {
+ if (net_ratelimit())
+ netif_err(priv, tx_err, net_dev, "HW csum error: %d\n",
+ err);
+ goto csum_failed;
+ }
+
+ sgt = (struct qm_sg_entry *)(sgt_buf + priv->tx_headroom);
+ qm_sg_entry_set_len(&sgt[0], skb_headlen(skb));
+ sgt[0].bpid = FSL_DPAA_BPID_INV;
+ sgt[0].offset = 0;
+ addr = dma_map_single(dev, skb->data,
+ skb_headlen(skb), dma_dir);
+ if (unlikely(dma_mapping_error(dev, addr))) {
+ dev_err(dev, "DMA mapping failed");
+ err = -EINVAL;
+ goto sg0_map_failed;
+ }
+ qm_sg_entry_set64(&sgt[0], addr);
+
+ /* populate the rest of SGT entries */
+ frag = &skb_shinfo(skb)->frags[0];
+ frag_len = frag->size;
+ for (i = 1; i <= nr_frags; i++, frag++) {
+ WARN_ON(!skb_frag_page(frag));
+ addr = skb_frag_dma_map(dev, frag, 0,
+ frag_len, dma_dir);
+ if (unlikely(dma_mapping_error(dev, addr))) {
+ dev_err(dev, "DMA mapping failed");
+ err = -EINVAL;
+ goto sg_map_failed;
+ }
+
+ qm_sg_entry_set_len(&sgt[i], frag_len);
+ sgt[i].bpid = FSL_DPAA_BPID_INV;
+ sgt[i].offset = 0;
+
+ /* keep the offset in the address */
+ qm_sg_entry_set64(&sgt[i], addr);
+ frag_len = frag->size;
+ }
+ qm_sg_entry_set_f(&sgt[i - 1], frag_len);
+
+ qm_fd_set_sg(fd, priv->tx_headroom, skb->len);
+
+ /* DMA map the SGT page */
+ buffer_start = (void *)sgt - priv->tx_headroom;
+ skbh = (struct sk_buff **)buffer_start;
+ *skbh = skb;
+
+ addr = dma_map_single(dev, buffer_start, priv->tx_headroom +
+ sizeof(struct qm_sg_entry) * (1 + nr_frags),
+ dma_dir);
+ if (unlikely(dma_mapping_error(dev, addr))) {
+ dev_err(dev, "DMA mapping failed");
+ err = -EINVAL;
+ goto sgt_map_failed;
+ }
+
+ fd->bpid = FSL_DPAA_BPID_INV;
+ fd->cmd |= FM_FD_CMD_FCO;
+ qm_fd_addr_set64(fd, addr);
+
+ return 0;
+
+sgt_map_failed:
+sg_map_failed:
+ for (j = 0; j < i; j++)
+ dma_unmap_page(dev, qm_sg_addr(&sgt[j]),
+ qm_sg_entry_get_len(&sgt[j]), dma_dir);
+sg0_map_failed:
+csum_failed:
+ skb_free_frag(sgt_buf);
+
+ return err;
+}
+
+static inline int dpaa_xmit(struct dpaa_priv *priv,
+ struct rtnl_link_stats64 *percpu_stats,
+ int queue,
+ struct qm_fd *fd)
+{
+ struct qman_fq *egress_fq;
+ int err, i;
+
+ egress_fq = priv->egress_fqs[queue];
+ if (fd->bpid == FSL_DPAA_BPID_INV)
+ fd->cmd |= qman_fq_fqid(priv->conf_fqs[queue]);
+
+ for (i = 0; i < DPAA_ENQUEUE_RETRIES; i++) {
+ err = qman_enqueue(egress_fq, fd);
+ if (err != -EBUSY)
+ break;
+ }
+
+ if (unlikely(err < 0)) {
+ percpu_stats->tx_errors++;
+ percpu_stats->tx_fifo_errors++;
+ return err;
+ }
+
+ percpu_stats->tx_packets++;
+ percpu_stats->tx_bytes += qm_fd_get_length(fd);
+
+ return 0;
+}
+
+static int dpaa_start_xmit(struct sk_buff *skb, struct net_device *net_dev)
+{
+ const int queue_mapping = skb_get_queue_mapping(skb);
+ bool nonlinear = skb_is_nonlinear(skb);
+ struct rtnl_link_stats64 *percpu_stats;
+ struct dpaa_percpu_priv *percpu_priv;
+ struct dpaa_priv *priv;
+ struct qm_fd fd;
+ int offset = 0;
+ int err = 0;
+
+ priv = netdev_priv(net_dev);
+ percpu_priv = this_cpu_ptr(priv->percpu_priv);
+ percpu_stats = &percpu_priv->stats;
+
+ qm_fd_clear_fd(&fd);
+
+ if (!nonlinear) {
+ /* We're going to store the skb backpointer at the beginning
+ * of the data buffer, so we need a privately owned skb
+ *
+ * We've made sure skb is not shared in dev->priv_flags,
+ * we need to verify the skb head is not cloned
+ */
+ if (skb_cow_head(skb, priv->tx_headroom))
+ goto enomem;
+
+ WARN_ON(skb_is_nonlinear(skb));
+ }
+
+ /* MAX_SKB_FRAGS is equal or larger than our dpaa_SGT_MAX_ENTRIES;
+ * make sure we don't feed FMan with more fragments than it supports.
+ */
+ if (nonlinear &&
+ likely(skb_shinfo(skb)->nr_frags < DPAA_SGT_MAX_ENTRIES)) {
+ /* Just create a S/G fd based on the skb */
+ err = skb_to_sg_fd(priv, skb, &fd);
+ } else {
+ /* If the egress skb contains more fragments than we support
+ * we have no choice but to linearize it ourselves.
+ */
+ if (unlikely(nonlinear) && __skb_linearize(skb))
+ goto enomem;
+
+ /* Finally, create a contig FD from this skb */
+ err = skb_to_contig_fd(priv, skb, &fd, &offset);
+ }
+ if (unlikely(err < 0))
+ goto skb_to_fd_failed;
+
+ if (likely(dpaa_xmit(priv, percpu_stats, queue_mapping, &fd) == 0))
+ return NETDEV_TX_OK;
+
+ dpaa_cleanup_tx_fd(priv, &fd);
+skb_to_fd_failed:
+enomem:
+ percpu_stats->tx_errors++;
+ dev_kfree_skb(skb);
+ return NETDEV_TX_OK;
+}
+
+static void dpaa_rx_error(struct net_device *net_dev,
+ const struct dpaa_priv *priv,
+ struct dpaa_percpu_priv *percpu_priv,
+ const struct qm_fd *fd,
+ u32 fqid)
+{
+ if (net_ratelimit())
+ netif_err(priv, hw, net_dev, "Err FD status = 0x%08x\n",
+ fd->status & FM_FD_STAT_RX_ERRORS);
+
+ percpu_priv->stats.rx_errors++;
+
+ dpaa_fd_release(net_dev, fd);
+}
+
+static void dpaa_tx_error(struct net_device *net_dev,
+ const struct dpaa_priv *priv,
+ struct dpaa_percpu_priv *percpu_priv,
+ const struct qm_fd *fd,
+ u32 fqid)
+{
+ struct sk_buff *skb;
+
+ if (net_ratelimit())
+ netif_warn(priv, hw, net_dev, "FD status = 0x%08x\n",
+ fd->status & FM_FD_STAT_TX_ERRORS);
+
+ percpu_priv->stats.tx_errors++;
+
+ skb = dpaa_cleanup_tx_fd(priv, fd);
+ dev_kfree_skb(skb);
+}
+
+static int dpaa_eth_poll(struct napi_struct *napi, int budget)
+{
+ struct dpaa_napi_portal *np =
+ container_of(napi, struct dpaa_napi_portal, napi);
+
+ int cleaned = qman_p_poll_dqrr(np->p, budget);
+
+ if (cleaned < budget) {
+ napi_complete(napi);
+ qman_p_irqsource_add(np->p, QM_PIRQ_DQRI);
+
+ } else if (np->down) {
+ qman_p_irqsource_add(np->p, QM_PIRQ_DQRI);
+ }
+
+ return cleaned;
+}
+
+static void dpaa_tx_conf(struct net_device *net_dev,
+ const struct dpaa_priv *priv,
+ struct dpaa_percpu_priv *percpu_priv,
+ const struct qm_fd *fd,
+ u32 fqid)
+{
+ struct sk_buff *skb;
+
+ if (unlikely(fd->status & FM_FD_STAT_TX_ERRORS) != 0) {
+ if (net_ratelimit())
+ netif_warn(priv, hw, net_dev, "FD status = 0x%08x\n",
+ fd->status & FM_FD_STAT_TX_ERRORS);
+
+ percpu_priv->stats.tx_errors++;
+ }
+
+ skb = dpaa_cleanup_tx_fd(priv, fd);
+
+ consume_skb(skb);
+}
+
+static inline int dpaa_eth_napi_schedule(struct dpaa_percpu_priv *percpu_priv,
+ struct qman_portal *portal)
+{
+ if (unlikely(in_irq() || !in_serving_softirq())) {
+ /* Disable QMan IRQ and invoke NAPI */
+ qman_p_irqsource_remove(portal, QM_PIRQ_DQRI);
+
+ percpu_priv->np.p = portal;
+ napi_schedule(&percpu_priv->np.napi);
+ return 1;
+ }
+ return 0;
+}
+
+static enum qman_cb_dqrr_result rx_error_dqrr(struct qman_portal *portal,
+ struct qman_fq *fq,
+ const struct qm_dqrr_entry *dq)
+{
+ struct dpaa_fq *dpaa_fq = container_of(fq, struct dpaa_fq, fq_base);
+ struct dpaa_percpu_priv *percpu_priv;
+ struct net_device *net_dev;
+ struct dpaa_bp *dpaa_bp;
+ struct dpaa_priv *priv;
+
+ net_dev = dpaa_fq->net_dev;
+ priv = netdev_priv(net_dev);
+ dpaa_bp = dpaa_bpid2pool(dq->fd.bpid);
+ if (!dpaa_bp)
+ return qman_cb_dqrr_consume;
+
+ percpu_priv = this_cpu_ptr(priv->percpu_priv);
+
+ if (dpaa_eth_napi_schedule(percpu_priv, portal))
+ return qman_cb_dqrr_stop;
+
+ if (dpaa_eth_refill_bpools(priv))
+ /* Unable to refill the buffer pool due to insufficient
+ * system memory. Just release the frame back into the pool,
+ * otherwise we'll soon end up with an empty buffer pool.
+ */
+ dpaa_fd_release(net_dev, &dq->fd);
+ else
+ dpaa_rx_error(net_dev, priv, percpu_priv, &dq->fd, fq->fqid);
+
+ return qman_cb_dqrr_consume;
+}
+
+static enum qman_cb_dqrr_result rx_default_dqrr(struct qman_portal *portal,
+ struct qman_fq *fq,
+ const struct qm_dqrr_entry *dq)
+{
+ struct rtnl_link_stats64 *percpu_stats;
+ struct dpaa_percpu_priv *percpu_priv;
+ const struct qm_fd *fd = &dq->fd;
+ dma_addr_t addr = qm_fd_addr(fd);
+ enum qm_fd_format fd_format;
+ struct net_device *net_dev;
+ u32 fd_status = fd->status;
+ struct dpaa_bp *dpaa_bp;
+ struct dpaa_priv *priv;
+ unsigned int skb_len;
+ struct sk_buff *skb;
+ int *count_ptr;
+
+ net_dev = ((struct dpaa_fq *)fq)->net_dev;
+ priv = netdev_priv(net_dev);
+ dpaa_bp = dpaa_bpid2pool(dq->fd.bpid);
+ if (!dpaa_bp)
+ return qman_cb_dqrr_consume;
+
+ percpu_priv = this_cpu_ptr(priv->percpu_priv);
+ percpu_stats = &percpu_priv->stats;
+
+ if (unlikely(dpaa_eth_napi_schedule(percpu_priv, portal)))
+ return qman_cb_dqrr_stop;
+
+ /* Make sure we didn't run out of buffers */
+ if (unlikely(dpaa_eth_refill_bpools(priv))) {
+ /* Unable to refill the buffer pool due to insufficient
+ * system memory. Just release the frame back into the pool,
+ * otherwise we'll soon end up with an empty buffer pool.
+ */
+ dpaa_fd_release(net_dev, &dq->fd);
+ return qman_cb_dqrr_consume;
+ }
+
+ if (unlikely(fd_status & FM_FD_STAT_RX_ERRORS) != 0) {
+ if (net_ratelimit())
+ netif_warn(priv, hw, net_dev, "FD status = 0x%08x\n",
+ fd_status & FM_FD_STAT_RX_ERRORS);
+
+ percpu_stats->rx_errors++;
+ dpaa_fd_release(net_dev, fd);
+ return qman_cb_dqrr_consume;
+ }
+
+ dpaa_bp = dpaa_bpid2pool(fd->bpid);
+ if (!dpaa_bp)
+ return qman_cb_dqrr_consume;
+
+ dma_unmap_single(dpaa_bp->dev, addr, dpaa_bp->size, DMA_FROM_DEVICE);
+
+ /* prefetch the first 64 bytes of the frame or the SGT start */
+ prefetch(phys_to_virt(addr) + qm_fd_get_offset(fd));
+
+ fd_format = qm_fd_get_format(fd);
+ /* The only FD types that we may receive are contig and S/G */
+ WARN_ON((fd_format != qm_fd_contig) && (fd_format != qm_fd_sg));
+
+ /* Account for either the contig buffer or the SGT buffer (depending on
+ * which case we were in) having been removed from the pool.
+ */
+ count_ptr = this_cpu_ptr(dpaa_bp->percpu_count);
+ (*count_ptr)--;
+
+ if (likely(fd_format == qm_fd_contig))
+ skb = contig_fd_to_skb(priv, fd);
+ else
+ skb = sg_fd_to_skb(priv, fd);
+ if (!skb)
+ return qman_cb_dqrr_consume;
+
+ skb->protocol = eth_type_trans(skb, net_dev);
+
+ skb_len = skb->len;
+
+ if (unlikely(netif_receive_skb(skb) == NET_RX_DROP))
+ return qman_cb_dqrr_consume;
+
+ percpu_stats->rx_packets++;
+ percpu_stats->rx_bytes += skb_len;
+
+ return qman_cb_dqrr_consume;
+}
+
+static enum qman_cb_dqrr_result conf_error_dqrr(struct qman_portal *portal,
+ struct qman_fq *fq,
+ const struct qm_dqrr_entry *dq)
+{
+ struct dpaa_percpu_priv *percpu_priv;
+ struct net_device *net_dev;
+ struct dpaa_priv *priv;
+
+ net_dev = ((struct dpaa_fq *)fq)->net_dev;
+ priv = netdev_priv(net_dev);
+
+ percpu_priv = this_cpu_ptr(priv->percpu_priv);
+
+ if (dpaa_eth_napi_schedule(percpu_priv, portal))
+ return qman_cb_dqrr_stop;
+
+ dpaa_tx_error(net_dev, priv, percpu_priv, &dq->fd, fq->fqid);
+
+ return qman_cb_dqrr_consume;
+}
+
+static enum qman_cb_dqrr_result conf_dflt_dqrr(struct qman_portal *portal,
+ struct qman_fq *fq,
+ const struct qm_dqrr_entry *dq)
+{
+ struct dpaa_percpu_priv *percpu_priv;
+ struct net_device *net_dev;
+ struct dpaa_priv *priv;
+
+ net_dev = ((struct dpaa_fq *)fq)->net_dev;
+ priv = netdev_priv(net_dev);
+
+ percpu_priv = this_cpu_ptr(priv->percpu_priv);
+
+ if (dpaa_eth_napi_schedule(percpu_priv, portal))
+ return qman_cb_dqrr_stop;
+
+ dpaa_tx_conf(net_dev, priv, percpu_priv, &dq->fd, fq->fqid);
+
+ return qman_cb_dqrr_consume;
+}
+
+static void egress_ern(struct qman_portal *portal,
+ struct qman_fq *fq,
+ const union qm_mr_entry *msg)
+{
+ const struct qm_fd *fd = &msg->ern.fd;
+ struct dpaa_percpu_priv *percpu_priv;
+ const struct dpaa_priv *priv;
+ struct net_device *net_dev;
+ struct sk_buff *skb;
+
+ net_dev = ((struct dpaa_fq *)fq)->net_dev;
+ priv = netdev_priv(net_dev);
+ percpu_priv = this_cpu_ptr(priv->percpu_priv);
+
+ percpu_priv->stats.tx_dropped++;
+ percpu_priv->stats.tx_fifo_errors++;
+
+ skb = dpaa_cleanup_tx_fd(priv, fd);
+ dev_kfree_skb_any(skb);
+}
+
+static const struct dpaa_fq_cbs dpaa_fq_cbs = {
+ .rx_defq = { .cb = { .dqrr = rx_default_dqrr } },
+ .tx_defq = { .cb = { .dqrr = conf_dflt_dqrr } },
+ .rx_errq = { .cb = { .dqrr = rx_error_dqrr } },
+ .tx_errq = { .cb = { .dqrr = conf_error_dqrr } },
+ .egress_ern = { .cb = { .ern = egress_ern } }
+};
+
+static void dpaa_eth_napi_enable(struct dpaa_priv *priv)
+{
+ struct dpaa_percpu_priv *percpu_priv;
+ int i;
+
+ for_each_possible_cpu(i) {
+ percpu_priv = per_cpu_ptr(priv->percpu_priv, i);
+
+ percpu_priv->np.down = 0;
+ napi_enable(&percpu_priv->np.napi);
+ }
+}
+
+static void dpaa_eth_napi_disable(struct dpaa_priv *priv)
+{
+ struct dpaa_percpu_priv *percpu_priv;
+ int i;
+
+ for_each_possible_cpu(i) {
+ percpu_priv = per_cpu_ptr(priv->percpu_priv, i);
+
+ percpu_priv->np.down = 1;
+ napi_disable(&percpu_priv->np.napi);
+ }
+}
+
+static int dpaa_open(struct net_device *net_dev)
+{
+ struct mac_device *mac_dev;
+ struct dpaa_priv *priv;
+ int err, i;
+
+ priv = netdev_priv(net_dev);
+ mac_dev = priv->mac_dev;
+ dpaa_eth_napi_enable(priv);
+
+ net_dev->phydev = mac_dev->init_phy(net_dev, priv->mac_dev);
+ if (!net_dev->phydev) {
+ netif_err(priv, ifup, net_dev, "init_phy() failed\n");
+ return -ENODEV;
+ }
+
+ for (i = 0; i < ARRAY_SIZE(mac_dev->port); i++) {
+ err = fman_port_enable(mac_dev->port[i]);
+ if (err)
+ goto mac_start_failed;
+ }
+
+ err = priv->mac_dev->start(mac_dev);
+ if (err < 0) {
+ netif_err(priv, ifup, net_dev, "mac_dev->start() = %d\n", err);
+ goto mac_start_failed;
+ }
+
+ netif_tx_start_all_queues(net_dev);
+
+ return 0;
+
+mac_start_failed:
+ for (i = 0; i < ARRAY_SIZE(mac_dev->port); i++)
+ fman_port_disable(mac_dev->port[i]);
+
+ dpaa_eth_napi_disable(priv);
+
+ return err;
+}
+
+static int dpaa_eth_stop(struct net_device *net_dev)
+{
+ struct dpaa_priv *priv;
+ int err;
+
+ err = dpaa_stop(net_dev);
+
+ priv = netdev_priv(net_dev);
+ dpaa_eth_napi_disable(priv);
+
+ return err;
+}
+
+static const struct net_device_ops dpaa_ops = {
+ .ndo_open = dpaa_open,
+ .ndo_start_xmit = dpaa_start_xmit,
+ .ndo_stop = dpaa_eth_stop,
+ .ndo_tx_timeout = dpaa_tx_timeout,
+ .ndo_get_stats64 = dpaa_get_stats64,
+ .ndo_set_mac_address = dpaa_set_mac_address,
+ .ndo_validate_addr = eth_validate_addr,
+ .ndo_set_rx_mode = dpaa_set_rx_mode,
+};
+
+static int dpaa_napi_add(struct net_device *net_dev)
+{
+ struct dpaa_priv *priv = netdev_priv(net_dev);
+ struct dpaa_percpu_priv *percpu_priv;
+ int cpu;
+
+ for_each_possible_cpu(cpu) {
+ percpu_priv = per_cpu_ptr(priv->percpu_priv, cpu);
+
+ netif_napi_add(net_dev, &percpu_priv->np.napi,
+ dpaa_eth_poll, NAPI_POLL_WEIGHT);
+ }
+
+ return 0;
+}
+
+static void dpaa_napi_del(struct net_device *net_dev)
+{
+ struct dpaa_priv *priv = netdev_priv(net_dev);
+ struct dpaa_percpu_priv *percpu_priv;
+ int cpu;
+
+ for_each_possible_cpu(cpu) {
+ percpu_priv = per_cpu_ptr(priv->percpu_priv, cpu);
+
+ netif_napi_del(&percpu_priv->np.napi);
+ }
+}
+
+static inline void dpaa_bp_free_pf(const struct dpaa_bp *bp,
+ struct bm_buffer *bmb)
+{
+ dma_addr_t addr = bm_buf_addr(bmb);
+
+ dma_unmap_single(bp->dev, addr, bp->size, DMA_FROM_DEVICE);
+
+ skb_free_frag(phys_to_virt(addr));
+}
+
+/* Alloc the dpaa_bp struct and configure default values */
+static struct dpaa_bp *dpaa_bp_alloc(struct device *dev)
+{
+ struct dpaa_bp *dpaa_bp;
+
+ dpaa_bp = devm_kzalloc(dev, sizeof(*dpaa_bp), GFP_KERNEL);
+ if (!dpaa_bp)
+ return ERR_PTR(-ENOMEM);
+
+ dpaa_bp->bpid = FSL_DPAA_BPID_INV;
+ dpaa_bp->percpu_count = devm_alloc_percpu(dev, *dpaa_bp->percpu_count);
+ dpaa_bp->config_count = FSL_DPAA_ETH_MAX_BUF_COUNT;
+
+ dpaa_bp->seed_cb = dpaa_bp_seed;
+ dpaa_bp->free_buf_cb = dpaa_bp_free_pf;
+
+ return dpaa_bp;
+}
+
+/* Place all ingress FQs (Rx Default, Rx Error) in a dedicated CGR.
+ * We won't be sending congestion notifications to FMan; for now, we just use
+ * this CGR to generate enqueue rejections to FMan in order to drop the frames
+ * before they reach our ingress queues and eat up memory.
+ */
+static int dpaa_ingress_cgr_init(struct dpaa_priv *priv)
+{
+ struct qm_mcc_initcgr initcgr;
+ u32 cs_th;
+ int err;
+
+ err = qman_alloc_cgrid(&priv->ingress_cgr.cgrid);
+ if (err < 0) {
+ if (netif_msg_drv(priv))
+ pr_err("Error %d allocating CGR ID\n", err);
+ goto out_error;
+ }
+
+ /* Enable CS TD, but disable Congestion State Change Notifications. */
+ initcgr.we_mask = QM_CGR_WE_CS_THRES;
+ initcgr.cgr.cscn_en = QM_CGR_EN;
+ cs_th = DPAA_INGRESS_CS_THRESHOLD;
+ qm_cgr_cs_thres_set64(&initcgr.cgr.cs_thres, cs_th, 1);
+
+ initcgr.we_mask |= QM_CGR_WE_CSTD_EN;
+ initcgr.cgr.cstd_en = QM_CGR_EN;
+
+ /* This CGR will be associated with the SWP affined to the current CPU.
+ * However, we'll place all our ingress FQs in it.
+ */
+ err = qman_create_cgr(&priv->ingress_cgr, QMAN_CGR_FLAG_USE_INIT,
+ &initcgr);
+ if (err < 0) {
+ if (netif_msg_drv(priv))
+ pr_err("Error %d creating ingress CGR with ID %d\n",
+ err, priv->ingress_cgr.cgrid);
+ qman_release_cgrid(priv->ingress_cgr.cgrid);
+ goto out_error;
+ }
+ if (netif_msg_drv(priv))
+ pr_debug("Created ingress CGR %d for netdev with hwaddr %pM\n",
+ priv->ingress_cgr.cgrid, priv->mac_dev->addr);
+
+ priv->use_ingress_cgr = true;
+
+out_error:
+ return err;
+}
+
+static const struct of_device_id dpaa_match[];
+
+static inline u16 dpaa_get_headroom(struct dpaa_buffer_layout *bl)
+{
+ u16 headroom;
+
+ /* The frame headroom must accommodate:
+ * - the driver private data area
+ * - parse results, hash results, timestamp if selected
+ * If either hash results or time stamp are selected, both will
+ * be copied to/from the frame headroom, as TS is located between PR and
+ * HR in the IC and IC copy size has a granularity of 16bytes
+ * (see description of FMBM_RICP and FMBM_TICP registers in DPAARM)
+ *
+ * Also make sure the headroom is a multiple of data_align bytes
+ */
+ headroom = (u16)(bl->priv_data_size + DPAA_PARSE_RESULTS_SIZE +
+ DPAA_TIME_STAMP_SIZE + DPAA_HASH_RESULTS_SIZE);
+
+ return DPAA_FD_DATA_ALIGNMENT ? ALIGN(headroom,
+ DPAA_FD_DATA_ALIGNMENT) :
+ headroom;
+}
+
+static int dpaa_eth_probe(struct platform_device *pdev)
+{
+ struct dpaa_bp *dpaa_bps[DPAA_BPS_NUM] = {NULL};
+ struct dpaa_percpu_priv *percpu_priv;
+ struct net_device *net_dev = NULL;
+ struct dpaa_fq *dpaa_fq, *tmp;
+ struct dpaa_priv *priv = NULL;
+ struct fm_port_fqs port_fqs;
+ struct mac_device *mac_dev;
+ int err = 0, i, channel;
+ struct device *dev;
+
+ dev = &pdev->dev;
+
+ /* Allocate this early, so we can store relevant information in
+ * the private area
+ */
+ net_dev = alloc_etherdev_mq(sizeof(*priv), DPAA_ETH_TXQ_NUM);
+ if (!net_dev) {
+ dev_err(dev, "alloc_etherdev_mq() failed\n");
+ goto alloc_etherdev_mq_failed;
+ }
+
+ /* Do this here, so we can be verbose early */
+ SET_NETDEV_DEV(net_dev, dev);
+ dev_set_drvdata(dev, net_dev);
+
+ priv = netdev_priv(net_dev);
+ priv->net_dev = net_dev;
+
+ priv->msg_enable = netif_msg_init(debug, DPAA_MSG_DEFAULT);
+
+ mac_dev = dpaa_mac_dev_get(pdev);
+ if (IS_ERR(mac_dev)) {
+ dev_err(dev, "dpaa_mac_dev_get() failed\n");
+ err = PTR_ERR(mac_dev);
+ goto mac_probe_failed;
+ }
+
+ /* If fsl_fm_max_frm is set to a higher value than the all-common 1500,
+ * we choose conservatively and let the user explicitly set a higher
+ * MTU via ifconfig. Otherwise, the user may end up with different MTUs
+ * in the same LAN.
+ * If on the other hand fsl_fm_max_frm has been chosen below 1500,
+ * start with the maximum allowed.
+ */
+ net_dev->mtu = min(dpaa_get_max_mtu(), ETH_DATA_LEN);
+
+ netdev_dbg(net_dev, "Setting initial MTU on net device: %d\n",
+ net_dev->mtu);
+
+ priv->buf_layout[RX].priv_data_size = DPAA_RX_PRIV_DATA_SIZE; /* Rx */
+ priv->buf_layout[TX].priv_data_size = DPAA_TX_PRIV_DATA_SIZE; /* Tx */
+
+ /* device used for DMA mapping */
+ arch_setup_dma_ops(dev, 0, 0, NULL, false);
+ err = dma_coerce_mask_and_coherent(dev, DMA_BIT_MASK(40));
+ if (err) {
+ dev_err(dev, "dma_coerce_mask_and_coherent() failed\n");
+ goto dev_mask_failed;
+ }
+
+ /* bp init */
+ for (i = 0; i < DPAA_BPS_NUM; i++) {
+ int err;
+
+ dpaa_bps[i] = dpaa_bp_alloc(dev);
+ if (IS_ERR(dpaa_bps[i]))
+ return PTR_ERR(dpaa_bps[i]);
+ /* the raw size of the buffers used for reception */
+ dpaa_bps[i]->raw_size = bpool_buffer_raw_size(i, DPAA_BPS_NUM);
+ /* avoid runtime computations by keeping the usable size here */
+ dpaa_bps[i]->size = dpaa_bp_size(dpaa_bps[i]->raw_size);
+ dpaa_bps[i]->dev = dev;
+
+ err = dpaa_bp_alloc_pool(dpaa_bps[i]);
+ if (err < 0) {
+ dpaa_bps_free(priv);
+ priv->dpaa_bps[i] = NULL;
+ goto bp_create_failed;
+ }
+ priv->dpaa_bps[i] = dpaa_bps[i];
+ }
+
+ INIT_LIST_HEAD(&priv->dpaa_fq_list);
+
+ memset(&port_fqs, 0, sizeof(port_fqs));
+
+ err = dpaa_alloc_all_fqs(dev, &priv->dpaa_fq_list, &port_fqs);
+ if (err < 0) {
+ dev_err(dev, "dpaa_alloc_all_fqs() failed\n");
+ goto fq_probe_failed;
+ }
+
+ priv->mac_dev = mac_dev;
+
+ channel = dpaa_get_channel();
+ if (channel < 0) {
+ dev_err(dev, "dpaa_get_channel() failed\n");
+ err = channel;
+ goto get_channel_failed;
+ }
+
+ priv->channel = (u16)channel;
+
+ /* Start a thread that will walk the CPUs with affine portals
+ * and add this pool channel to each's dequeue mask.
+ */
+ dpaa_eth_add_channel(priv->channel);
+
+ dpaa_fq_setup(priv, &dpaa_fq_cbs, priv->mac_dev->port[TX]);
+
+ /* Create a congestion group for this netdev, with
+ * dynamically-allocated CGR ID.
+ * Must be executed after probing the MAC, but before
+ * assigning the egress FQs to the CGRs.
+ */
+ err = dpaa_eth_cgr_init(priv);
+ if (err < 0) {
+ dev_err(dev, "Error initializing CGR\n");
+ goto tx_cgr_init_failed;
+ }
+
+ err = dpaa_ingress_cgr_init(priv);
+ if (err < 0) {
+ dev_err(dev, "Error initializing ingress CGR\n");
+ goto rx_cgr_init_failed;
+ }
+
+ /* Add the FQs to the interface, and make them active */
+ list_for_each_entry_safe(dpaa_fq, tmp, &priv->dpaa_fq_list, list) {
+ err = dpaa_fq_init(dpaa_fq, false);
+ if (err < 0)
+ goto fq_alloc_failed;
+ }
+
+ priv->tx_headroom = dpaa_get_headroom(&priv->buf_layout[TX]);
+ priv->rx_headroom = dpaa_get_headroom(&priv->buf_layout[RX]);
+
+ /* All real interfaces need their ports initialized */
+ dpaa_eth_init_ports(mac_dev, dpaa_bps, DPAA_BPS_NUM, &port_fqs,
+ &priv->buf_layout[0], dev);
+
+ priv->percpu_priv = devm_alloc_percpu(dev, *priv->percpu_priv);
+ if (!priv->percpu_priv) {
+ dev_err(dev, "devm_alloc_percpu() failed\n");
+ err = -ENOMEM;
+ goto alloc_percpu_failed;
+ }
+ for_each_possible_cpu(i) {
+ percpu_priv = per_cpu_ptr(priv->percpu_priv, i);
+ memset(percpu_priv, 0, sizeof(*percpu_priv));
+ }
+
+ /* Initialize NAPI */
+ err = dpaa_napi_add(net_dev);
+ if (err < 0)
+ goto napi_add_failed;
+
+ err = dpaa_netdev_init(net_dev, &dpaa_ops, tx_timeout);
+ if (err < 0)
+ goto netdev_init_failed;
+
+ netif_info(priv, probe, net_dev, "Probed interface %s\n",
+ net_dev->name);
+
+ return 0;
+
+netdev_init_failed:
+napi_add_failed:
+ dpaa_napi_del(net_dev);
+alloc_percpu_failed:
+ dpaa_fq_free(dev, &priv->dpaa_fq_list);
+fq_alloc_failed:
+ qman_delete_cgr_safe(&priv->ingress_cgr);
+ qman_release_cgrid(priv->ingress_cgr.cgrid);
+rx_cgr_init_failed:
+ qman_delete_cgr_safe(&priv->cgr_data.cgr);
+ qman_release_cgrid(priv->cgr_data.cgr.cgrid);
+tx_cgr_init_failed:
+get_channel_failed:
+ dpaa_bps_free(priv);
+bp_create_failed:
+fq_probe_failed:
+dev_mask_failed:
+mac_probe_failed:
+ dev_set_drvdata(dev, NULL);
+ free_netdev(net_dev);
+alloc_etherdev_mq_failed:
+ for (i = 0; i < DPAA_BPS_NUM && dpaa_bps[i]; i++) {
+ if (atomic_read(&dpaa_bps[i]->refs) == 0)
+ devm_kfree(dev, dpaa_bps[i]);
+ }
+ return err;
+}
+
+static int dpaa_remove(struct platform_device *pdev)
+{
+ struct net_device *net_dev;
+ struct dpaa_priv *priv;
+ struct device *dev;
+ int err;
+
+ dev = &pdev->dev;
+ net_dev = dev_get_drvdata(dev);
+
+ priv = netdev_priv(net_dev);
+
+ dev_set_drvdata(dev, NULL);
+ unregister_netdev(net_dev);
+
+ err = dpaa_fq_free(dev, &priv->dpaa_fq_list);
+
+ qman_delete_cgr_safe(&priv->ingress_cgr);
+ qman_release_cgrid(priv->ingress_cgr.cgrid);
+ qman_delete_cgr_safe(&priv->cgr_data.cgr);
+ qman_release_cgrid(priv->cgr_data.cgr.cgrid);
+
+ dpaa_napi_del(net_dev);
+
+ dpaa_bps_free(priv);
+
+ free_netdev(net_dev);
+
+ return err;
+}
+
+static struct platform_device_id dpaa_devtype[] = {
+ {
+ .name = "dpaa-ethernet",
+ .driver_data = 0,
+ }, {
+ }
+};
+MODULE_DEVICE_TABLE(platform, dpaa_devtype);
+
+static struct platform_driver dpaa_driver = {
+ .driver = {
+ .name = KBUILD_MODNAME,
+ },
+ .id_table = dpaa_devtype,
+ .probe = dpaa_eth_probe,
+ .remove = dpaa_remove
+};
+
+static int __init dpaa_load(void)
+{
+ int err;
+
+ pr_debug("FSL DPAA Ethernet driver\n");
+
+ /* initialize dpaa_eth mirror values */
+ dpaa_rx_extra_headroom = fman_get_rx_extra_headroom();
+ dpaa_max_frm = fman_get_max_frm();
+
+ err = platform_driver_register(&dpaa_driver);
+ if (err < 0)
+ pr_err("Error, platform_driver_register() = %d\n", err);
+
+ return err;
+}
+module_init(dpaa_load);
+
+static void __exit dpaa_unload(void)
+{
+ platform_driver_unregister(&dpaa_driver);
+
+ /* Only one channel is used and needs to be released after all
+ * interfaces are removed
+ */
+ dpaa_release_channel();
+}
+module_exit(dpaa_unload);
+
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_DESCRIPTION("FSL DPAA Ethernet driver");
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
new file mode 100644
index 0000000..fe98e08
--- /dev/null
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
@@ -0,0 +1,144 @@
+/* Copyright 2008 - 2016 Freescale Semiconductor Inc.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * * Neither the name of Freescale Semiconductor nor the
+ * names of its contributors may be used to endorse or promote products
+ * derived from this software without specific prior written permission.
+ *
+ * ALTERNATIVELY, this software may be distributed under the terms of the
+ * GNU General Public License ("GPL") as published by the Free Software
+ * Foundation, either version 2 of that License or (at your option) any
+ * later version.
+ *
+ * THIS SOFTWARE IS PROVIDED BY Freescale Semiconductor ``AS IS'' AND ANY
+ * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ * DISCLAIMED. IN NO EVENT SHALL Freescale Semiconductor BE LIABLE FOR ANY
+ * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+ * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef __DPAA_H
+#define __DPAA_H
+
+#include <linux/netdevice.h>
+#include <soc/fsl/qman.h>
+#include <soc/fsl/bman.h>
+
+#include "fman.h"
+#include "mac.h"
+
+#define DPAA_ETH_TXQ_NUM NR_CPUS
+
+#define DPAA_BPS_NUM 3 /* number of bpools per interface */
+
+/* More detailed FQ types - used for fine-grained WQ assignments */
+enum dpaa_fq_type {
+ FQ_TYPE_RX_DEFAULT = 1, /* Rx Default FQs */
+ FQ_TYPE_RX_ERROR, /* Rx Error FQs */
+ FQ_TYPE_TX, /* "Real" Tx FQs */
+ FQ_TYPE_TX_CONFIRM, /* Tx default Conf FQ (actually an Rx FQ) */
+ FQ_TYPE_TX_CONF_MQ, /* Tx conf FQs (one for each Tx FQ) */
+ FQ_TYPE_TX_ERROR, /* Tx Error FQs (these are actually Rx FQs) */
+};
+
+struct dpaa_fq {
+ struct qman_fq fq_base;
+ struct list_head list;
+ struct net_device *net_dev;
+ bool init;
+ u32 fqid;
+ u32 flags;
+ u16 channel;
+ u8 wq;
+ enum dpaa_fq_type fq_type;
+};
+
+struct dpaa_fq_cbs {
+ struct qman_fq rx_defq;
+ struct qman_fq tx_defq;
+ struct qman_fq rx_errq;
+ struct qman_fq tx_errq;
+ struct qman_fq egress_ern;
+};
+
+struct dpaa_bp {
+ /* device used in the DMA mapping operations */
+ struct device *dev;
+ /* current number of buffers in the buffer pool alloted to each CPU */
+ int __percpu *percpu_count;
+ /* all buffers allocated for this pool have this raw size */
+ size_t raw_size;
+ /* all buffers in this pool have this same usable size */
+ size_t size;
+ /* the buffer pools are initialized with config_count buffers for each
+ * CPU; at runtime the number of buffers per CPU is constantly brought
+ * back to this level
+ */
+ u16 config_count;
+ u8 bpid;
+ struct bman_pool *pool;
+ /* bpool can be seeded before use by this cb */
+ int (*seed_cb)(struct dpaa_bp *);
+ /* bpool can be emptied before freeing by this cb */
+ void (*free_buf_cb)(const struct dpaa_bp *, struct bm_buffer *);
+ atomic_t refs;
+};
+
+struct dpaa_napi_portal {
+ struct napi_struct napi;
+ struct qman_portal *p;
+ bool down;
+};
+
+struct dpaa_percpu_priv {
+ struct net_device *net_dev;
+ struct dpaa_napi_portal np;
+ struct rtnl_link_stats64 stats;
+};
+
+struct dpaa_buffer_layout {
+ u16 priv_data_size;
+};
+
+struct dpaa_priv {
+ struct dpaa_percpu_priv __percpu *percpu_priv;
+ struct dpaa_bp *dpaa_bps[DPAA_BPS_NUM];
+ /* Store here the needed Tx headroom for convenience and speed
+ * (even though it can be computed based on the fields of buf_layout)
+ */
+ u16 tx_headroom;
+ struct net_device *net_dev;
+ struct mac_device *mac_dev;
+ struct qman_fq *egress_fqs[DPAA_ETH_TXQ_NUM];
+ struct qman_fq *conf_fqs[DPAA_ETH_TXQ_NUM];
+
+ u16 channel;
+ struct list_head dpaa_fq_list;
+
+ u32 msg_enable; /* net_device message level */
+
+ struct {
+ /* All egress queues to a given net device belong to one
+ * (and the same) congestion group.
+ */
+ struct qman_cgr cgr;
+ } cgr_data;
+ /* Use a per-port CGR for ingress traffic. */
+ bool use_ingress_cgr;
+ struct qman_cgr ingress_cgr;
+
+ struct dpaa_buffer_layout buf_layout[2];
+ u16 rx_headroom;
+};
+#endif /* __DPAA_H */
--
2.1.0
^ permalink raw reply related
* [PATCH net-next v8 1/9] devres: add devm_alloc_percpu()
From: Madalin Bucur @ 2016-11-15 8:41 UTC (permalink / raw)
To: netdev
Cc: linuxppc-dev, linux-kernel, davem, oss, ppc, joe, pebolle,
joakim.tjernlund
In-Reply-To: <1479199269-9748-1-git-send-email-madalin.bucur@nxp.com>
Introduce managed counterparts for alloc_percpu() and free_percpu().
Add devm_alloc_percpu() and devm_free_percpu() into the managed
interfaces list.
Signed-off-by: Madalin Bucur <madalin.bucur@nxp.com>
---
Documentation/driver-model/devres.txt | 4 +++
drivers/base/devres.c | 66 +++++++++++++++++++++++++++++++++++
include/linux/device.h | 19 ++++++++++
3 files changed, 89 insertions(+)
diff --git a/Documentation/driver-model/devres.txt b/Documentation/driver-model/devres.txt
index 1670708..ca9d1eb 100644
--- a/Documentation/driver-model/devres.txt
+++ b/Documentation/driver-model/devres.txt
@@ -332,6 +332,10 @@ MEM
MFD
devm_mfd_add_devices()
+PER-CPU MEM
+ devm_alloc_percpu()
+ devm_free_percpu()
+
PCI
pcim_enable_device() : after success, all PCI ops become managed
pcim_pin_device() : keep PCI device enabled after release
diff --git a/drivers/base/devres.c b/drivers/base/devres.c
index 8fc654f..71d5770 100644
--- a/drivers/base/devres.c
+++ b/drivers/base/devres.c
@@ -10,6 +10,7 @@
#include <linux/device.h>
#include <linux/module.h>
#include <linux/slab.h>
+#include <linux/percpu.h>
#include "base.h"
@@ -985,3 +986,68 @@ void devm_free_pages(struct device *dev, unsigned long addr)
&devres));
}
EXPORT_SYMBOL_GPL(devm_free_pages);
+
+static void devm_percpu_release(struct device *dev, void *pdata)
+{
+ void __percpu *p;
+
+ p = *(void __percpu **)pdata;
+ free_percpu(p);
+}
+
+static int devm_percpu_match(struct device *dev, void *data, void *p)
+{
+ struct devres *devr = container_of(data, struct devres, data);
+
+ return *(void **)devr->data == p;
+}
+
+/**
+ * __devm_alloc_percpu - Resource-managed alloc_percpu
+ * @dev: Device to allocate per-cpu memory for
+ * @size: Size of per-cpu memory to allocate
+ * @align: Alignment of per-cpu memory to allocate
+ *
+ * Managed alloc_percpu. Per-cpu memory allocated with this function is
+ * automatically freed on driver detach.
+ *
+ * RETURNS:
+ * Pointer to allocated memory on success, NULL on failure.
+ */
+void __percpu *__devm_alloc_percpu(struct device *dev, size_t size,
+ size_t align)
+{
+ void *p;
+ void __percpu *pcpu;
+
+ pcpu = __alloc_percpu(size, align);
+ if (!pcpu)
+ return NULL;
+
+ p = devres_alloc(devm_percpu_release, sizeof(void *), GFP_KERNEL);
+ if (!p) {
+ free_percpu(pcpu);
+ return NULL;
+ }
+
+ *(void __percpu **)p = pcpu;
+
+ devres_add(dev, p);
+
+ return pcpu;
+}
+EXPORT_SYMBOL_GPL(__devm_alloc_percpu);
+
+/**
+ * devm_free_percpu - Resource-managed free_percpu
+ * @dev: Device this memory belongs to
+ * @pdata: Per-cpu memory to free
+ *
+ * Free memory allocated with devm_alloc_percpu().
+ */
+void devm_free_percpu(struct device *dev, void __percpu *pdata)
+{
+ WARN_ON(devres_destroy(dev, devm_percpu_release, devm_percpu_match,
+ (void *)pdata));
+}
+EXPORT_SYMBOL_GPL(devm_free_percpu);
diff --git a/include/linux/device.h b/include/linux/device.h
index bc41e87..a00105c 100644
--- a/include/linux/device.h
+++ b/include/linux/device.h
@@ -698,6 +698,25 @@ static inline int devm_add_action_or_reset(struct device *dev,
return ret;
}
+/**
+ * devm_alloc_percpu - Resource-managed alloc_percpu
+ * @dev: Device to allocate per-cpu memory for
+ * @type: Type to allocate per-cpu memory for
+ *
+ * Managed alloc_percpu. Per-cpu memory allocated with this function is
+ * automatically freed on driver detach.
+ *
+ * RETURNS:
+ * Pointer to allocated memory on success, NULL on failure.
+ */
+#define devm_alloc_percpu(dev, type) \
+ ((typeof(type) __percpu *)__devm_alloc_percpu((dev), sizeof(type), \
+ __alignof__(type)))
+
+void __percpu *__devm_alloc_percpu(struct device *dev, size_t size,
+ size_t align);
+void devm_free_percpu(struct device *dev, void __percpu *pdata);
+
struct device_dma_parameters {
/*
* a low level driver may set these to teach IOMMU code about
--
2.1.0
^ permalink raw reply related
* [PATCH net-next v8 0/9] dpaa_eth: Add the QorIQ DPAA Ethernet driver
From: Madalin Bucur @ 2016-11-15 8:41 UTC (permalink / raw)
To: netdev
Cc: linuxppc-dev, linux-kernel, davem, oss, ppc, joe, pebolle,
joakim.tjernlund
This patch series adds the Ethernet driver for the Freescale
QorIQ Data Path Acceleration Architecture (DPAA).
This version includes changes following the feedback received
on previous versions from Eric Dumazet, Bob Cochran, Joe Perches,
Paul Bolle, Joakim Tjernlund, Scott Wood, David Miller - thank you.
Together with the driver a managed version of alloc_percpu
is provided that simplifies the release of per-CPU memory.
The Freescale DPAA architecture consists in a series of hardware
blocks that support the Ethernet connectivity. The Ethernet driver
depends upon the following drivers that are currently in the Linux
kernel:
- Peripheral Access Memory Unit (PAMU)
drivers/iommu/fsl_*
- Frame Manager (FMan) added in v4.4
drivers/net/ethernet/freescale/fman
- Queue Manager (QMan), Buffer Manager (BMan) added in v4.9-rc1
drivers/soc/fsl/qbman
dpaa_eth interfaces mapping to FMan MACs:
dpaa_eth /eth0\ ... /ethN\
driver | | | |
------------- ---- ----------- ---- -------------
-Ports / Tx Rx \ ... / Tx Rx \
FMan | | | |
-MACs | MAC0 | | MACN |
/ dtsec0 \ ... / dtsecN \ (or tgec)
/ \ / \(or memac)
--------- -------------- --- -------------- ---------
FMan, FMan Port, FMan SP, FMan MURAM drivers
---------------------------------------------------------
FMan HW blocks: MURAM, MACs, Ports, SP
---------------------------------------------------------
dpaa_eth relation to QMan, FMan:
________________________________
dpaa_eth / eth0 \
driver / \
--------- -^- -^- -^- --- ---------
QMan driver / \ / \ / \ \ / | BMan |
|Rx | |Rx | |Tx | |Tx | | driver |
--------- |Dfl| |Err| |Cnf| |FQs| | |
QMan HW |FQ | |FQ | |FQ | | | | |
/ \ / \ / \ \ / | |
--------- --- --- --- -v- ---------
| FMan QMI | |
| FMan HW FMan BMI | BMan HW |
----------------------- --------
where the acronyms used above (and in the code) are:
DPAA = Data Path Acceleration Architecture
FMan = DPAA Frame Manager
QMan = DPAA Queue Manager
BMan = DPAA Buffers Manager
QMI = QMan interface in FMan
BMI = BMan interface in FMan
FMan SP = FMan Storage Profiles
MURAM = Multi-user RAM in FMan
FQ = QMan Frame Queue
Rx Dfl FQ = default reception FQ
Rx Err FQ = Rx error frames FQ
Tx Cnf FQ = Tx confirmation FQ
Tx FQs = transmission frame queues
dtsec = datapath three speed Ethernet controller (10/100/1000 Mbps)
tgec = ten gigabit Ethernet controller (10 Gbps)
memac = multirate Ethernet MAC (10/100/1000/10000)
Changes from v7:
- remove the debug option to use a common buffer pool for all the
interfaces
Changed from v6:
- fixed an issue on an error path in dpaa_set_mac_address()
- removed NDO operation definitions that were not needed
- sorted the local variable declarations
- cleaned up a few checkpatch checks
- removed friendly network interface naming code
Changes from v5:
- adapt to the latest Q/BMan drivers API
- use build_skb() on Rx path instead of buffer pool refill path
- proper support for multiple buffer pools
- align function, variable names, code cleanup
- driver file structure cleanup
Changes from v4:
- addressed feedback from Scott Wood and Joe Perches
- fixed spelling
- fixed leak of uninitialized stack to userspace
- fix prints
- replace raw_cpu_ptr() with this_cpu_ptr()
- remove _s from the end of structure names
- remove underscores at start of functions, goto labels
- remove likely in error paths
- use container_of() instead of open casts
- remove priv from the driver name
- move return type on same line with function name
- drop DPA_READ_SKB_PTR/DPA_WRITE_SKB_PTR
Changes from v3:
- removed bogus delay and comment in .ndo_stop implementation
- addressed minor issues reported by David Miller
Changes from v2:
- removed debugfs, moved exports to ethtool statistics
- removed congestion groups Kconfig params
Changes from v1:
- bpool level Kconfig options removed
- print format using pr_fmt, cleaned up prints
- __hot/__cold removed
- gratuitous unlikely() removed
- code style aligned, consistent spacing for declarations
- comment formatting
The changes are also available in the public git repository at
git://git.freescale.com/ppc/upstream/linux.git on the branch dpaa_eth-next.
Madalin Bucur (9):
devres: add devm_alloc_percpu()
dpaa_eth: add support for DPAA Ethernet
dpaa_eth: add ethtool functionality
dpaa_eth: add ethtool statistics
dpaa_eth: add sysfs exports
dpaa_eth: add trace points
arch/powerpc: Enable FSL_PAMU
arch/powerpc: Enable FSL_FMAN
arch/powerpc: Enable dpaa_eth
Documentation/driver-model/devres.txt | 4 +
arch/powerpc/configs/dpaa.config | 3 +
drivers/base/devres.c | 66 +
drivers/net/ethernet/freescale/Kconfig | 2 +
drivers/net/ethernet/freescale/Makefile | 1 +
drivers/net/ethernet/freescale/dpaa/Kconfig | 10 +
drivers/net/ethernet/freescale/dpaa/Makefile | 12 +
drivers/net/ethernet/freescale/dpaa/dpaa_eth.c | 2753 ++++++++++++++++++++
drivers/net/ethernet/freescale/dpaa/dpaa_eth.h | 185 ++
.../net/ethernet/freescale/dpaa/dpaa_eth_sysfs.c | 165 ++
.../net/ethernet/freescale/dpaa/dpaa_eth_trace.h | 141 +
drivers/net/ethernet/freescale/dpaa/dpaa_ethtool.c | 417 +++
include/linux/device.h | 19 +
13 files changed, 3778 insertions(+)
create mode 100644 drivers/net/ethernet/freescale/dpaa/Kconfig
create mode 100644 drivers/net/ethernet/freescale/dpaa/Makefile
create mode 100644 drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
create mode 100644 drivers/net/ethernet/freescale/dpaa/dpaa_eth.h
create mode 100644 drivers/net/ethernet/freescale/dpaa/dpaa_eth_sysfs.c
create mode 100644 drivers/net/ethernet/freescale/dpaa/dpaa_eth_trace.h
create mode 100644 drivers/net/ethernet/freescale/dpaa/dpaa_ethtool.c
--
2.1.0
^ permalink raw reply
* Re: [PATCH 1/3] tuntap: rx batching
From: Jason Wang @ 2016-11-15 8:08 UTC (permalink / raw)
To: Michael S. Tsirkin; +Cc: John Fastabend, netdev, linux-kernel
In-Reply-To: <20161115053435-mutt-send-email-mst@kernel.org>
On 2016年11月15日 11:41, Michael S. Tsirkin wrote:
> On Tue, Nov 15, 2016 at 11:14:48AM +0800, Jason Wang wrote:
>> >
>> >
>> >On 2016年11月12日 00:20, Michael S. Tsirkin wrote:
>>> > >On Fri, Nov 11, 2016 at 12:28:38PM +0800, Jason Wang wrote:
>>>> > > >
>>>> > > >On 2016年11月11日 12:17, John Fastabend wrote:
>>>>> > > > >On 16-11-10 07:31 PM, Michael S. Tsirkin wrote:
>>>>>>> > > > > > >On Fri, Nov 11, 2016 at 10:07:44AM +0800, Jason Wang wrote:
>>>>>>>>> > > > > > > > >
>>>>>>>>> > > > > > > > >On 2016年11月10日 00:38, Michael S. Tsirkin wrote:
>>>>>>>>>>> > > > > > > > > > >On Wed, Nov 09, 2016 at 03:38:31PM +0800, Jason Wang wrote:
>>>>>>>>>>>>> > > > > > > > > > > > >Backlog were used for tuntap rx, but it can only process 1 packet at
>>>>>>>>>>>>> > > > > > > > > > > > >one time since it was scheduled during sendmsg() synchronously in
>>>>>>>>>>>>> > > > > > > > > > > > >process context. This lead bad cache utilization so this patch tries
>>>>>>>>>>>>> > > > > > > > > > > > >to do some batching before call rx NAPI. This is done through:
>>>>>>>>>>>>> > > > > > > > > > > > >
>>>>>>>>>>>>> > > > > > > > > > > > >- accept MSG_MORE as a hint from sendmsg() caller, if it was set,
>>>>>>>>>>>>> > > > > > > > > > > > > batch the packet temporarily in a linked list and submit them all
>>>>>>>>>>>>> > > > > > > > > > > > > once MSG_MORE were cleared.
>>>>>>>>>>>>> > > > > > > > > > > > >- implement a tuntap specific NAPI handler for processing this kind of
>>>>>>>>>>>>> > > > > > > > > > > > > possible batching. (This could be done by extending backlog to
>>>>>>>>>>>>> > > > > > > > > > > > > support skb like, but using a tun specific one looks cleaner and
>>>>>>>>>>>>> > > > > > > > > > > > > easier for future extension).
>>>>>>>>>>>>> > > > > > > > > > > > >
>>>>>>>>>>>>> > > > > > > > > > > > >Signed-off-by: Jason Wang<jasowang@redhat.com>
>>>>>>>>>>> > > > > > > > > > >So why do we need an extra queue?
>>>>>>>>> > > > > > > > >The idea was borrowed from backlog to allow some kind of bulking and avoid
>>>>>>>>> > > > > > > > >spinlock on each dequeuing.
>>>>>>>>> > > > > > > > >
>>>>>>>>>>> > > > > > > > > > > This is not what hardware devices do.
>>>>>>>>>>> > > > > > > > > > >How about adding the packet to queue unconditionally, deferring
>>>>>>>>>>> > > > > > > > > > >signalling until we get sendmsg without MSG_MORE?
>>>>>>>>> > > > > > > > >Then you need touch spinlock when dequeuing each packet.
>>>>> > > > >Random thought, I have a cmpxchg ring I am using for the qdisc work that
>>>>> > > > >could possibly replace the spinlock implementation. I haven't figured
>>>>> > > > >out the resizing API yet because I did not need it but I assume it could
>>>>> > > > >help here and let you dequeue multiple skbs in one operation.
>>>>> > > > >
>>>>> > > > >I can post the latest version if useful or an older version is
>>>>> > > > >somewhere on patchworks as well.
>>>>> > > > >
>>>>> > > > >.John
>>>>> > > > >
>>>>> > > > >
>>>> > > >Look useful here, and I can compare the performance if you post.
>>>> > > >
>>>> > > >A question is can we extend the skb_array to support that?
>>>> > > >
>>>> > > >Thanks
>>> > >I'd like to start with simple patch adding napi with one queue, then add
>>> > >optimization patches on top.
>> >
>> >The point is tun is using backlog who uses two queues (process_queue and
>> >input_pkt_queue).
>> >
>> >How about something like:
>> >
>> >1) NAPI support with skb_array
> I would start with just write queue linked list. It all runs on a single
> CPU normally,
True for virt but I'm not sure the others. If we have multiple senders
at the same time, current code scales very well.
> so the nice reductions of cache line bounces due to skb
> array should never materialize.
>
> While we are at it, limiting the size of the queue might
> be a good idea. Kind of like TUNSETSNDBUF but 1. actually
> working where instead of tracking packets within net stack
> we make sndbuf track the internal buffer
Get your point, will start from simple skb list.
Thanks
^ permalink raw reply
* [PATCH net] ipv6 addrconf: Implemented enhanced DAD (RFC7527)
From: Erik Nordmark @ 2016-11-15 7:57 UTC (permalink / raw)
To: netdev
Implemented RFC7527 Enhanced DAD.
IPv6 duplicate address detection can fail if there is some temporary
loopback of Ethernet frames. RFC7527 solves this by including a random
nonce in the NS messages used for DAD, and if an NS is received with the
same nonce it is assumed to be a looped back DAD probe and is ignored.
RFC7527 is disabled by default. Can be enabled by setting either one of
conf/{all,interface}/ipv6_rfc7527 to non-zero.
Signed-off-by: Erik Nordmark <nordmark@arista.com>
Index: linux-stable/Documentation/networking/ip-sysctl.txt
===================================================================
--- linux-stable.orig/Documentation/networking/ip-sysctl.txt
+++ linux-stable/Documentation/networking/ip-sysctl.txt
@@ -1713,6 +1713,15 @@ drop_unsolicited_na - BOOLEAN
By default this is turned off.
+ipv6_rfc7527 - BOOLEAN
+ Include a nonce option in the IPv6 neighbor solicitation messages
used for
+ duplicate address detection per RFC7527. A received DAD NS will
only signal
+ a duplicate address if the nonce is different. This avoids any false
+ detection of duplicates due to loopback of the NS messages that we
send.
+ The nonce option will be sent on an interface if either one of
+ conf/{all,interface}/ipv6_rfc7527 are TRUE.
+ Default: FALSE
+
icmp/*:
ratelimit - INTEGER
Limit the maximal rates for sending ICMPv6 packets.
Index: linux-stable/include/linux/ipv6.h
===================================================================
--- linux-stable.orig/include/linux/ipv6.h
+++ linux-stable/include/linux/ipv6.h
@@ -63,6 +63,7 @@ struct ipv6_devconf {
} stable_secret;
__s32 use_oif_addrs_only;
__s32 keep_addr_on_down;
+ __u32 ipv6_rfc7527;
struct ctl_table_header *sysctl_header;
};
Index: linux-stable/include/net/if_inet6.h
===================================================================
--- linux-stable.orig/include/net/if_inet6.h
+++ linux-stable/include/net/if_inet6.h
@@ -55,6 +55,7 @@ struct inet6_ifaddr {
__u8 stable_privacy_retry;
__u16 scope;
+ __u64 dad_nonce;
unsigned long cstamp; /* created timestamp */
unsigned long tstamp; /* updated timestamp */
Index: linux-stable/include/net/ndisc.h
===================================================================
--- linux-stable.orig/include/net/ndisc.h
+++ linux-stable/include/net/ndisc.h
@@ -31,6 +31,7 @@ enum {
ND_OPT_PREFIX_INFO = 3, /* RFC2461 */
ND_OPT_REDIRECT_HDR = 4, /* RFC2461 */
ND_OPT_MTU = 5, /* RFC2461 */
+ ND_OPT_NONCE = 14, /* RFC7527 */
__ND_OPT_ARRAY_MAX,
ND_OPT_ROUTE_INFO = 24, /* RFC4191 */
ND_OPT_RDNSS = 25, /* RFC5006 */
@@ -121,6 +122,7 @@ struct ndisc_options {
#define nd_opts_pi_end nd_opt_array[__ND_OPT_PREFIX_INFO_END]
#define nd_opts_rh nd_opt_array[ND_OPT_REDIRECT_HDR]
#define nd_opts_mtu nd_opt_array[ND_OPT_MTU]
+#define nd_opts_nonce nd_opt_array[ND_OPT_NONCE]
#define nd_802154_opts_src_lladdr
nd_802154_opt_array[ND_OPT_SOURCE_LL_ADDR]
#define nd_802154_opts_tgt_lladdr
nd_802154_opt_array[ND_OPT_TARGET_LL_ADDR]
@@ -398,7 +400,8 @@ void ndisc_cleanup(void);
int ndisc_rcv(struct sk_buff *skb);
void ndisc_send_ns(struct net_device *dev, const struct in6_addr *solicit,
- const struct in6_addr *daddr, const struct in6_addr *saddr);
+ const struct in6_addr *daddr, const struct in6_addr *saddr,
+ u64 nonce);
void ndisc_send_rs(struct net_device *dev,
const struct in6_addr *saddr, const struct in6_addr *daddr);
Index: linux-stable/include/uapi/linux/ipv6.h
===================================================================
--- linux-stable.orig/include/uapi/linux/ipv6.h
+++ linux-stable/include/uapi/linux/ipv6.h
@@ -177,6 +177,7 @@ enum {
DEVCONF_DROP_UNICAST_IN_L2_MULTICAST,
DEVCONF_DROP_UNSOLICITED_NA,
DEVCONF_KEEP_ADDR_ON_DOWN,
+ DEVCONF_IPV6_RFC7527,
DEVCONF_MAX
};
Index: linux-stable/net/ipv6/addrconf.c
===================================================================
--- linux-stable.orig/net/ipv6/addrconf.c
+++ linux-stable/net/ipv6/addrconf.c
@@ -217,6 +217,7 @@ static struct ipv6_devconf ipv6_devconf
.use_oif_addrs_only = 0,
.ignore_routes_with_linkdown = 0,
.keep_addr_on_down = 0,
+ .ipv6_rfc7527 = 0,
};
static struct ipv6_devconf ipv6_devconf_dflt __read_mostly = {
@@ -262,6 +263,7 @@ static struct ipv6_devconf ipv6_devconf_
.use_oif_addrs_only = 0,
.ignore_routes_with_linkdown = 0,
.keep_addr_on_down = 0,
+ .ipv6_rfc7527 = 0,
};
/* Check if a valid qdisc is available */
@@ -3722,12 +3724,18 @@ static void addrconf_dad_kick(struct ine
{
unsigned long rand_num;
struct inet6_dev *idev = ifp->idev;
+ u64 nonce;
if (ifp->flags & IFA_F_OPTIMISTIC)
rand_num = 0;
else
rand_num = prandom_u32() % (idev->cnf.rtr_solicit_delay ? : 1);
+ nonce = 0;
+ if (ifp->idev->cnf.ipv6_rfc7527 ||
+ dev_net((ifp->idev)->dev)->ipv6.devconf_all->ipv6_rfc7527)
+ get_random_bytes(&nonce, 6);
+ ifp->dad_nonce = nonce;
ifp->dad_probes = idev->cnf.dad_transmits;
addrconf_mod_dad_work(ifp, rand_num);
}
@@ -3903,7 +3911,8 @@ static void addrconf_dad_work(struct wor
/* send a neighbour solicitation for our addr */
addrconf_addr_solict_mult(&ifp->addr, &mcaddr);
- ndisc_send_ns(ifp->idev->dev, &ifp->addr, &mcaddr, &in6addr_any);
+ ndisc_send_ns(ifp->idev->dev, &ifp->addr, &mcaddr, &in6addr_any,
+ ifp->dad_nonce);
out:
in6_ifa_put(ifp);
rtnl_unlock();
@@ -4937,6 +4946,7 @@ static inline void ipv6_store_devconf(st
array[DEVCONF_DROP_UNICAST_IN_L2_MULTICAST] =
cnf->drop_unicast_in_l2_multicast;
array[DEVCONF_DROP_UNSOLICITED_NA] = cnf->drop_unsolicited_na;
array[DEVCONF_KEEP_ADDR_ON_DOWN] = cnf->keep_addr_on_down;
+ array[DEVCONF_IPV6_RFC7527] = cnf->ipv6_rfc7527;
}
static inline size_t inet6_ifla6_size(void)
@@ -6027,6 +6037,13 @@ static const struct ctl_table addrconf_s
},
{
+ .procname = "ipv6_rfc7527",
+ .data = &ipv6_devconf.ipv6_rfc7527,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec,
+ },
+ {
/* sentinel */
}
};
Index: linux-stable/net/ipv6/ndisc.c
===================================================================
--- linux-stable.orig/net/ipv6/ndisc.c
+++ linux-stable/net/ipv6/ndisc.c
@@ -234,6 +234,7 @@ struct ndisc_options *ndisc_parse_option
case ND_OPT_SOURCE_LL_ADDR:
case ND_OPT_TARGET_LL_ADDR:
case ND_OPT_MTU:
+ case ND_OPT_NONCE:
case ND_OPT_REDIRECT_HDR:
if (ndopts->nd_opt_array[nd_opt->nd_opt_type]) {
ND_PRINTK(2, warn,
@@ -571,7 +572,8 @@ static void ndisc_send_unsol_na(struct n
}
void ndisc_send_ns(struct net_device *dev, const struct in6_addr *solicit,
- const struct in6_addr *daddr, const struct in6_addr *saddr)
+ const struct in6_addr *daddr, const struct in6_addr *saddr,
+ u64 nonce)
{
struct sk_buff *skb;
struct in6_addr addr_buf;
@@ -591,6 +593,8 @@ void ndisc_send_ns(struct net_device *de
if (inc_opt)
optlen += ndisc_opt_addr_space(dev,
NDISC_NEIGHBOUR_SOLICITATION);
+ if (nonce != 0)
+ optlen += 8;
skb = ndisc_alloc_skb(dev, sizeof(*msg) + optlen);
if (!skb)
@@ -608,6 +612,13 @@ void ndisc_send_ns(struct net_device *de
ndisc_fill_addr_option(skb, ND_OPT_SOURCE_LL_ADDR,
dev->dev_addr,
NDISC_NEIGHBOUR_SOLICITATION);
+ if (nonce != 0) {
+ u8 *opt = skb_put(skb, 8);
+
+ opt[0] = ND_OPT_NONCE;
+ opt[1] = 8 >> 3;
+ memcpy(opt + 2, &nonce, 6);
+ }
ndisc_send_skb(skb, daddr, saddr);
}
@@ -696,12 +707,12 @@ static void ndisc_solicit(struct neighbo
"%s: trying to ucast probe in NUD_INVALID: %pI6\n",
__func__, target);
}
- ndisc_send_ns(dev, target, target, saddr);
+ ndisc_send_ns(dev, target, target, saddr, 0);
} else if ((probes -= NEIGH_VAR(neigh->parms, APP_PROBES)) < 0) {
neigh_app_ns(neigh);
} else {
addrconf_addr_solict_mult(target, &mcaddr);
- ndisc_send_ns(dev, target, &mcaddr, saddr);
+ ndisc_send_ns(dev, target, &mcaddr, saddr, 0);
}
}
@@ -745,6 +756,7 @@ static void ndisc_recv_ns(struct sk_buff
int dad = ipv6_addr_any(saddr);
bool inc;
int is_router = -1;
+ u64 nonce;
if (skb->len < sizeof(struct nd_msg)) {
ND_PRINTK(2, warn, "NS: packet too short\n");
@@ -789,6 +801,8 @@ static void ndisc_recv_ns(struct sk_buff
return;
}
}
+ if (ndopts.nd_opts_nonce)
+ memcpy(&nonce, (u8 *)(ndopts.nd_opts_nonce + 1), 6);
inc = ipv6_addr_is_multicast(daddr);
@@ -797,6 +811,16 @@ static void ndisc_recv_ns(struct sk_buff
have_ifp:
if (ifp->flags & (IFA_F_TENTATIVE|IFA_F_OPTIMISTIC)) {
if (dad) {
+ if (nonce != 0 && ifp->dad_nonce == nonce) {
+ /* Matching nonce if looped back */
+ if (net_ratelimit())
+ ND_PRINTK(2, notice,
+ "%s: IPv6 DAD loopback for address %pI6c
nonce %llu ignored\n",
+ ifp->idev->dev->name,
+ &ifp->addr,
+ nonce);
+ goto out;
+ }
/*
* We are colliding with another node
* who is doing DAD
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox