* Re: [PATCH 1/2] net: phy: leds: Clear phy_num_led_triggers on failure to avoid crash
From: Florian Fainelli @ 2017-01-24 20:52 UTC (permalink / raw)
To: Geert Uytterhoeven, David S. Miller, Zach Brown,
Volodymyr Bendiuga
Cc: Andrew Lunn, Magnus Öberg, netdev, linux-renesas-soc,
linux-kernel
In-Reply-To: <1485272470-3497-2-git-send-email-geert+renesas@glider.be>
On 01/24/2017 07:41 AM, Geert Uytterhoeven wrote:
> phy_attach_direct() ignores errors returned by
> phy_led_triggers_register(). I think that's OK, as LED triggers can be
> considered a non-critical feature.
>
> However, this causes problems later:
> - phy_led_trigger_change_speed() will access the array
> phy_device.phy_led_triggers, which has been freed in the error path
> of phy_led_triggers_register(), which may lead to a crash.
>
> - phy_led_triggers_unregister() will access the same array, leading to
> crashes during s2ram or poweroff, like:
>
> Unable to handle kernel NULL pointer dereference at virtual address
> 00000000
> ...
> [<c04116d4>] (__list_del_entry_valid) from [<c05e8948>] (led_trigger_unregister+0x34/0xcc)
> [<c05e8948>] (led_trigger_unregister) from [<c05336c4>] (phy_led_triggers_unregister+0x28/0x34)
> [<c05336c4>] (phy_led_triggers_unregister) from [<c0531d44>] (phy_detach+0x30/0x74)
> [<c0531d44>] (phy_detach) from [<c0538bdc>] (sh_eth_close+0x64/0x9c)
> [<c0538bdc>] (sh_eth_close) from [<c04d4ce0>] (dpm_run_callback+0x48/0xc8)
>
> or:
>
> list_del corruption. prev->next should be dede6540, but was 2e323931
> ------------[ cut here ]------------
> kernel BUG at lib/list_debug.c:52!
> ...
> [<c02f6d70>] (__list_del_entry_valid) from [<c0425168>] (led_trigger_unregister+0x34/0xcc)
> [<c0425168>] (led_trigger_unregister) from [<c03a05a0>] (phy_led_triggers_unregister+0x28/0x34)
> [<c03a05a0>] (phy_led_triggers_unregister) from [<c039ec04>] (phy_detach+0x30/0x74)
> [<c039ec04>] (phy_detach) from [<c03a4fc0>] (sh_eth_close+0x6c/0xa4)
> [<c03a4fc0>] (sh_eth_close) from [<c0483234>] (__dev_close_many+0xac/0xd0)
>
> To fix this, clear phy_device.phy_num_led_triggers in the error path of
> phy_led_triggers_register() fails.
>
> Note that the "No phy led trigger registered for speed" message will
> still be printed on link speed changes, which is a good cue that
> something went wrong with the LED triggers.
>
> Fixes: 2e0bc452f4721520 ("net: phy: leds: add support for led triggers on phy link state change")
> Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
--
Florian
^ permalink raw reply
* Re: [PATCH v2] virtio_net: fix PAGE_SIZE > 64k
From: Michael S. Tsirkin @ 2017-01-24 20:45 UTC (permalink / raw)
To: David Miller
Cc: linux-kernel, jasowang, virtualization, netdev, john.fastabend
In-Reply-To: <20170124.150959.1708982899683381291.davem@davemloft.net>
On Tue, Jan 24, 2017 at 03:09:59PM -0500, David Miller wrote:
> From: "Michael S. Tsirkin" <mst@redhat.com>
> Date: Tue, 24 Jan 2017 21:53:13 +0200
>
> > I didn't realise. Why can't we? I thought that adjust_header is an
> > optional feature that userspace can test for, so no rush.
>
> No, we want the base set of XDP features to be present in all drivers
> supporting XDP.
I see, I didn't realize this. In light of this, is there any
guidance *how much* head room is required to be considered
valid? We already have 12 bytes of headroom.
I'm generally sorry it's taking long, a large part of that is difficulty
figuring out the requirements: when we discussed this on LPC my
take-away was that one of the first users will be fighting DDOS attacks.
In light of this, I assumed that
- supporting just DROP (or DROP+TX) would already be useful
- XDP_PASS shouldn't be too slow as some people will run all
their traffic with XDP enabled
- people actually want this in virtio because they run in a vm
Since then I heard opinions that seem to imply that
- you must support all features, not just DROP, otherwise it's useless
- XDP_PASS is a slow path fallback as people are not expected to mix XDP
with regular sockets
- DDOS protection and by extension XDP in virtio is a developer's toy anyway
Thus, expect some slowness while I figure it all out.
--
MST
^ permalink raw reply
* Re: [PATCH net-next 0/5] External MDIO support for mv88e6xxx
From: David Miller @ 2017-01-24 20:35 UTC (permalink / raw)
To: andrew; +Cc: netdev, vivien.didelot, f.fainelli
In-Reply-To: <1485266031-4980-1-git-send-email-andrew@lunn.ch>
From: Andrew Lunn <andrew@lunn.ch>
Date: Tue, 24 Jan 2017 14:53:46 +0100
> The mv88e6390 family of switches has two MDIO busses, one internal to
> the switch and a second one for external usage. Older generations of switches
> have a single MDIO bus, which is used both internally and externally.
>
> Refactor the existing MDIO driver code to allow for multiple MDIO
> busses, and implement the second MDIO bus on mv88e6390.
>
> This is a rewrite of a patch previously submitted as part of "Batch
> 3". It has been broken up into 5 smaller patches. A compatible string
> is now used in the device tree to indicate the external MDIO bus.
Series applied, thanks Andrew.
There seems to be a bit of inconsistency wrt. tab vs. space for
indentation in the binding files, just FYI. I had to make some small
adjustments to this patch series in order to keep GIT from warning
when I applied this stuff.
^ permalink raw reply
* Re: [PATCH net-next v6 1/4] net: dsa: mv88e6xxx: Don't forbid MDIO I/Os for PHY addr >= num_of_ports
From: Gregory CLEMENT @ 2017-01-24 20:31 UTC (permalink / raw)
To: Andrew Lunn
Cc: Vivien Didelot, Florian Fainelli, netdev, linux-kernel,
David S. Miller, Jason Cooper, Sebastian Hesselbarth,
Thomas Petazzoni, linux-arm-kernel, Nadav Haklai, Wilson Ding,
Kostya Porotchkin, Joe Zhou, Jon Pannell
In-Reply-To: <7665c54a8da0d4338d5935ab272952ea6547d815.1485288602.git-series.gregory.clement@free-electrons.com>
Hi,
On mar., janv. 24 2017, Gregory CLEMENT <gregory.clement@free-electrons.com> wrote:
> From: Romain Perier <romain.perier@free-electrons.com>
>
> Some Marvell ethernet switches have internal ethernet transceivers with
> hardcoded phy addresses. These addresses can be greater than the number
> of ports or its value might be different than the associated port number.
> This is for example the case for MV88E6341 that has 6 ports and internal
> Port 1 to Port4 PHYs mapped at SMI addresses from 0x11 to 0x14.
>
> This commits fixes the issue by removing the condition in MDIO callbacks.
>
> Signed-off-by: Romain Perier <romain.perier@free-electrons.com>
> Reviewed-by: Andrew Lunn <andrew@lunn.ch>
This flag is missing:
Reviewed-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Actually I added during an interactive rebase, but I had to abort it
and it was lost.
Gregory
> Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
> ---
> drivers/net/dsa/mv88e6xxx/chip.c | 6 ------
> 1 file changed, 6 deletions(-)
>
> diff --git a/drivers/net/dsa/mv88e6xxx/chip.c b/drivers/net/dsa/mv88e6xxx/chip.c
> index c7e08e13bb54..7d942f8a42a7 100644
> --- a/drivers/net/dsa/mv88e6xxx/chip.c
> +++ b/drivers/net/dsa/mv88e6xxx/chip.c
> @@ -2882,9 +2882,6 @@ static int mv88e6xxx_mdio_read(struct mii_bus *bus, int phy, int reg)
> u16 val;
> int err;
>
> - if (phy >= mv88e6xxx_num_ports(chip))
> - return 0xffff;
> -
> mutex_lock(&chip->reg_lock);
> err = mv88e6xxx_phy_read(chip, phy, reg, &val);
> mutex_unlock(&chip->reg_lock);
> @@ -2897,9 +2894,6 @@ static int mv88e6xxx_mdio_write(struct mii_bus *bus, int phy, int reg, u16 val)
> struct mv88e6xxx_chip *chip = bus->priv;
> int err;
>
> - if (phy >= mv88e6xxx_num_ports(chip))
> - return 0xffff;
> -
> mutex_lock(&chip->reg_lock);
> err = mv88e6xxx_phy_write(chip, phy, reg, val);
> mutex_unlock(&chip->reg_lock);
> --
> git-series 0.9.1
--
Gregory Clement, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com
^ permalink raw reply
* Re: [PATCH v2] bpf: Restrict cgroup bpf hooks to the init netns
From: David Ahern @ 2017-01-24 20:29 UTC (permalink / raw)
To: Andy Lutomirski, Alexei Starovoitov, Tejun Heo
Cc: Andy Lutomirski, Network Development, David S. Miller,
Daniel Borkmann, Alexei Starovoitov
In-Reply-To: <CALCETrW-Sj9eM+Nm9PVpqJky_30YSGhzc8hnzmddPkhHzb5_Fg@mail.gmail.com>
On 1/24/17 11:54 AM, Andy Lutomirski wrote:
>>> My point is that, in 4.10-rc, it doesn't work right, and I doubt this
>>> problem is restricted to just 'ip vrf'. Without some kind of change
>>> to the way that netns and cgroup+bpf interact, anything that uses
>>> sk_bound_dev_if or reads the ifindex on an skb will be subject to a
>>> huge footgun that unprivileged programs can trigger and any future
>>> attempt to make the cgroup+bpf work for unprivileged users is going to
>>> be more complicated than it deserves to be.
>>
>> For n-th time, the current BPF_PROG_TYPE_CGROUP* is root only and
>> speculation about unprivileged usage are not helping the discussion.
>
> ...which has nothing to do with my example of how it's broken. I used
> 'ip vrf' the way it was intended to be used and then I ran code in it
> that uses only APIs that predate eBPF. The result was that the
> program obtained a broken socket that had sk_bound_dev_if filled in
> with a nonsense index. There's no speculation -- it's just broken.
Again, there are many ways to arrive at this point where a socket is bound to a device index that is invalid. Nothing stops you from downloading a bpf program with an invalid ifindex. That is up to the user.
Users do not run around exec'ing commands in random network contexts (namespace, vrf, device, whatever) and expect them to just work.
>
> Maybe you can argue that this is a missing feature in cgroup+bpf (no
> API to query which netns is in use) and a corresponding bug in 'ip
> vrf', but I see this as evidence that cgroup+bpf as it exists in 4.10
> is not carefully enough throught through. The only non-example user
> of it that I can find (ip vrf) is buggy and can't really be fixed
> using mechanisms that exist in 4.10-rc.
The argument is that cgroups and namespaces are completely disjoint infrastructure and that users need to know what they are doing. If one uses a management tool (ip in this case) in a consistent manner it will work. If 'ip netns' is used to change namespaces, vrf is reset on the switch. And, I have a patch now to properly nest namespaces and vrfs so that mgmt vrf in multiple namespaces will work properly.
>
>>
>>> things up so that unshare will malfunction. It should avoid
>>> malfunctioning when running Linux programs that are unaware of it.
>>
>> I agree that for VRF use case it will help to make programs netns
>> aware by adding new bpf_get_current_netns_id() or something helper,
>> but it's up to the program to function properly or be broken.
>
> This will cause David's code to run slower, and I think he wants very
> high performance.
This is a socket create path not a packet path. While overhead should be contained, a few extra cycles should be fine.
Adding the capability to allow users to check the netns id would offer a solution to the namespace problem, but there is nothing that *requires* a bpf program to do it.
Who's to say an admin does not *want* all processes in a group to have sockets bound to a non-existent device? 'ip vrf' restricts the device index to a VRF device because as a management tool I want it to be user friendly, but generically the BPF code does not have any restrictions. ifindex can be any u32 value.
^ permalink raw reply
* Re: [PATCH net 0/3] alx: fix fallout from multi queue conversion
From: David Miller @ 2017-01-24 20:28 UTC (permalink / raw)
To: tobias.regnery; +Cc: netdev, jcliburn, chris.snook
In-Reply-To: <cover.1485263833.git.tobias.regnery@gmail.com>
From: Tobias Regnery <tobias.regnery@gmail.com>
Date: Tue, 24 Jan 2017 14:34:21 +0100
> Here are 3 fixes for the multi queue conversion in v4.10.
>
> The first patch fixes a wrong condition in an if statement.
>
> Patches 2 and 3 fixes regressions in the corner case when requesting msi-x
> interrupts fails and we fall back to msi or legacy interrupts.
Series applied, thanks.
^ permalink raw reply
* Re: [PATCH net-next v6 4/4] phy: marvell: Add support for the PHY embedded in the topaz switch
From: Andrew Lunn @ 2017-01-24 20:28 UTC (permalink / raw)
To: Gregory CLEMENT
Cc: Vivien Didelot, Florian Fainelli, netdev, linux-kernel,
David S. Miller, Jason Cooper, Sebastian Hesselbarth,
Thomas Petazzoni, linux-arm-kernel, Nadav Haklai, Wilson Ding,
Kostya Porotchkin, Joe Zhou, Jon Pannell
In-Reply-To: <f8526f4a5d99223fc90fe5579b4958eaa0950a3a.1485288602.git-series.gregory.clement@free-electrons.com>
On Tue, Jan 24, 2017 at 09:10:26PM +0100, Gregory CLEMENT wrote:
> The PHY with the ID 0x1410C00
:-(
I don't have a better reference, but
Linux/Documentation/devicetree/bindings/net/phy.txt says:
22 If the phy's identifier is known then the list may contain an entry
23 of the form: "ethernet-phy-idAAAA.BBBB" where
24 AAAA - The value of the 16 bit Phy Identifier 1 register as
25 4 hex digits. This is the chip vendor OUI bits 3:18
26 BBBB - The value of the 16 bit Phy Identifier 2 register as
27 4 hex digits. This is the chip vendor OUI bits 19:24,
28 followed by 10 bits of a vendor specific ID.
So the lower 10 bits of 0x1410C00 are 0. So we know it is a Marvell
PHY from the OUI, but the vendor specific bits are all 0.
Please take a look at:
https://marc.info/?l=linux-netdev&m=148495522620757&w=1
and
https://marc.info/?l=linux-netdev&m=148495510320714&w=1
Maybe i should submit these two independently, so you can extend it
for the 88E6341 family.
Andrew
^ permalink raw reply
* Re: [PATCH] cxgbit: use T6 specific macro to set force bit
From: David Miller @ 2017-01-24 20:26 UTC (permalink / raw)
To: varun; +Cc: bart.vanassche, target-devel, netdev, indranil
In-Reply-To: <1485257822-729-1-git-send-email-varun@chelsio.com>
From: Varun Prakash <varun@chelsio.com>
Date: Tue, 24 Jan 2017 17:07:02 +0530
> For T6 adapters use T6 specific macro to set
> force bit.
>
> Signed-off-by: Varun Prakash <varun@chelsio.com>
What tree is this for?
You must always indicate this in your PATCH subject text.
^ permalink raw reply
* Re: [RESENT PATCH net,stable] qmi_wwan/cdc_ether: add device ID for HP lt2523 (Novatel E371) WWAN card
From: David Miller @ 2017-01-24 20:25 UTC (permalink / raw)
To: bjorn; +Cc: netdev, oliver, linux-usb, dcbw
In-Reply-To: <20170124094538.20351-1-bjorn@mork.no>
From: Bjørn Mork <bjorn@mork.no>
Date: Tue, 24 Jan 2017 10:45:38 +0100
> Another rebranded Novatel E371. qmi_wwan should drive this device, while
> cdc_ether should ignore it. Even though the USB descriptors are plain
> CDC-ETHER that USB interface is a QMI interface. Ref commit 7fdb7846c9ca
> ("qmi_wwan/cdc_ether: add device IDs for Dell 5804 (Novatel E371) WWAN
> card")
>
> Cc: Dan Williams <dcbw@redhat.com>
> Signed-off-by: Bjørn Mork <bjorn@mork.no>
Applied.
^ permalink raw reply
* Re: [PATCH][V3][net-next] net: sctp: fix array overrun read on sctp_timer_tbl
From: David Miller @ 2017-01-24 20:24 UTC (permalink / raw)
To: colin.king; +Cc: vyasevich, nhorman, linux-sctp, netdev, linux-kernel
In-Reply-To: <20170124092554.27019-1-colin.king@canonical.com>
From: Colin King <colin.king@canonical.com>
Date: Tue, 24 Jan 2017 09:25:54 +0000
> From: Colin Ian King <colin.king@canonical.com>
>
> Table sctp_timer_tbl is missing a TIMEOUT_RECONF string so
> add this in. Also compare timeout with the size of the array
> sctp_timer_tbl rather than SCTP_EVENT_TIMEOUT_MAX. Also add
> a build time check that SCTP_EVENT_TIMEOUT_MAX is correct
> so we don't ever get this kind of mismatch between the table
> and SCTP_EVENT_TIMEOUT_MAX in the future.
>
> Kudos to Marcelo Ricardo Leitner for spotting the missing string
> and suggesting the build time sanity check.
>
> Fixes CoverityScan CID#1397639 ("Out-of-bounds read")
>
> Fixes: 7b9438de0cd4 ("sctp: add stream reconf timer")
> Signed-off-by: Colin Ian King <colin.king@canonical.com>
Applied.
^ permalink raw reply
* Re: [PATCH 5/5] gtp: let userspace handle packets for invalid tunnels
From: Pablo Neira Ayuso @ 2017-01-24 20:19 UTC (permalink / raw)
To: Andreas Schultz; +Cc: netdev, Lionel Gauthier, openbsc, laforge
In-Reply-To: <1060167008.597625.1485288151511.JavaMail.zimbra@tpip.net>
On Tue, Jan 24, 2017 at 09:02:31PM +0100, Andreas Schultz wrote:
> Hi Pablo,
>
> ----- On Jan 24, 2017, at 8:03 PM, pablo pablo@netfilter.org wrote:
>
> > Hi Andreas,
> >
> > On Tue, Jan 24, 2017 at 06:24:02PM +0100, Andreas Schultz wrote:
> >> enable userspace to send error replies for invalid tunnels
> >>
> >> Signed-off-by: Andreas Schultz <aschultz@tpip.net>
> >> ---
> >> drivers/net/gtp.c | 8 ++++----
> >> 1 file changed, 4 insertions(+), 4 deletions(-)
> >>
> >> diff --git a/drivers/net/gtp.c b/drivers/net/gtp.c
> >> index 912721e..c607333 100644
> >> --- a/drivers/net/gtp.c
> >> +++ b/drivers/net/gtp.c
> >> @@ -198,12 +198,12 @@ static int gtp0_udp_encap_recv(struct gtp_dev *gtp, struct
> >> sk_buff *skb,
> >> pctx = gtp0_pdp_find(gtp, be64_to_cpu(gtp0->tid));
> >> if (!pctx) {
> >> netdev_dbg(gtp->dev, "No PDP ctx to decap skb=%p\n", skb);
> >> - return -1;
> >> + return 1;
> >> }
> >>
> >> if (!gtp_check_src_ms(skb, pctx, hdrlen)) {
> >> netdev_dbg(gtp->dev, "No PDP ctx for this MS\n");
> >> - return -1;
> >> + return 1;
> >
> > So userspace gets the packet that we cannot forward. I guess your
> > userspace codebase performs this sanity checks again so you can send
> > the appropriate error reply?
>
> For TEID /= 0, the only reply is a T-PDU of type error indication. There
> is no cause specified. So I don't actually have to repeat the check.
> TEID == 0 is more interesting, this tells userspace that it tried to
> send on an invalid tunnel and should tear it down.
Ah, I see, thanks for explaining, I guess this is good to debug
misconfigurations.
> If you like, you can have a look at the userspace code. The relevant piece
> is at https://github.com/travelping/gtp_u_kmod/blob/master/src/gtp_u_kmod_port.erl#L231
>
> But be warned, it's written in Erlang ;-)
Will have a look, but you taking me away from my usual domains :).
^ permalink raw reply
* [net-next 10/12] IB/mlx5: Enable Eth VFs to query their min-inline value for user-space
From: Saeed Mahameed @ 2017-01-24 20:16 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, Or Gerlitz, Saeed Mahameed
In-Reply-To: <20170124201652.2920-1-saeedm@mellanox.com>
From: Or Gerlitz <ogerlitz@mellanox.com>
For some mlx5 HW models (CX4, CX4Lx), the VF driver needs to put part
of the packet headers on the TX descriptor so the e-switch can do proper
matching and steering. This is called "min-inline", it's advertized to
the VF by the FW and also enforced on them by the HW, such that if they
don't obey, their packets are dropped.
SRIOV VF libmlx5 instances should take into account the min-inline
value of their vports. For that end, we provide this value through
the vendor response part of init_ucontext command.
The min inline value is reported in a way which will let newer libmlx5
instances realize that they are running over an older kernel and act
accordingly (e.g apply some educated guess).
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/infiniband/hw/mlx5/main.c | 9 +++++++++
include/uapi/rdma/mlx5-abi.h | 14 +++++++++++++-
2 files changed, 22 insertions(+), 1 deletion(-)
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
index 8727116a4cab..9d8535385bb8 100644
--- a/drivers/infiniband/hw/mlx5/main.c
+++ b/drivers/infiniband/hw/mlx5/main.c
@@ -53,6 +53,7 @@
#include <linux/in.h>
#include <linux/etherdevice.h>
#include <linux/mlx5/fs.h>
+#include <linux/mlx5/vport.h>
#include "mlx5_ib.h"
#define DRIVER_NAME "mlx5_ib"
@@ -1202,6 +1203,14 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev,
resp.response_length += sizeof(resp.cmds_supp_uhw);
}
+ if (field_avail(typeof(resp), eth_min_inline, udata->outlen)) {
+ if (mlx5_ib_port_link_layer(ibdev, 1) == IB_LINK_LAYER_ETHERNET) {
+ mlx5_query_min_inline(dev->mdev, &resp.eth_min_inline);
+ resp.eth_min_inline++;
+ }
+ resp.response_length += sizeof(resp.eth_min_inline);
+ }
+
/*
* We don't want to expose information from the PCI bar that is located
* after 4096 bytes, so if the arch only supports larger pages, let's
diff --git a/include/uapi/rdma/mlx5-abi.h b/include/uapi/rdma/mlx5-abi.h
index 85dc966ea70b..da7cd62bace7 100644
--- a/include/uapi/rdma/mlx5-abi.h
+++ b/include/uapi/rdma/mlx5-abi.h
@@ -90,6 +90,17 @@ enum mlx5_user_cmds_supp_uhw {
MLX5_USER_CMDS_SUPP_UHW_CREATE_AH = 1 << 1,
};
+/* The eth_min_inline response value is set to off-by-one vs the FW
+ * returned value to allow user-space to deal with older kernels.
+ */
+enum mlx5_user_inline_mode {
+ MLX5_USER_INLINE_MODE_NA,
+ MLX5_USER_INLINE_MODE_NONE,
+ MLX5_USER_INLINE_MODE_L2,
+ MLX5_USER_INLINE_MODE_IP,
+ MLX5_USER_INLINE_MODE_TCP_UDP,
+};
+
struct mlx5_ib_alloc_ucontext_resp {
__u32 qp_tab_size;
__u32 bf_reg_size;
@@ -106,7 +117,8 @@ struct mlx5_ib_alloc_ucontext_resp {
__u32 response_length;
__u8 cqe_version;
__u8 cmds_supp_uhw;
- __u16 reserved2;
+ __u8 eth_min_inline;
+ __u8 reserved2;
__u64 hca_core_clock_offset;
__u32 log_uar_size;
__u32 num_uars_per_page;
--
2.11.0
^ permalink raw reply related
* [pull request][net-next 00/12] Mellanox mlx5 updates 2017-01-24
From: Saeed Mahameed @ 2017-01-24 20:16 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, Or Gerlitz, Saeed Mahameed
Hi Dave,
This pull request includes one new feature to support offloading IPv6
tunnels in switchdev mode, in addition to some small mlx5 updates.
Details are down bleow.
Please pull and let me know if there's any problem.
Thanks,
Saeed.
---
The following changes since commit 4548b683b78137f8eadeb312b94e20bb0d4a7141:
Introduce a sysctl that modifies the value of PROT_SOCK. (2017-01-24 12:10:51 -0500)
are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux.git tags/mlx5-updates-2017-01-24
for you to fetch changes up to 5eb0249b4352c813f0a3c31b967f6cb4b9869a50:
net/mlx5e: CQE compression control code reuse (2017-01-24 21:14:08 +0200)
----------------------------------------------------------------
mlx5-updates-2017-24-01
The first seven patches from Or Gerlitz in this series further enhances
the mlx5 SRIOV switchdev mode to support offloading IPv6 tunnels using the
TC tunnel key set (encap) and unset (decap) actions.
Or Gerlitz says:
========================
As part of doing this change, few cleanups are done in the IPv4 code,
later we move to use the full tunnel key info provided to the driver as
the key for our internal hashing which is used to identify cases where
the same tunnel is used for encapsulating multiple flows. As done in the
IPv4 case, the control path for offloading IPv6 tunnels uses route/neigh
lookups and construction of the IPv6 tunnel headers on the encap path and
matching on the outer hears in the decap path.
The last patch of the series enlarges the HW FDB size for the switchdev mode,
so it has now room to contain offloaded flows as many as min(max number
of HW flow counters supported, max HW table size supported).
========================
Next to Or's series you can find several patches handling several topics.
>From Mohamad, add support for SRIOV VF min rate guarantee by using the
TSAR BW share weights mechanism.
>From Or, Two patches to enable Eth VFs to query their min-inline value for
user-space.
for that we move a mlx5 low level min inline helper function from mlx5
ethernet driver into the core driver and then use it in mlx5_ib to expose
the inline mode to rdma applications through libmlx5.
>From Kamal Heib, Reduce memory consumption on kdump kernel.
>From Shaker Daibes, code reuse in CQE compression control logic
Thanks,
Saeed.
----------------------------------------------------------------
Kamal Heib (1):
net/mlx5e: Reduce memory consumption on kdump kernel
Mohamad Haj Yahia (1):
net/mlx5: Add support for setting VF min rate
Or Gerlitz (9):
net/mlx5: Use exact encap header size for the FW input buffer
net/mlx5e: Add TC offloads matching on IPv6 encapsulation headers
net/mlx5e: TC ipv4 tunnel encap offload cosmetic changes
net/mlx5e: Use the full tunnel key info for encapsulation offload house-keeping
net/mlx5e: Maximize ip tunnel key usage on the TC offloading path
net/mlx5e: Support SRIOV TC encapsulation offloads for IPv6 tunnels
net/mlx5: E-Switch, Enlarge the FDB size for the switchdev mode
net/mlx5: Push min-inline mode resolution helper into the core
IB/mlx5: Enable Eth VFs to query their min-inline value for user-space
Shaker Daibes (1):
net/mlx5e: CQE compression control code reuse
drivers/infiniband/hw/mlx5/main.c | 9 +
drivers/net/ethernet/mellanox/mlx5/core/en.h | 9 +-
drivers/net/ethernet/mellanox/mlx5/core/en_clock.c | 7 +-
.../net/ethernet/mellanox/mlx5/core/en_ethtool.c | 17 +-
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 44 ++--
drivers/net/ethernet/mellanox/mlx5/core/en_rx.c | 8 +-
drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 242 +++++++++++++++++----
drivers/net/ethernet/mellanox/mlx5/core/eswitch.c | 97 ++++++++-
drivers/net/ethernet/mellanox/mlx5/core/eswitch.h | 20 +-
.../ethernet/mellanox/mlx5/core/eswitch_offloads.c | 13 +-
drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c | 7 +-
drivers/net/ethernet/mellanox/mlx5/core/vport.c | 17 ++
include/linux/mlx5/mlx5_ifc.h | 4 +-
include/linux/mlx5/vport.h | 1 +
include/uapi/rdma/mlx5-abi.h | 14 +-
15 files changed, 389 insertions(+), 120 deletions(-)
^ permalink raw reply
* [net-next 08/12] net/mlx5: Add support for setting VF min rate
From: Saeed Mahameed @ 2017-01-24 20:16 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, Or Gerlitz, Mohamad Haj Yahia, Saeed Mahameed
In-Reply-To: <20170124201652.2920-1-saeedm@mellanox.com>
From: Mohamad Haj Yahia <mohamad@mellanox.com>
Add support for SRIOV VF min rate guarantee by using the TSAR BW share
weights mechanism.
The TSAR BW share vport attribute represents the weight of that vport
among the other vports weights which means that the actual vport BW
percentage is the same vport weight percentage among the total vports
weights sum.
Signed-off-by: Mohamad Haj Yahia <mohamad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 5 +-
drivers/net/ethernet/mellanox/mlx5/core/eswitch.c | 97 +++++++++++++++++++++--
drivers/net/ethernet/mellanox/mlx5/core/eswitch.h | 11 ++-
include/linux/mlx5/mlx5_ifc.h | 4 +-
4 files changed, 104 insertions(+), 13 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 3a06c81ef85e..c819d07fbdb3 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -3021,11 +3021,8 @@ static int mlx5e_set_vf_rate(struct net_device *dev, int vf, int min_tx_rate,
struct mlx5e_priv *priv = netdev_priv(dev);
struct mlx5_core_dev *mdev = priv->mdev;
- if (min_tx_rate)
- return -EOPNOTSUPP;
-
return mlx5_eswitch_set_vport_rate(mdev->priv.eswitch, vf + 1,
- max_tx_rate);
+ max_tx_rate, min_tx_rate);
}
static int mlx5_vport_link2ifla(u8 esw_link)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
index 4b3b60be319d..efa1a7a76d8a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
@@ -1415,7 +1415,7 @@ static void esw_destroy_tsar(struct mlx5_eswitch *esw)
}
static int esw_vport_enable_qos(struct mlx5_eswitch *esw, int vport_num,
- u32 initial_max_rate)
+ u32 initial_max_rate, u32 initial_bw_share)
{
u32 sched_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {0};
struct mlx5_vport *vport = &esw->vports[vport_num];
@@ -1439,6 +1439,7 @@ static int esw_vport_enable_qos(struct mlx5_eswitch *esw, int vport_num,
esw->qos.root_tsar_id);
MLX5_SET(scheduling_context, &sched_ctx, max_average_bw,
initial_max_rate);
+ MLX5_SET(scheduling_context, &sched_ctx, bw_share, initial_bw_share);
err = mlx5_create_scheduling_element_cmd(dev,
SCHEDULING_HIERARCHY_E_SWITCH,
@@ -1473,7 +1474,7 @@ static void esw_vport_disable_qos(struct mlx5_eswitch *esw, int vport_num)
}
static int esw_vport_qos_config(struct mlx5_eswitch *esw, int vport_num,
- u32 max_rate)
+ u32 max_rate, u32 bw_share)
{
u32 sched_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {0};
struct mlx5_vport *vport = &esw->vports[vport_num];
@@ -1497,7 +1498,9 @@ static int esw_vport_qos_config(struct mlx5_eswitch *esw, int vport_num,
esw->qos.root_tsar_id);
MLX5_SET(scheduling_context, &sched_ctx, max_average_bw,
max_rate);
+ MLX5_SET(scheduling_context, &sched_ctx, bw_share, bw_share);
bitmask |= MODIFY_SCHEDULING_ELEMENT_IN_MODIFY_BITMASK_MAX_AVERAGE_BW;
+ bitmask |= MODIFY_SCHEDULING_ELEMENT_IN_MODIFY_BITMASK_BW_SHARE;
err = mlx5_modify_scheduling_element_cmd(dev,
SCHEDULING_HIERARCHY_E_SWITCH,
@@ -1563,7 +1566,8 @@ static void esw_enable_vport(struct mlx5_eswitch *esw, int vport_num,
esw_apply_vport_conf(esw, vport);
/* Attach vport to the eswitch rate limiter */
- if (esw_vport_enable_qos(esw, vport_num, vport->info.max_rate))
+ if (esw_vport_enable_qos(esw, vport_num, vport->info.max_rate,
+ vport->qos.bw_share))
esw_warn(esw->dev, "Failed to attach vport %d to eswitch rate limiter", vport_num);
/* Sync with current vport context */
@@ -1952,6 +1956,7 @@ int mlx5_eswitch_get_vport_config(struct mlx5_eswitch *esw,
ivi->qos = evport->info.qos;
ivi->spoofchk = evport->info.spoofchk;
ivi->trusted = evport->info.trusted;
+ ivi->min_tx_rate = evport->info.min_rate;
ivi->max_tx_rate = evport->info.max_rate;
mutex_unlock(&esw->state_lock);
@@ -2046,23 +2051,103 @@ int mlx5_eswitch_set_vport_trust(struct mlx5_eswitch *esw,
return 0;
}
-int mlx5_eswitch_set_vport_rate(struct mlx5_eswitch *esw,
- int vport, u32 max_rate)
+static u32 calculate_vports_min_rate_divider(struct mlx5_eswitch *esw)
{
+ u32 fw_max_bw_share = MLX5_CAP_QOS(esw->dev, max_tsar_bw_share);
struct mlx5_vport *evport;
+ u32 max_guarantee = 0;
+ int i;
+
+ for (i = 0; i <= esw->total_vports; i++) {
+ evport = &esw->vports[i];
+ if (!evport->enabled || evport->info.min_rate < max_guarantee)
+ continue;
+ max_guarantee = evport->info.min_rate;
+ }
+
+ return max_t(u32, max_guarantee / fw_max_bw_share, 1);
+}
+
+static int normalize_vports_min_rate(struct mlx5_eswitch *esw, u32 divider)
+{
+ u32 fw_max_bw_share = MLX5_CAP_QOS(esw->dev, max_tsar_bw_share);
+ struct mlx5_vport *evport;
+ u32 vport_max_rate;
+ u32 vport_min_rate;
+ u32 bw_share;
+ int err;
+ int i;
+
+ for (i = 0; i <= esw->total_vports; i++) {
+ evport = &esw->vports[i];
+ if (!evport->enabled)
+ continue;
+ vport_min_rate = evport->info.min_rate;
+ vport_max_rate = evport->info.max_rate;
+ bw_share = MLX5_MIN_BW_SHARE;
+
+ if (vport_min_rate)
+ bw_share = MLX5_RATE_TO_BW_SHARE(vport_min_rate,
+ divider,
+ fw_max_bw_share);
+
+ if (bw_share == evport->qos.bw_share)
+ continue;
+
+ err = esw_vport_qos_config(esw, i, vport_max_rate,
+ bw_share);
+ if (!err)
+ evport->qos.bw_share = bw_share;
+ else
+ return err;
+ }
+
+ return 0;
+}
+
+int mlx5_eswitch_set_vport_rate(struct mlx5_eswitch *esw, int vport,
+ u32 max_rate, u32 min_rate)
+{
+ u32 fw_max_bw_share = MLX5_CAP_QOS(esw->dev, max_tsar_bw_share);
+ bool min_rate_supported = MLX5_CAP_QOS(esw->dev, esw_bw_share) &&
+ fw_max_bw_share >= MLX5_MIN_BW_SHARE;
+ bool max_rate_supported = MLX5_CAP_QOS(esw->dev, esw_rate_limit);
+ struct mlx5_vport *evport;
+ u32 previous_min_rate;
+ u32 divider;
int err = 0;
if (!ESW_ALLOWED(esw))
return -EPERM;
if (!LEGAL_VPORT(esw, vport))
return -EINVAL;
+ if ((min_rate && !min_rate_supported) || (max_rate && !max_rate_supported))
+ return -EOPNOTSUPP;
mutex_lock(&esw->state_lock);
evport = &esw->vports[vport];
- err = esw_vport_qos_config(esw, vport, max_rate);
+
+ if (min_rate == evport->info.min_rate)
+ goto set_max_rate;
+
+ previous_min_rate = evport->info.min_rate;
+ evport->info.min_rate = min_rate;
+ divider = calculate_vports_min_rate_divider(esw);
+ err = normalize_vports_min_rate(esw, divider);
+ if (err) {
+ evport->info.min_rate = previous_min_rate;
+ goto unlock;
+ }
+
+set_max_rate:
+ if (max_rate == evport->info.max_rate)
+ goto unlock;
+
+ err = esw_vport_qos_config(esw, vport, max_rate, evport->qos.bw_share);
if (!err)
evport->info.max_rate = max_rate;
+unlock:
mutex_unlock(&esw->state_lock);
return err;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
index cea1660d59ac..5b78883d5654 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
@@ -50,6 +50,11 @@
#define FDB_UPLINK_VPORT 0xffff
+#define MLX5_MIN_BW_SHARE 1
+
+#define MLX5_RATE_TO_BW_SHARE(rate, divider, limit) \
+ min_t(u32, max_t(u32, (rate) / (divider), MLX5_MIN_BW_SHARE), limit)
+
/* L2 -mac address based- hash helpers */
struct l2addr_node {
struct hlist_node hlist;
@@ -116,6 +121,7 @@ struct mlx5_vport_info {
u8 qos;
u64 node_guid;
int link_state;
+ u32 min_rate;
u32 max_rate;
bool spoofchk;
bool trusted;
@@ -138,6 +144,7 @@ struct mlx5_vport {
struct {
bool enabled;
u32 esw_tsar_ix;
+ u32 bw_share;
} qos;
bool enabled;
@@ -249,8 +256,8 @@ int mlx5_eswitch_set_vport_spoofchk(struct mlx5_eswitch *esw,
int vport, bool spoofchk);
int mlx5_eswitch_set_vport_trust(struct mlx5_eswitch *esw,
int vport_num, bool setting);
-int mlx5_eswitch_set_vport_rate(struct mlx5_eswitch *esw,
- int vport, u32 max_rate);
+int mlx5_eswitch_set_vport_rate(struct mlx5_eswitch *esw, int vport,
+ u32 max_rate, u32 min_rate);
int mlx5_eswitch_get_vport_config(struct mlx5_eswitch *esw,
int vport, struct ifla_vf_info *ivi);
int mlx5_eswitch_get_vport_stats(struct mlx5_eswitch *esw,
diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h
index d96ebc319d63..a919dfb920ae 100644
--- a/include/linux/mlx5/mlx5_ifc.h
+++ b/include/linux/mlx5/mlx5_ifc.h
@@ -547,7 +547,9 @@ struct mlx5_ifc_e_switch_cap_bits {
struct mlx5_ifc_qos_cap_bits {
u8 packet_pacing[0x1];
u8 esw_scheduling[0x1];
- u8 reserved_at_2[0x1e];
+ u8 esw_bw_share[0x1];
+ u8 esw_rate_limit[0x1];
+ u8 reserved_at_4[0x1c];
u8 reserved_at_20[0x20];
--
2.11.0
^ permalink raw reply related
* [net-next 04/12] net/mlx5e: Use the full tunnel key info for encapsulation offload house-keeping
From: Saeed Mahameed @ 2017-01-24 20:16 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, Or Gerlitz, Saeed Mahameed
In-Reply-To: <20170124201652.2920-1-saeedm@mellanox.com>
From: Or Gerlitz <ogerlitz@mellanox.com>
Currently we use subset of the input tunnel key fields (id, ip daddr,
dst port) which are provided by upper layers to indentify flows that should
go through the same encapsulation and maintain the HW encapsulation table.
This is redundant and can get us wrong.
Instead, keep a copy of the ip tunnel info provided by the user
through TC and have the tunnel key part as the key to our internal hash.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Hadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 38 +++++++++--------------
drivers/net/ethernet/mellanox/mlx5/core/eswitch.h | 9 ++----
2 files changed, 17 insertions(+), 30 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index a515444d2489..477b2796c12e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -668,15 +668,15 @@ static int parse_tc_nic_actions(struct mlx5e_priv *priv, struct tcf_exts *exts,
return 0;
}
-static inline int cmp_encap_info(struct mlx5_encap_info *a,
- struct mlx5_encap_info *b)
+static inline int cmp_encap_info(struct ip_tunnel_key *a,
+ struct ip_tunnel_key *b)
{
return memcmp(a, b, sizeof(*a));
}
-static inline int hash_encap_info(struct mlx5_encap_info *info)
+static inline int hash_encap_info(struct ip_tunnel_key *key)
{
- return jhash(info, sizeof(*info), 0);
+ return jhash(key, sizeof(*key), 0);
}
static int mlx5e_route_lookup_ipv4(struct mlx5e_priv *priv,
@@ -762,6 +762,7 @@ static int mlx5e_create_encap_header_ipv4(struct mlx5e_priv *priv,
struct net_device **out_dev)
{
int max_encap_size = MLX5_CAP_ESW(priv->mdev, max_encap_header_size);
+ struct ip_tunnel_key *tun_key = &e->tun_info.key;
struct neighbour *n = NULL;
struct flowi4 fl4 = {};
char *encap_header;
@@ -777,13 +778,13 @@ static int mlx5e_create_encap_header_ipv4(struct mlx5e_priv *priv,
switch (e->tunnel_type) {
case MLX5_HEADER_TYPE_VXLAN:
fl4.flowi4_proto = IPPROTO_UDP;
- fl4.fl4_dport = e->tun_info.tp_dst;
+ fl4.fl4_dport = tun_key->tp_dst;
break;
default:
err = -EOPNOTSUPP;
goto out;
}
- fl4.daddr = e->tun_info.daddr;
+ fl4.daddr = tun_key->u.ipv4.dst;
err = mlx5e_route_lookup_ipv4(priv, mirred_dev, out_dev,
&fl4, &n, &saddr, &ttl);
@@ -805,9 +806,9 @@ static int mlx5e_create_encap_header_ipv4(struct mlx5e_priv *priv,
case MLX5_HEADER_TYPE_VXLAN:
encap_size = gen_vxlan_header_ipv4(*out_dev, encap_header,
e->h_dest, ttl,
- e->tun_info.daddr,
- saddr, e->tun_info.tp_dst,
- e->tun_info.tun_id);
+ tun_key->u.ipv4.dst,
+ saddr, tun_key->tp_dst,
+ tunnel_id_to_key32(tun_key->tun_id));
break;
default:
err = -EOPNOTSUPP;
@@ -831,13 +832,11 @@ static int mlx5e_attach_encap(struct mlx5e_priv *priv,
struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
unsigned short family = ip_tunnel_info_af(tun_info);
struct ip_tunnel_key *key = &tun_info->key;
- struct mlx5_encap_info info;
struct mlx5_encap_entry *e;
struct net_device *out_dev;
+ int tunnel_type, err;
uintptr_t hash_key;
bool found = false;
- int tunnel_type;
- int err;
/* udp dst port must be set */
if (!memchr_inv(&key->tp_dst, 0, sizeof(key->tp_dst)))
@@ -853,8 +852,6 @@ static int mlx5e_attach_encap(struct mlx5e_priv *priv,
if (mlx5e_vxlan_lookup_port(priv, be16_to_cpu(key->tp_dst)) &&
MLX5_CAP_ESW(priv->mdev, vxlan_encap_decap)) {
- info.tp_dst = key->tp_dst;
- info.tun_id = tunnel_id_to_key32(key->tun_id);
tunnel_type = MLX5_HEADER_TYPE_VXLAN;
} else {
netdev_warn(priv->netdev,
@@ -862,22 +859,17 @@ static int mlx5e_attach_encap(struct mlx5e_priv *priv,
return -EOPNOTSUPP;
}
- switch (family) {
- case AF_INET:
- info.daddr = key->u.ipv4.dst;
- break;
- case AF_INET6:
+ if (family == AF_INET6) {
netdev_warn(priv->netdev,
"IPv6 tunnel encap offload isn't supported\n");
- default:
return -EOPNOTSUPP;
}
- hash_key = hash_encap_info(&info);
+ hash_key = hash_encap_info(key);
hash_for_each_possible_rcu(esw->offloads.encap_tbl, e,
encap_hlist, hash_key) {
- if (!cmp_encap_info(&e->tun_info, &info)) {
+ if (!cmp_encap_info(&e->tun_info.key, key)) {
found = true;
break;
}
@@ -892,7 +884,7 @@ static int mlx5e_attach_encap(struct mlx5e_priv *priv,
if (!e)
return -ENOMEM;
- e->tun_info = info;
+ e->tun_info = *tun_info;
e->tunnel_type = tunnel_type;
INIT_LIST_HEAD(&e->flows);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
index 8661dd3f542c..cea1660d59ac 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
@@ -36,6 +36,7 @@
#include <linux/if_ether.h>
#include <linux/if_link.h>
#include <net/devlink.h>
+#include <net/ip_tunnels.h>
#include <linux/mlx5/device.h>
#define MLX5_MAX_UC_PER_VPORT(dev) \
@@ -274,18 +275,12 @@ enum {
#define MLX5_FLOW_CONTEXT_ACTION_VLAN_POP 0x40
#define MLX5_FLOW_CONTEXT_ACTION_VLAN_PUSH 0x80
-struct mlx5_encap_info {
- __be32 daddr;
- __be32 tun_id;
- __be16 tp_dst;
-};
-
struct mlx5_encap_entry {
struct hlist_node encap_hlist;
struct list_head flows;
u32 encap_id;
struct neighbour *n;
- struct mlx5_encap_info tun_info;
+ struct ip_tunnel_info tun_info;
unsigned char h_dest[ETH_ALEN]; /* destination eth addr */
struct net_device *out_dev;
--
2.11.0
^ permalink raw reply related
* [net-next 05/12] net/mlx5e: Maximize ip tunnel key usage on the TC offloading path
From: Saeed Mahameed @ 2017-01-24 20:16 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, Or Gerlitz, Saeed Mahameed
In-Reply-To: <20170124201652.2920-1-saeedm@mellanox.com>
From: Or Gerlitz <ogerlitz@mellanox.com>
Use more fields out of the tunnel key (e.g the tunnel source IP address)
provided by upper layers for the route lookup done on the encap offload path.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Hadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 15 ++++++---------
1 file changed, 6 insertions(+), 9 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index 477b2796c12e..3d0dbfb018ca 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -684,7 +684,6 @@ static int mlx5e_route_lookup_ipv4(struct mlx5e_priv *priv,
struct net_device **out_dev,
struct flowi4 *fl4,
struct neighbour **out_n,
- __be32 *saddr,
int *out_ttl)
{
struct rtable *rt;
@@ -714,7 +713,6 @@ static int mlx5e_route_lookup_ipv4(struct mlx5e_priv *priv,
return -ENOMEM;
*out_n = n;
- *saddr = fl4->saddr;
*out_dev = rt->dst.dev;
return 0;
@@ -763,13 +761,10 @@ static int mlx5e_create_encap_header_ipv4(struct mlx5e_priv *priv,
{
int max_encap_size = MLX5_CAP_ESW(priv->mdev, max_encap_header_size);
struct ip_tunnel_key *tun_key = &e->tun_info.key;
+ int encap_size, ttl, err;
struct neighbour *n = NULL;
struct flowi4 fl4 = {};
char *encap_header;
- int encap_size;
- __be32 saddr;
- int ttl;
- int err;
encap_header = kzalloc(max_encap_size, GFP_KERNEL);
if (!encap_header)
@@ -784,10 +779,12 @@ static int mlx5e_create_encap_header_ipv4(struct mlx5e_priv *priv,
err = -EOPNOTSUPP;
goto out;
}
+ fl4.flowi4_tos = tun_key->tos;
fl4.daddr = tun_key->u.ipv4.dst;
+ fl4.saddr = tun_key->u.ipv4.src;
err = mlx5e_route_lookup_ipv4(priv, mirred_dev, out_dev,
- &fl4, &n, &saddr, &ttl);
+ &fl4, &n, &ttl);
if (err)
goto out;
@@ -806,8 +803,8 @@ static int mlx5e_create_encap_header_ipv4(struct mlx5e_priv *priv,
case MLX5_HEADER_TYPE_VXLAN:
encap_size = gen_vxlan_header_ipv4(*out_dev, encap_header,
e->h_dest, ttl,
- tun_key->u.ipv4.dst,
- saddr, tun_key->tp_dst,
+ fl4.daddr,
+ fl4.saddr, tun_key->tp_dst,
tunnel_id_to_key32(tun_key->tun_id));
break;
default:
--
2.11.0
^ permalink raw reply related
* [net-next 11/12] net/mlx5e: Reduce memory consumption on kdump kernel
From: Saeed Mahameed @ 2017-01-24 20:16 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, Or Gerlitz, Kamal Heib, Saeed Mahameed
In-Reply-To: <20170124201652.2920-1-saeedm@mellanox.com>
From: Kamal Heib <kamalh@mellanox.com>
Reduce memory consumption on kdump kernel by decreasing the number of
channels to 1 and the size of RQs and SQs to the minimal values.
Signed-off-by: Kamal Heib <kamalh@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en.h | 7 +------
.../net/ethernet/mellanox/mlx5/core/en_ethtool.c | 4 ++--
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 21 ++++++++++++++++++---
3 files changed, 21 insertions(+), 11 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index d603a30ad639..8d660a018bf8 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -101,6 +101,7 @@
#define MLX5E_LOG_INDIR_RQT_SIZE 0x7
#define MLX5E_INDIR_RQT_SIZE BIT(MLX5E_LOG_INDIR_RQT_SIZE)
+#define MLX5E_MIN_NUM_CHANNELS 0x1
#define MLX5E_MAX_NUM_CHANNELS (MLX5E_INDIR_RQT_SIZE >> 1)
#define MLX5E_MAX_NUM_SQS (MLX5E_MAX_NUM_CHANNELS * MLX5E_MAX_NUM_TC)
#define MLX5E_TX_CQ_POLL_BUDGET 128
@@ -847,12 +848,6 @@ static inline u32 mlx5e_get_wqe_mtt_offset(struct mlx5e_rq *rq, u16 wqe_ix)
return wqe_ix * ALIGN(MLX5_MPWRQ_PAGES_PER_WQE, 8);
}
-static inline int mlx5e_get_max_num_channels(struct mlx5_core_dev *mdev)
-{
- return min_t(int, mdev->priv.eq_table.num_comp_vectors,
- MLX5E_MAX_NUM_CHANNELS);
-}
-
extern const struct ethtool_ops mlx5e_ethtool_ops;
#ifdef CONFIG_MLX5_CORE_EN_DCB
extern const struct dcbnl_rtnl_ops mlx5e_dcbnl_ops;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
index 4021863e3840..9d520f8d6d1a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
@@ -552,7 +552,7 @@ static void mlx5e_get_channels(struct net_device *dev,
{
struct mlx5e_priv *priv = netdev_priv(dev);
- ch->max_combined = mlx5e_get_max_num_channels(priv->mdev);
+ ch->max_combined = priv->profile->max_nch(priv->mdev);
ch->combined_count = priv->params.num_channels;
}
@@ -560,7 +560,7 @@ static int mlx5e_set_channels(struct net_device *dev,
struct ethtool_channels *ch)
{
struct mlx5e_priv *priv = netdev_priv(dev);
- int ncv = mlx5e_get_max_num_channels(priv->mdev);
+ int ncv = priv->profile->max_nch(priv->mdev);
unsigned int count = ch->combined_count;
bool arfs_enabled;
bool was_opened;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 636372873e64..e829143efc14 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -31,6 +31,7 @@
*/
#include <net/tc_act/tc_gact.h>
+#include <linux/crash_dump.h>
#include <net/pkt_cls.h>
#include <linux/mlx5/fs.h>
#include <net/vxlan.h>
@@ -83,7 +84,9 @@ static void mlx5e_set_rq_type_params(struct mlx5e_priv *priv, u8 rq_type)
priv->params.rq_wq_type = rq_type;
switch (priv->params.rq_wq_type) {
case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
- priv->params.log_rq_size = MLX5E_PARAMS_DEFAULT_LOG_RQ_SIZE_MPW;
+ priv->params.log_rq_size = is_kdump_kernel() ?
+ MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE_MPW :
+ MLX5E_PARAMS_DEFAULT_LOG_RQ_SIZE_MPW;
priv->params.mpwqe_log_stride_sz =
MLX5E_GET_PFLAG(priv, MLX5E_PFLAG_RX_CQE_COMPRESS) ?
MLX5_MPWRQ_LOG_STRIDE_SIZE_CQE_COMPRESS :
@@ -92,7 +95,9 @@ static void mlx5e_set_rq_type_params(struct mlx5e_priv *priv, u8 rq_type)
priv->params.mpwqe_log_stride_sz;
break;
default: /* MLX5_WQ_TYPE_LINKED_LIST */
- priv->params.log_rq_size = MLX5E_PARAMS_DEFAULT_LOG_RQ_SIZE;
+ priv->params.log_rq_size = is_kdump_kernel() ?
+ MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE :
+ MLX5E_PARAMS_DEFAULT_LOG_RQ_SIZE;
}
priv->params.min_rx_wqes = mlx5_min_rx_wqes(priv->params.rq_wq_type,
BIT(priv->params.log_rq_size));
@@ -1508,6 +1513,14 @@ static int mlx5e_set_tx_maxrate(struct net_device *dev, int index, u32 rate)
return err;
}
+static inline int mlx5e_get_max_num_channels(struct mlx5_core_dev *mdev)
+{
+ return is_kdump_kernel() ?
+ MLX5E_MIN_NUM_CHANNELS :
+ min_t(int, mdev->priv.eq_table.num_comp_vectors,
+ MLX5E_MAX_NUM_CHANNELS);
+}
+
static int mlx5e_open_channel(struct mlx5e_priv *priv, int ix,
struct mlx5e_channel_param *cparam,
struct mlx5e_channel **cp)
@@ -3491,7 +3504,9 @@ static void mlx5e_build_nic_netdev_priv(struct mlx5_core_dev *mdev,
priv->params.lro_timeout =
mlx5e_choose_lro_timeout(mdev, MLX5E_DEFAULT_LRO_TIMEOUT);
- priv->params.log_sq_size = MLX5E_PARAMS_DEFAULT_LOG_SQ_SIZE;
+ priv->params.log_sq_size = is_kdump_kernel() ?
+ MLX5E_PARAMS_MINIMUM_LOG_SQ_SIZE :
+ MLX5E_PARAMS_DEFAULT_LOG_SQ_SIZE;
/* set CQE compression */
priv->params.rx_cqe_compress_def = false;
--
2.11.0
^ permalink raw reply related
* [net-next 09/12] net/mlx5: Push min-inline mode resolution helper into the core
From: Saeed Mahameed @ 2017-01-24 20:16 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, Or Gerlitz, Saeed Mahameed
In-Reply-To: <20170124201652.2920-1-saeedm@mellanox.com>
From: Or Gerlitz <ogerlitz@mellanox.com>
So we can use that from the IB driver too in downstream patches.
This patch doesn't change any functionality.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 18 +-----------------
drivers/net/ethernet/mellanox/mlx5/core/vport.c | 17 +++++++++++++++++
include/linux/mlx5/vport.h | 1 +
3 files changed, 19 insertions(+), 17 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index c819d07fbdb3..636372873e64 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -3458,22 +3458,6 @@ void mlx5e_set_rx_cq_mode_params(struct mlx5e_params *params, u8 cq_period_mode)
MLX5E_PARAMS_DEFAULT_RX_CQ_MODERATION_USEC_FROM_CQE;
}
-static void mlx5e_query_min_inline(struct mlx5_core_dev *mdev,
- u8 *min_inline_mode)
-{
- switch (MLX5_CAP_ETH(mdev, wqe_inline_mode)) {
- case MLX5_CAP_INLINE_MODE_L2:
- *min_inline_mode = MLX5_INLINE_MODE_L2;
- break;
- case MLX5_CAP_INLINE_MODE_VPORT_CONTEXT:
- mlx5_query_nic_vport_min_inline(mdev, 0, min_inline_mode);
- break;
- case MLX5_CAP_INLINE_MODE_NOT_REQUIRED:
- *min_inline_mode = MLX5_INLINE_MODE_NONE;
- break;
- }
-}
-
u32 mlx5e_choose_lro_timeout(struct mlx5_core_dev *mdev, u32 wanted_timeout)
{
int i;
@@ -3533,7 +3517,7 @@ static void mlx5e_build_nic_netdev_priv(struct mlx5_core_dev *mdev,
priv->params.tx_cq_moderation.pkts =
MLX5E_PARAMS_DEFAULT_TX_CQ_MODERATION_PKTS;
priv->params.tx_max_inline = mlx5e_get_max_inline_cap(mdev);
- mlx5e_query_min_inline(mdev, &priv->params.tx_min_inline_mode);
+ mlx5_query_min_inline(mdev, &priv->params.tx_min_inline_mode);
priv->params.num_tc = 1;
priv->params.rss_hfunc = ETH_RSS_HASH_XOR;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/vport.c b/drivers/net/ethernet/mellanox/mlx5/core/vport.c
index 269e4401c342..b49cfc895c10 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/vport.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/vport.c
@@ -127,6 +127,23 @@ int mlx5_query_nic_vport_min_inline(struct mlx5_core_dev *mdev,
}
EXPORT_SYMBOL_GPL(mlx5_query_nic_vport_min_inline);
+void mlx5_query_min_inline(struct mlx5_core_dev *mdev,
+ u8 *min_inline_mode)
+{
+ switch (MLX5_CAP_ETH(mdev, wqe_inline_mode)) {
+ case MLX5_CAP_INLINE_MODE_L2:
+ *min_inline_mode = MLX5_INLINE_MODE_L2;
+ break;
+ case MLX5_CAP_INLINE_MODE_VPORT_CONTEXT:
+ mlx5_query_nic_vport_min_inline(mdev, 0, min_inline_mode);
+ break;
+ case MLX5_CAP_INLINE_MODE_NOT_REQUIRED:
+ *min_inline_mode = MLX5_INLINE_MODE_NONE;
+ break;
+ }
+}
+EXPORT_SYMBOL_GPL(mlx5_query_min_inline);
+
int mlx5_modify_nic_vport_min_inline(struct mlx5_core_dev *mdev,
u16 vport, u8 min_inline)
{
diff --git a/include/linux/mlx5/vport.h b/include/linux/mlx5/vport.h
index ec35157ea725..656c70b65dd2 100644
--- a/include/linux/mlx5/vport.h
+++ b/include/linux/mlx5/vport.h
@@ -51,6 +51,7 @@ int mlx5_query_nic_vport_mac_address(struct mlx5_core_dev *mdev,
u16 vport, u8 *addr);
int mlx5_query_nic_vport_min_inline(struct mlx5_core_dev *mdev,
u16 vport, u8 *min_inline);
+void mlx5_query_min_inline(struct mlx5_core_dev *mdev, u8 *min_inline);
int mlx5_modify_nic_vport_min_inline(struct mlx5_core_dev *mdev,
u16 vport, u8 min_inline);
int mlx5_modify_nic_vport_mac_address(struct mlx5_core_dev *dev,
--
2.11.0
^ permalink raw reply related
* [net-next 07/12] net/mlx5: E-Switch, Enlarge the FDB size for the switchdev mode
From: Saeed Mahameed @ 2017-01-24 20:16 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, Or Gerlitz, Saeed Mahameed
In-Reply-To: <20170124201652.2920-1-saeedm@mellanox.com>
From: Or Gerlitz <ogerlitz@mellanox.com>
The E-Switch FDB size was hard coded to 8k. Change it to be
min(max eswitch table size, max flow counters * num flow groups)
where the max values are read from the firmware and the number of
flow groups is hard-coded as before this change.
We don't know upfront the division of flows to group. This setup allows
each group to be of size up to the where we want to support (we mandate
pairing of flows with counters for offloading). Thus, we don't expect
multiple occurences for a group which in turn adds steering hops.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Roi Dayan <roid@mellanox.com>
Tested-by: Roi Dayan <roid@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c | 13 ++++++++-----
1 file changed, 8 insertions(+), 5 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index 03293ed1cc22..348182599294 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -402,19 +402,18 @@ static int esw_add_fdb_miss_rule(struct mlx5_eswitch *esw)
}
#define MAX_PF_SQ 256
-#define ESW_OFFLOADS_NUM_ENTRIES (1 << 13) /* 8K */
#define ESW_OFFLOADS_NUM_GROUPS 4
static int esw_create_offloads_fdb_table(struct mlx5_eswitch *esw, int nvports)
{
int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
+ int table_size, ix, esw_size, err = 0;
struct mlx5_core_dev *dev = esw->dev;
struct mlx5_flow_namespace *root_ns;
struct mlx5_flow_table *fdb = NULL;
struct mlx5_flow_group *g;
u32 *flow_group_in;
void *match_criteria;
- int table_size, ix, err = 0;
u32 flags = 0;
flow_group_in = mlx5_vzalloc(inlen);
@@ -427,15 +426,19 @@ static int esw_create_offloads_fdb_table(struct mlx5_eswitch *esw, int nvports)
goto ns_err;
}
- esw_debug(dev, "Create offloads FDB table, log_max_size(%d)\n",
- MLX5_CAP_ESW_FLOWTABLE_FDB(dev, log_max_ft_size));
+ esw_debug(dev, "Create offloads FDB table, min (max esw size(2^%d), max counters(%d)*groups(%d))\n",
+ MLX5_CAP_ESW_FLOWTABLE_FDB(dev, log_max_ft_size),
+ MLX5_CAP_GEN(dev, max_flow_counter), ESW_OFFLOADS_NUM_GROUPS);
+
+ esw_size = min_t(int, MLX5_CAP_GEN(dev, max_flow_counter) * ESW_OFFLOADS_NUM_GROUPS,
+ 1 << MLX5_CAP_ESW_FLOWTABLE_FDB(dev, log_max_ft_size));
if (MLX5_CAP_ESW_FLOWTABLE_FDB(dev, encap) &&
MLX5_CAP_ESW_FLOWTABLE_FDB(dev, decap))
flags |= MLX5_FLOW_TABLE_TUNNEL_EN;
fdb = mlx5_create_auto_grouped_flow_table(root_ns, FDB_FAST_PATH,
- ESW_OFFLOADS_NUM_ENTRIES,
+ esw_size,
ESW_OFFLOADS_NUM_GROUPS, 0,
flags);
if (IS_ERR(fdb)) {
--
2.11.0
^ permalink raw reply related
* [net-next 03/12] net/mlx5e: TC ipv4 tunnel encap offload cosmetic changes
From: Saeed Mahameed @ 2017-01-24 20:16 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, Or Gerlitz, Saeed Mahameed
In-Reply-To: <20170124201652.2920-1-saeedm@mellanox.com>
From: Or Gerlitz <ogerlitz@mellanox.com>
Move around some settings of variables as pre-step to make things
more robust and clear for the ipv6 case in down-stream patch.
This patch doesn't change any functionality.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Hadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 10 ++++------
1 file changed, 4 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index e9f0854e9509..a515444d2489 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -689,7 +689,6 @@ static int mlx5e_route_lookup_ipv4(struct mlx5e_priv *priv,
{
struct rtable *rt;
struct neighbour *n = NULL;
- int ttl;
#if IS_ENABLED(CONFIG_INET)
int ret;
@@ -708,7 +707,7 @@ static int mlx5e_route_lookup_ipv4(struct mlx5e_priv *priv,
return -EOPNOTSUPP;
}
- ttl = ip4_dst_hoplimit(&rt->dst);
+ *out_ttl = ip4_dst_hoplimit(&rt->dst);
n = dst_neigh_lookup(&rt->dst, &fl4->daddr);
ip_rt_put(rt);
if (!n)
@@ -716,7 +715,6 @@ static int mlx5e_route_lookup_ipv4(struct mlx5e_priv *priv,
*out_n = n;
*saddr = fl4->saddr;
- *out_ttl = ttl;
*out_dev = rt->dst.dev;
return 0;
@@ -792,15 +790,15 @@ static int mlx5e_create_encap_header_ipv4(struct mlx5e_priv *priv,
if (err)
goto out;
- e->n = n;
- e->out_dev = *out_dev;
-
if (!(n->nud_state & NUD_VALID)) {
pr_warn("%s: can't offload, neighbour to %pI4 invalid\n", __func__, &fl4.daddr);
err = -EOPNOTSUPP;
goto out;
}
+ e->n = n;
+ e->out_dev = *out_dev;
+
neigh_ha_snapshot(e->h_dest, n, *out_dev);
switch (e->tunnel_type) {
--
2.11.0
^ permalink raw reply related
* [net-next 06/12] net/mlx5e: Support SRIOV TC encapsulation offloads for IPv6 tunnels
From: Saeed Mahameed @ 2017-01-24 20:16 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, Or Gerlitz, Saeed Mahameed
In-Reply-To: <20170124201652.2920-1-saeedm@mellanox.com>
From: Or Gerlitz <ogerlitz@mellanox.com>
Add the missing parts for offloading IPv6 tunnels. This includes
route and neigh lookups and construnction of the IPv6 tunnel headers.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Hadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 159 ++++++++++++++++++++++--
1 file changed, 151 insertions(+), 8 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index 3d0dbfb018ca..640f10f2e994 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -718,6 +718,47 @@ static int mlx5e_route_lookup_ipv4(struct mlx5e_priv *priv,
return 0;
}
+static int mlx5e_route_lookup_ipv6(struct mlx5e_priv *priv,
+ struct net_device *mirred_dev,
+ struct net_device **out_dev,
+ struct flowi6 *fl6,
+ struct neighbour **out_n,
+ int *out_ttl)
+{
+ struct neighbour *n = NULL;
+ struct dst_entry *dst;
+
+#if IS_ENABLED(CONFIG_INET) && IS_ENABLED(CONFIG_IPV6)
+ struct mlx5_eswitch *esw = priv->mdev->priv.eswitch;
+ int ret;
+
+ dst = ip6_route_output(dev_net(mirred_dev), NULL, fl6);
+ if (dst->error) {
+ ret = dst->error;
+ dst_release(dst);
+ return ret;
+ }
+
+ *out_ttl = ip6_dst_hoplimit(dst);
+
+ /* if the egress device isn't on the same HW e-switch, we use the uplink */
+ if (!switchdev_port_same_parent_id(priv->netdev, dst->dev))
+ *out_dev = mlx5_eswitch_get_uplink_netdev(esw);
+ else
+ *out_dev = dst->dev;
+#else
+ return -EOPNOTSUPP;
+#endif
+
+ n = dst_neigh_lookup(dst, &fl6->daddr);
+ dst_release(dst);
+ if (!n)
+ return -ENOMEM;
+
+ *out_n = n;
+ return 0;
+}
+
static int gen_vxlan_header_ipv4(struct net_device *out_dev,
char buf[],
unsigned char h_dest[ETH_ALEN],
@@ -754,6 +795,41 @@ static int gen_vxlan_header_ipv4(struct net_device *out_dev,
return encap_size;
}
+static int gen_vxlan_header_ipv6(struct net_device *out_dev,
+ char buf[],
+ unsigned char h_dest[ETH_ALEN],
+ int ttl,
+ struct in6_addr *daddr,
+ struct in6_addr *saddr,
+ __be16 udp_dst_port,
+ __be32 vx_vni)
+{
+ int encap_size = VXLAN_HLEN + sizeof(struct ipv6hdr) + ETH_HLEN;
+ struct ethhdr *eth = (struct ethhdr *)buf;
+ struct ipv6hdr *ip6h = (struct ipv6hdr *)((char *)eth + sizeof(struct ethhdr));
+ struct udphdr *udp = (struct udphdr *)((char *)ip6h + sizeof(struct ipv6hdr));
+ struct vxlanhdr *vxh = (struct vxlanhdr *)((char *)udp + sizeof(struct udphdr));
+
+ memset(buf, 0, encap_size);
+
+ ether_addr_copy(eth->h_dest, h_dest);
+ ether_addr_copy(eth->h_source, out_dev->dev_addr);
+ eth->h_proto = htons(ETH_P_IPV6);
+
+ ip6_flow_hdr(ip6h, 0, 0);
+ /* the HW fills up ipv6 payload len */
+ ip6h->nexthdr = IPPROTO_UDP;
+ ip6h->hop_limit = ttl;
+ ip6h->daddr = *daddr;
+ ip6h->saddr = *saddr;
+
+ udp->dest = udp_dst_port;
+ vxh->vx_flags = VXLAN_HF_VNI;
+ vxh->vx_vni = vxlan_vni_field(vx_vni);
+
+ return encap_size;
+}
+
static int mlx5e_create_encap_header_ipv4(struct mlx5e_priv *priv,
struct net_device *mirred_dev,
struct mlx5_encap_entry *e,
@@ -821,6 +897,75 @@ static int mlx5e_create_encap_header_ipv4(struct mlx5e_priv *priv,
return err;
}
+static int mlx5e_create_encap_header_ipv6(struct mlx5e_priv *priv,
+ struct net_device *mirred_dev,
+ struct mlx5_encap_entry *e,
+ struct net_device **out_dev)
+
+{
+ int max_encap_size = MLX5_CAP_ESW(priv->mdev, max_encap_header_size);
+ struct ip_tunnel_key *tun_key = &e->tun_info.key;
+ int encap_size, err, ttl = 0;
+ struct neighbour *n = NULL;
+ struct flowi6 fl6 = {};
+ char *encap_header;
+
+ encap_header = kzalloc(max_encap_size, GFP_KERNEL);
+ if (!encap_header)
+ return -ENOMEM;
+
+ switch (e->tunnel_type) {
+ case MLX5_HEADER_TYPE_VXLAN:
+ fl6.flowi6_proto = IPPROTO_UDP;
+ fl6.fl6_dport = tun_key->tp_dst;
+ break;
+ default:
+ err = -EOPNOTSUPP;
+ goto out;
+ }
+
+ fl6.flowlabel = ip6_make_flowinfo(RT_TOS(tun_key->tos), tun_key->label);
+ fl6.daddr = tun_key->u.ipv6.dst;
+ fl6.saddr = tun_key->u.ipv6.src;
+
+ err = mlx5e_route_lookup_ipv6(priv, mirred_dev, out_dev,
+ &fl6, &n, &ttl);
+ if (err)
+ goto out;
+
+ if (!(n->nud_state & NUD_VALID)) {
+ pr_warn("%s: can't offload, neighbour to %pI6 invalid\n", __func__, &fl6.daddr);
+ err = -EOPNOTSUPP;
+ goto out;
+ }
+
+ e->n = n;
+ e->out_dev = *out_dev;
+
+ neigh_ha_snapshot(e->h_dest, n, *out_dev);
+
+ switch (e->tunnel_type) {
+ case MLX5_HEADER_TYPE_VXLAN:
+ encap_size = gen_vxlan_header_ipv6(*out_dev, encap_header,
+ e->h_dest, ttl,
+ &fl6.daddr,
+ &fl6.saddr, tun_key->tp_dst,
+ tunnel_id_to_key32(tun_key->tun_id));
+ break;
+ default:
+ err = -EOPNOTSUPP;
+ goto out;
+ }
+
+ err = mlx5_encap_alloc(priv->mdev, e->tunnel_type,
+ encap_size, encap_header, &e->encap_id);
+out:
+ if (err && n)
+ neigh_release(n);
+ kfree(encap_header);
+ return err;
+}
+
static int mlx5e_attach_encap(struct mlx5e_priv *priv,
struct ip_tunnel_info *tun_info,
struct net_device *mirred_dev,
@@ -831,7 +976,7 @@ static int mlx5e_attach_encap(struct mlx5e_priv *priv,
struct ip_tunnel_key *key = &tun_info->key;
struct mlx5_encap_entry *e;
struct net_device *out_dev;
- int tunnel_type, err;
+ int tunnel_type, err = -EOPNOTSUPP;
uintptr_t hash_key;
bool found = false;
@@ -856,12 +1001,6 @@ static int mlx5e_attach_encap(struct mlx5e_priv *priv,
return -EOPNOTSUPP;
}
- if (family == AF_INET6) {
- netdev_warn(priv->netdev,
- "IPv6 tunnel encap offload isn't supported\n");
- return -EOPNOTSUPP;
- }
-
hash_key = hash_encap_info(key);
hash_for_each_possible_rcu(esw->offloads.encap_tbl, e,
@@ -885,7 +1024,11 @@ static int mlx5e_attach_encap(struct mlx5e_priv *priv,
e->tunnel_type = tunnel_type;
INIT_LIST_HEAD(&e->flows);
- err = mlx5e_create_encap_header_ipv4(priv, mirred_dev, e, &out_dev);
+ if (family == AF_INET)
+ err = mlx5e_create_encap_header_ipv4(priv, mirred_dev, e, &out_dev);
+ else if (family == AF_INET6)
+ err = mlx5e_create_encap_header_ipv6(priv, mirred_dev, e, &out_dev);
+
if (err)
goto out_err;
--
2.11.0
^ permalink raw reply related
* [net-next 02/12] net/mlx5e: Add TC offloads matching on IPv6 encapsulation headers
From: Saeed Mahameed @ 2017-01-24 20:16 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, Or Gerlitz, Saeed Mahameed
In-Reply-To: <20170124201652.2920-1-saeedm@mellanox.com>
From: Or Gerlitz <ogerlitz@mellanox.com>
Enhance the parsing of offloaded TC rules to set HW matching on outer
IPv6 encapsulation headers. This effectively adds support for TC tunnel
key release action (decapsulation) of SRIOV offloads over IPv6 tunnels.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Hadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 30 ++++++++++++++++++++++---
1 file changed, 27 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
index d4af5507679f..e9f0854e9509 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
@@ -298,6 +298,32 @@ static int parse_tunnel_attr(struct mlx5e_priv *priv,
MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c, ethertype);
MLX5_SET(fte_match_set_lyr_2_4, headers_v, ethertype, ETH_P_IP);
+ } else if (enc_control->addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS) {
+ struct flow_dissector_key_ipv6_addrs *key =
+ skb_flow_dissector_target(f->dissector,
+ FLOW_DISSECTOR_KEY_ENC_IPV6_ADDRS,
+ f->key);
+ struct flow_dissector_key_ipv6_addrs *mask =
+ skb_flow_dissector_target(f->dissector,
+ FLOW_DISSECTOR_KEY_ENC_IPV6_ADDRS,
+ f->mask);
+
+ memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+ src_ipv4_src_ipv6.ipv6_layout.ipv6),
+ &mask->src, MLX5_FLD_SZ_BYTES(ipv6_layout, ipv6));
+ memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
+ src_ipv4_src_ipv6.ipv6_layout.ipv6),
+ &key->src, MLX5_FLD_SZ_BYTES(ipv6_layout, ipv6));
+
+ memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_c,
+ dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+ &mask->dst, MLX5_FLD_SZ_BYTES(ipv6_layout, ipv6));
+ memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, headers_v,
+ dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
+ &key->dst, MLX5_FLD_SZ_BYTES(ipv6_layout, ipv6));
+
+ MLX5_SET_TO_ONES(fte_match_set_lyr_2_4, headers_c, ethertype);
+ MLX5_SET(fte_match_set_lyr_2_4, headers_v, ethertype, ETH_P_IPV6);
}
/* Enforce DMAC when offloading incoming tunneled flows.
@@ -358,12 +384,10 @@ static int __parse_cls_flower(struct mlx5e_priv *priv,
f->key);
switch (key->addr_type) {
case FLOW_DISSECTOR_KEY_IPV4_ADDRS:
+ case FLOW_DISSECTOR_KEY_IPV6_ADDRS:
if (parse_tunnel_attr(priv, spec, f))
return -EOPNOTSUPP;
break;
- case FLOW_DISSECTOR_KEY_IPV6_ADDRS:
- netdev_warn(priv->netdev,
- "IPv6 tunnel decap offload isn't supported\n");
default:
return -EOPNOTSUPP;
}
--
2.11.0
^ permalink raw reply related
* [net-next 12/12] net/mlx5e: CQE compression control code reuse
From: Saeed Mahameed @ 2017-01-24 20:16 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, Or Gerlitz, Shaker Daibes, Saeed Mahameed
In-Reply-To: <20170124201652.2920-1-saeedm@mellanox.com>
From: Shaker Daibes <shakerd@mellanox.com>
This patch is intended for code reuse of mlx5e_modify_rx_cqe_compression
function.
Signed-off-by: Shaker Daibes <shakerd@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en.h | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/en_clock.c | 7 +++++--
drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c | 13 ++-----------
drivers/net/ethernet/mellanox/mlx5/core/en_rx.c | 8 ++------
4 files changed, 10 insertions(+), 20 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index 8d660a018bf8..46f728de9e76 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -787,7 +787,7 @@ void mlx5e_pps_event_handler(struct mlx5e_priv *priv,
struct ptp_clock_event *event);
int mlx5e_hwstamp_set(struct net_device *dev, struct ifreq *ifr);
int mlx5e_hwstamp_get(struct net_device *dev, struct ifreq *ifr);
-void mlx5e_modify_rx_cqe_compression(struct mlx5e_priv *priv, bool val);
+void mlx5e_modify_rx_cqe_compression_locked(struct mlx5e_priv *priv, bool val);
int mlx5e_vlan_rx_add_vid(struct net_device *dev, __always_unused __be16 proto,
u16 vid);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_clock.c b/drivers/net/ethernet/mellanox/mlx5/core/en_clock.c
index 349dc72cd2b6..37e66eef6fb5 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_clock.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_clock.c
@@ -106,11 +106,12 @@ int mlx5e_hwstamp_set(struct net_device *dev, struct ifreq *ifr)
return -ERANGE;
}
+ mutex_lock(&priv->state_lock);
/* RX HW timestamp */
switch (config.rx_filter) {
case HWTSTAMP_FILTER_NONE:
/* Reset CQE compression to Admin default */
- mlx5e_modify_rx_cqe_compression(priv, priv->params.rx_cqe_compress_def);
+ mlx5e_modify_rx_cqe_compression_locked(priv, priv->params.rx_cqe_compress_def);
break;
case HWTSTAMP_FILTER_ALL:
case HWTSTAMP_FILTER_SOME:
@@ -128,14 +129,16 @@ int mlx5e_hwstamp_set(struct net_device *dev, struct ifreq *ifr)
case HWTSTAMP_FILTER_PTP_V2_DELAY_REQ:
/* Disable CQE compression */
netdev_warn(dev, "Disabling cqe compression");
- mlx5e_modify_rx_cqe_compression(priv, false);
+ mlx5e_modify_rx_cqe_compression_locked(priv, false);
config.rx_filter = HWTSTAMP_FILTER_ALL;
break;
default:
+ mutex_unlock(&priv->state_lock);
return -ERANGE;
}
memcpy(&priv->tstamp.hwtstamp_config, &config, sizeof(config));
+ mutex_unlock(&priv->state_lock);
return copy_to_user(ifr->ifr_data, &config,
sizeof(config)) ? -EFAULT : 0;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
index 9d520f8d6d1a..6c1a5cb43f8c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
@@ -1476,8 +1476,6 @@ static int set_pflag_rx_cqe_compress(struct net_device *netdev,
{
struct mlx5e_priv *priv = netdev_priv(netdev);
struct mlx5_core_dev *mdev = priv->mdev;
- int err = 0;
- bool reset;
if (!MLX5_CAP_GEN(mdev, cqe_compression))
return -ENOTSUPP;
@@ -1487,17 +1485,10 @@ static int set_pflag_rx_cqe_compress(struct net_device *netdev,
return -EINVAL;
}
- reset = test_bit(MLX5E_STATE_OPENED, &priv->state);
-
- if (reset)
- mlx5e_close_locked(netdev);
-
- MLX5E_SET_PFLAG(priv, MLX5E_PFLAG_RX_CQE_COMPRESS, enable);
+ mlx5e_modify_rx_cqe_compression_locked(priv, enable);
priv->params.rx_cqe_compress_def = enable;
- if (reset)
- err = mlx5e_open_locked(netdev);
- return err;
+ return 0;
}
static int mlx5e_handle_pflag(struct net_device *netdev,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
index 20f116f8c457..ba50583ea3ed 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
@@ -155,17 +155,15 @@ static inline u32 mlx5e_decompress_cqes_start(struct mlx5e_rq *rq,
return mlx5e_decompress_cqes_cont(rq, cq, 1, budget_rem) - 1;
}
-void mlx5e_modify_rx_cqe_compression(struct mlx5e_priv *priv, bool val)
+void mlx5e_modify_rx_cqe_compression_locked(struct mlx5e_priv *priv, bool val)
{
bool was_opened;
if (!MLX5_CAP_GEN(priv->mdev, cqe_compression))
return;
- mutex_lock(&priv->state_lock);
-
if (MLX5E_GET_PFLAG(priv, MLX5E_PFLAG_RX_CQE_COMPRESS) == val)
- goto unlock;
+ return;
was_opened = test_bit(MLX5E_STATE_OPENED, &priv->state);
if (was_opened)
@@ -176,8 +174,6 @@ void mlx5e_modify_rx_cqe_compression(struct mlx5e_priv *priv, bool val)
if (was_opened)
mlx5e_open_locked(priv->netdev);
-unlock:
- mutex_unlock(&priv->state_lock);
}
#define RQ_PAGE_SIZE(rq) ((1 << rq->buff.page_order) << PAGE_SHIFT)
--
2.11.0
^ permalink raw reply related
* [net-next 01/12] net/mlx5: Use exact encap header size for the FW input buffer
From: Saeed Mahameed @ 2017-01-24 20:16 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev, Or Gerlitz, Saeed Mahameed
In-Reply-To: <20170124201652.2920-1-saeedm@mellanox.com>
From: Or Gerlitz <ogerlitz@mellanox.com>
The current code is allocating the max encap size supported by
the firmware and not the size requested by the caller, fix that.
Also, spare a warning when the size of the encapsulation headers
is bigger from what is supported by the firmware.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Hadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c | 7 +++++--
1 file changed, 5 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
index c4478ecd8056..b5253b59e8fe 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_cmd.c
@@ -473,10 +473,13 @@ int mlx5_encap_alloc(struct mlx5_core_dev *dev,
int err;
u32 *in;
- if (size > MLX5_CAP_ESW(dev, max_encap_header_size))
+ if (size > max_encap_size) {
+ mlx5_core_warn(dev, "encap size %zd too big, max supported is %d\n",
+ size, max_encap_size);
return -EINVAL;
+ }
- in = kzalloc(MLX5_ST_SZ_BYTES(alloc_encap_header_in) + max_encap_size,
+ in = kzalloc(MLX5_ST_SZ_BYTES(alloc_encap_header_in) + size,
GFP_KERNEL);
if (!in)
return -ENOMEM;
--
2.11.0
^ permalink raw reply related
* Re: [PATCH] ibmveth: Add a proper check for the availability of the checksum features
From: David Miller @ 2017-01-24 20:15 UTC (permalink / raw)
To: thuth; +Cc: netdev, tlfalcon, linuxppc-dev
In-Reply-To: <1485239321-14947-1-git-send-email-thuth@redhat.com>
From: Thomas Huth <thuth@redhat.com>
Date: Tue, 24 Jan 2017 07:28:41 +0100
> When using the ibmveth driver in a KVM/QEMU based VM, it currently
> always prints out a scary error message like this when it is started:
>
> ibmveth 71000003 (unregistered net_device): unable to change
> checksum offload settings. 1 rc=-2 ret_attr=71000003
>
> This happens because the driver always tries to enable the checksum
> offloading without checking for the availability of this feature first.
> QEMU does not support checksum offloading for the spapr-vlan device,
> thus we always get the error message here.
> According to the LoPAPR specification, the "ibm,illan-options" property
> of the corresponding device tree node should be checked first to see
> whether the H_ILLAN_ATTRIUBTES hypercall and thus the checksum offloading
> feature is available. Thus let's do this in the ibmveth driver, too, so
> that the error message is really only limited to cases where something
> goes wrong, and does not occur if the feature is just missing.
>
> Signed-off-by: Thomas Huth <thuth@redhat.com>
Applied.
^ 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