OSDN Git Service

tomoyo/tomoyo-test1.git
4 years agoBluetooth: During le_conn_timeout disable EXT_ADV
Sathish Narsimman [Mon, 24 Feb 2020 05:23:40 +0000 (10:53 +0530)]
Bluetooth: During le_conn_timeout disable EXT_ADV

Disabling LE_LEGACY_ADV when LE_EXT_ADV is enabled causes 'command
disallowed . This patch fixes that issue and disables EXT_ADV if
enabled.

Signed-off-by: Sathish Narsimman <sathish.narasimman@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: Remove adv set for directed advertising
Sathish Narsimman [Mon, 24 Feb 2020 05:32:24 +0000 (11:02 +0530)]
Bluetooth: Remove adv set for directed advertising

Extended advertising Data is set during bluetooth initialization
by default which causes InvalidHCICommandParameters when setting
Extended advertising parameters.

As per Core Spec 5.2 Vol 2, PART E, Sec 7.8.53, for
advertising_event_property LE_LEGACY_ADV_DIRECT_IND does not
supports advertising data when the advertising set already
contains some, the controller shall return erroc code
'InvalidHCICommandParameters(0x12).

So it is required to remove adv set for handle 0x00. since we use
instance 0 for directed adv.

Signed-off-by: Sathish Narsimman <sathish.narasimman@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: Use list_for_each_entry_rcu() to traverse RCU list in RCU read-side CS
Madhuparna Bhowmik [Tue, 25 Feb 2020 13:17:53 +0000 (18:47 +0530)]
Bluetooth: Use list_for_each_entry_rcu() to traverse RCU list in RCU read-side CS

In function hci_is_blocked_key() RCU list is traversed with
list_for_each_entry() in RCU read-side CS.
Use list_for_each_entry_rcu() instead.

Signed-off-by: Madhuparna Bhowmik <madhuparnabhowmik10@gmail.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: Fix Suspicious RCU usage warnings
Madhuparna Bhowmik [Tue, 25 Feb 2020 13:08:09 +0000 (18:38 +0530)]
Bluetooth: Fix Suspicious RCU usage warnings

The following functions in hci_core are always called with
hdev->lock held. No need to use list_for_each_entry_rcu(), therefore
change the usage of list_for_each_entry_rcu() in these functions
to list_for_each_entry().

hci_link_keys_clear()
hci_smp_ltks_clear()
hci_smp_irks_clear()
hci_blocked_keys_clear()

Warning encountered with CONFIG_PROVE_RCU_LIST:

[   72.213184] =============================
[   72.213188] WARNING: suspicious RCU usage
[   72.213192] 5.6.0-rc1+ #5 Not tainted
[   72.213195] -----------------------------
[   72.213198] net/bluetooth/hci_core.c:2288 RCU-list traversed in non-reader section!!

[   72.213676] =============================
[   72.213679] WARNING: suspicious RCU usage
[   72.213683] 5.6.0-rc1+ #5 Not tainted
[   72.213685] -----------------------------
[   72.213689] net/bluetooth/hci_core.c:2298 RCU-list traversed in non-reader section!!

[   72.214195] =============================
[   72.214198] WARNING: suspicious RCU usage
[   72.214201] 5.6.0-rc1+ #5 Not tainted
[   72.214204] -----------------------------
[   72.214208] net/bluetooth/hci_core.c:2308 RCU-list traversed in non-reader section!!

[  333.456972] =============================
[  333.456979] WARNING: suspicious RCU usage
[  333.457001] 5.6.0-rc1+ #5 Not tainted
[  333.457007] -----------------------------
[  333.457014] net/bluetooth/hci_core.c:2318 RCU-list traversed in non-reader section!!

Signed-off-by: Madhuparna Bhowmik <madhuparnabhowmik10@gmail.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: Replace zero-length array with flexible-array member
Gustavo A. R. Silva [Wed, 26 Feb 2020 23:02:27 +0000 (17:02 -0600)]
Bluetooth: Replace zero-length array with flexible-array member

The current codebase makes use of the zero-length array language
extension to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:

struct foo {
        int stuff;
        struct boo array[];
};

By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on.

Also, notice that, dynamic memory allocations won't be affected by
this change:

"Flexible array members have incomplete type, and so the sizeof operator
may not be applied. As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]

This issue was found with the help of Coccinelle.

[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 76497732932f ("cxgb3/l2t: Fix undefined behaviour")

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: fix passkey uninitialized when used
Howard Chung [Thu, 20 Feb 2020 03:17:29 +0000 (11:17 +0800)]
Bluetooth: fix passkey uninitialized when used

This patch fix the issue: warning:variable 'passkey' is uninitialized
when used here

Link: https://groups.google.com/forum/#!topic/clang-built-linux/kyRKCjRsGoU
Fixes: cee5f20fece3 ("Bluetooth: secure bluetooth stack from bluedump attack")
Reported-by: kbuild test robot <lkp@intel.com>
Suggested-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Howard Chung <howardchung@google.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: RFCOMM: Use MTU auto tune logic
Luiz Augusto von Dentz [Thu, 20 Feb 2020 05:31:55 +0000 (21:31 -0800)]
Bluetooth: RFCOMM: Use MTU auto tune logic

This reuse the L2CAP MTU auto logic to select the MTU used for RFCOMM
channels, this should increase the maximum from 1013 to 1021 when 3-DH5
is supported.

Since it does not set an L2CAP MTU we no longer need a debugfs so that
is removed.

Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: Fix crash when using new BT_PHY option
Luiz Augusto von Dentz [Tue, 18 Feb 2020 18:33:20 +0000 (10:33 -0800)]
Bluetooth: Fix crash when using new BT_PHY option

This fixes the invalid check for connected socket which causes the
following trace due to sco_pi(sk)->conn being NULL:

RIP: 0010:sco_sock_getsockopt+0x2ff/0x800 net/bluetooth/sco.c:966

L2CAP has also been fixed since it has the same problem.

Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: hci_h5: Move variable into local scope
Marcel Holtmann [Tue, 18 Feb 2020 08:28:58 +0000 (09:28 +0100)]
Bluetooth: hci_h5: Move variable into local scope

The variable was declared in an unnecessarily broad scope.

Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
4 years agoBluetooth: hci_qca: Bug fixes while collecting controller memory dump
Venkata Lakshmi Narayana Gubba [Fri, 14 Feb 2020 16:17:15 +0000 (21:47 +0530)]
Bluetooth: hci_qca: Bug fixes while collecting controller memory dump

This patch will fix the below issues
 1. Discarding memory dump events if memdump state is moved to
    MEMDUMP_TIMEOUT.
 2. Fixed race conditions between qca_hw_error() and qca_controller_memdump
    while free memory dump buffers using mutex lock
 3. Moved timeout timer to delayed work queue
 4. Injecting HW error event in a case when dumps failed to receive and HW
    error event is not yet received.
 5. Clearing hw error and command timeout function callbacks before
    sending pre shutdown command.

 Collecting memory dump will follow any of the below sequence.

 Sequence 1:
   Receiving Memory dump events from the controller
   Received entire dump in stipulated time
   Received HW error event from the controller
   Controller Reset from HOST

 Sequence 2:
   Receiving Memory dump events from the controller
   Failed to Receive entire dump in stipulated time
   A Timeout schedules and if no HW error event received a fake HW
     error event will be injected.
   Controller Reset from HOST.

 Sequence 3:
   Received HW error event
   HOST trigger SSR by sending crash packet to controller.
   Received entire dump in stipulated time
   Controller Reset from HOST

Fixes: d841502c79e3 ("Bluetooth: hci_qca: Collect controller memory dump during SSR")
Reported-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Signed-off-by: Venkata Lakshmi Narayana Gubba <gubbaven@codeaurora.org>
Reviewed-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: Disable Extended Adv if enabled
Sathish Narsimman [Mon, 17 Feb 2020 09:07:44 +0000 (14:37 +0530)]
Bluetooth: Disable Extended Adv if enabled

Disabling LEGACY_ADV when EXT_ADV is enabled causes
'command disallowed' during DIRECTED_ADV. This Patch fixes this
issue.

Signed-off-by: Sathish Narsimman <sathish.narasimman@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: hci_h5: btrtl: Add support for RTL8822C
Max Chou [Mon, 17 Feb 2020 08:14:55 +0000 (16:14 +0800)]
Bluetooth: hci_h5: btrtl: Add support for RTL8822C

Add new compatible and FW loading support for RTL8822C.

Signed-off-by: Max Chou <max.chou@realtek.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: Fix a typo in Kconfig
Christophe JAILLET [Sun, 16 Feb 2020 11:20:31 +0000 (12:20 +0100)]
Bluetooth: Fix a typo in Kconfig

'internface' has an extra 'n'.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: Add BT_PHY socket option
Luiz Augusto von Dentz [Fri, 14 Feb 2020 18:08:57 +0000 (10:08 -0800)]
Bluetooth: Add BT_PHY socket option

This adds BT_PHY socket option (read-only) which can be used to read
the PHYs in use by the underline connection.

Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: secure bluetooth stack from bluedump attack
Howard Chung [Fri, 14 Feb 2020 11:16:41 +0000 (19:16 +0800)]
Bluetooth: secure bluetooth stack from bluedump attack

Attack scenario:
1. A Chromebook (let's call this device A) is paired to a legitimate
   Bluetooth classic device (e.g. a speaker) (let's call this device
   B).
2. A malicious device (let's call this device C) pretends to be the
   Bluetooth speaker by using the same BT address.
3. If device A is not currently connected to device B, device A will
   be ready to accept connection from device B in the background
   (technically, doing Page Scan).
4. Therefore, device C can initiate connection to device A
   (because device A is doing Page Scan) and device A will accept the
   connection because device A trusts device C's address which is the
   same as device B's address.
5. Device C won't be able to communicate at any high level Bluetooth
   profile with device A because device A enforces that device C is
   encrypted with their common Link Key, which device C doesn't have.
   But device C can initiate pairing with device A with just-works
   model without requiring user interaction (there is only pairing
   notification). After pairing, device A now trusts device C with a
   new different link key, common between device A and C.
6. From now on, device A trusts device C, so device C can at anytime
   connect to device A to do any kind of high-level hijacking, e.g.
   speaker hijack or mouse/keyboard hijack.

Since we don't know whether the repairing is legitimate or not,
leave the decision to user space if all the conditions below are met.
- the pairing is initialized by peer
- the authorization method is just-work
- host already had the link key to the peer

Signed-off-by: Howard Chung <howardchung@google.com>
Acked-by: Johan Hedberg <johan.hedberg@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: hci_uart: Replace zero-length array with flexible-array member
Gustavo A. R. Silva [Wed, 12 Feb 2020 19:30:19 +0000 (13:30 -0600)]
Bluetooth: hci_uart: Replace zero-length array with flexible-array member

The current codebase makes use of the zero-length array language
extension to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:

struct foo {
        int stuff;
        struct boo array[];
};

By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertenly introduced[3] to the codebase from now on.

Also, notice that, dynamic memory allocations won't be affected by
this change:

"Flexible array members have incomplete type, and so the sizeof operator
may not be applied. As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]

This issue was found with the help of Coccinelle.

[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 76497732932f ("cxgb3/l2t: Fix undefined behaviour")

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: hci_intel: Replace zero-length array with flexible-array member
Gustavo A. R. Silva [Wed, 12 Feb 2020 19:31:19 +0000 (13:31 -0600)]
Bluetooth: hci_intel: Replace zero-length array with flexible-array member

The current codebase makes use of the zero-length array language
extension to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:

struct foo {
        int stuff;
        struct boo array[];
};

By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertenly introduced[3] to the codebase from now on.

Also, notice that, dynamic memory allocations won't be affected by
this change:

"Flexible array members have incomplete type, and so the sizeof operator
may not be applied. As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]

This issue was found with the help of Coccinelle.

[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 76497732932f ("cxgb3/l2t: Fix undefined behaviour")

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: btintel: Replace zero-length array with flexible-array member
Gustavo A. R. Silva [Wed, 12 Feb 2020 19:28:07 +0000 (13:28 -0600)]
Bluetooth: btintel: Replace zero-length array with flexible-array member

The current codebase makes use of the zero-length array language
extension to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:

struct foo {
        int stuff;
        struct boo array[];
};

By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertenly introduced[3] to the codebase from now on.

Also, notice that, dynamic memory allocations won't be affected by
this change:

"Flexible array members have incomplete type, and so the sizeof operator
may not be applied. As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]

This issue was found with the help of Coccinelle.

[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 76497732932f ("cxgb3/l2t: Fix undefined behaviour")

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: btusb: Add support for 13d3:3548 Realtek 8822CE device
Sergey Shatunov [Sat, 8 Feb 2020 15:53:15 +0000 (23:53 +0800)]
Bluetooth: btusb: Add support for 13d3:3548 Realtek 8822CE device

The ASUS FX505DV laptop contains RTL8822CE device with an
associated BT chip using a USB ID of 13d3:3548.
This patch add fw download support for it.

T:  Bus=03 Lev=01 Prnt=01 Port=03 Cnt=03 Dev#=  4 Spd=12   MxCh= 0
D:  Ver= 1.00 Cls=e0(wlcon) Sub=01 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=13d3 ProdID=3548 Rev= 0.00
S:  Manufacturer=Realtek
S:  Product=Bluetooth Radio
S:  SerialNumber=00e04c000001
C:* #Ifs= 2 Cfg#= 1 Atr=a0 MxPwr=500mA
I:* If#= 0 Alt= 0 #EPs= 3 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=81(I) Atr=03(Int.) MxPS=  16 Ivl=1ms
E:  Ad=02(O) Atr=02(Bulk) MxPS=  64 Ivl=0ms
E:  Ad=82(I) Atr=02(Bulk) MxPS=  64 Ivl=0ms
I:* If#= 1 Alt= 0 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=   0 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=   0 Ivl=1ms
I:  If#= 1 Alt= 1 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=   9 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=   9 Ivl=1ms
I:  If#= 1 Alt= 2 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=  17 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=  17 Ivl=1ms
I:  If#= 1 Alt= 3 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=  25 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=  25 Ivl=1ms
I:  If#= 1 Alt= 4 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=  33 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=  33 Ivl=1ms
I:  If#= 1 Alt= 5 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=  49 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=  49 Ivl=1ms

Signed-off-by: Sergey Shatunov <me@prok.pw>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: hci_qca: Optimized code while enabling clocks for BT SOC
Venkata Lakshmi Narayana Gubba [Wed, 5 Feb 2020 10:51:43 +0000 (16:21 +0530)]
Bluetooth: hci_qca: Optimized code while enabling clocks for BT SOC

* Directly passing clock pointer to clock code without checking for NULL
  as clock code takes care of it
* Removed the comment which was not necessary
* Updated code for return in qca_regulator_enable()

Signed-off-by: Venkata Lakshmi Narayana Gubba <gubbaven@codeaurora.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: remove __get_channel/dir and __dir
Alex Shi [Wed, 5 Feb 2020 03:23:27 +0000 (11:23 +0800)]
Bluetooth: remove __get_channel/dir and __dir

These 3 macros are never used from first git commit Linux-2.6.12-rc2.
let's remove them.

Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: prefetch channel before killing sock
Hillf Danton [Wed, 5 Feb 2020 02:31:59 +0000 (10:31 +0800)]
Bluetooth: prefetch channel before killing sock

Prefetch channel before killing sock in order to fix UAF like

 BUG: KASAN: use-after-free in l2cap_sock_release+0x24c/0x290 net/bluetooth/l2cap_sock.c:1212
 Read of size 8 at addr ffff8880944904a0 by task syz-fuzzer/9751

Reported-by: syzbot+c3c5bdea7863886115dc@syzkaller.appspotmail.com
Fixes: 6c08fc896b60 ("Bluetooth: Fix refcount use-after-free issue")
Cc: Manish Mandlik <mmandlik@google.com>
Signed-off-by: Hillf Danton <hdanton@sina.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agodt-bindings: net: bluetooth: Add device tree bindings for QTI chip WCN3991
Venkata Lakshmi Narayana Gubba [Mon, 3 Feb 2020 10:40:41 +0000 (16:10 +0530)]
dt-bindings: net: bluetooth: Add device tree bindings for QTI chip WCN3991

Add compatible string for the Qualcomm WCN3991 Bluetooth controller

Signed-off-by: Venkata Lakshmi Narayana Gubba <gubbaven@codeaurora.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: hci_qca: Enable clocks required for BT SOC
Venkata Lakshmi Narayana Gubba [Mon, 3 Feb 2020 10:40:40 +0000 (16:10 +0530)]
Bluetooth: hci_qca: Enable clocks required for BT SOC

Instead of relying on other subsytem to turn ON clocks
required for BT SoC to operate, voting them from the driver.

Signed-off-by: Venkata Lakshmi Narayana Gubba <gubbaven@codeaurora.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: optimize barrier usage for Rmw atomics
Davidlohr Bueso [Wed, 29 Jan 2020 18:10:41 +0000 (10:10 -0800)]
Bluetooth: optimize barrier usage for Rmw atomics

Use smp_mb__before_atomic() instead of smp_mb() and avoid the
unnecessary barrier for non LL/SC architectures, such as x86.

Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: Fix refcount use-after-free issue
Manish Mandlik [Tue, 28 Jan 2020 18:54:14 +0000 (10:54 -0800)]
Bluetooth: Fix refcount use-after-free issue

There is no lock preventing both l2cap_sock_release() and
chan->ops->close() from running at the same time.

If we consider Thread A running l2cap_chan_timeout() and Thread B running
l2cap_sock_release(), expected behavior is:
  A::l2cap_chan_timeout()->l2cap_chan_close()->l2cap_sock_teardown_cb()
  A::l2cap_chan_timeout()->l2cap_sock_close_cb()->l2cap_sock_kill()
  B::l2cap_sock_release()->sock_orphan()
  B::l2cap_sock_release()->l2cap_sock_kill()

where,
sock_orphan() clears "sk->sk_socket" and l2cap_sock_teardown_cb() marks
socket as SOCK_ZAPPED.

In l2cap_sock_kill(), there is an "if-statement" that checks if both
sock_orphan() and sock_teardown() has been run i.e. sk->sk_socket is NULL
and socket is marked as SOCK_ZAPPED. Socket is killed if the condition is
satisfied.

In the race condition, following occurs:
  A::l2cap_chan_timeout()->l2cap_chan_close()->l2cap_sock_teardown_cb()
  B::l2cap_sock_release()->sock_orphan()
  B::l2cap_sock_release()->l2cap_sock_kill()
  A::l2cap_chan_timeout()->l2cap_sock_close_cb()->l2cap_sock_kill()

In this scenario, "if-statement" is true in both B::l2cap_sock_kill() and
A::l2cap_sock_kill() and we hit "refcount: underflow; use-after-free" bug.

Similar condition occurs at other places where teardown/sock_kill is
happening:
  l2cap_disconnect_rsp()->l2cap_chan_del()->l2cap_sock_teardown_cb()
  l2cap_disconnect_rsp()->l2cap_sock_close_cb()->l2cap_sock_kill()

  l2cap_conn_del()->l2cap_chan_del()->l2cap_sock_teardown_cb()
  l2cap_conn_del()->l2cap_sock_close_cb()->l2cap_sock_kill()

  l2cap_disconnect_req()->l2cap_chan_del()->l2cap_sock_teardown_cb()
  l2cap_disconnect_req()->l2cap_sock_close_cb()->l2cap_sock_kill()

  l2cap_sock_cleanup_listen()->l2cap_chan_close()->l2cap_sock_teardown_cb()
  l2cap_sock_cleanup_listen()->l2cap_sock_kill()

Protect teardown/sock_kill and orphan/sock_kill by adding hold_lock on
l2cap channel to ensure that the socket is killed only after marked as
zapped and orphan.

Signed-off-by: Manish Mandlik <mmandlik@google.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: SMP: Fix SALT value in some comments
Christophe JAILLET [Mon, 27 Jan 2020 22:36:09 +0000 (23:36 +0100)]
Bluetooth: SMP: Fix SALT value in some comments

Salts are 16 bytes long.
Remove some extra and erroneous '0' in the human readable format used
in comments.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoMerge branch 'qed-Utilize-FW-8.42.2.0'
David S. Miller [Mon, 27 Jan 2020 13:35:40 +0000 (14:35 +0100)]
Merge branch 'qed-Utilize-FW-8.42.2.0'

Michal Kalderon says:

====================
qed*: Utilize FW 8.42.2.0

This FW contains several fixes and features, main ones listed below.
We have taken into consideration past comments on previous FW versions
that were uploaded and tried to separate this one to smaller patches to
ease review.

- RoCE
- SRIOV support
- Fixes in following flows:
- latency optimization flow for inline WQEs
- iwarp OOO packed DDPs flow
- tx-dif workaround calculations flow
- XRC-SRQ exceed cache num

- iSCSI
- Fixes:
- iSCSI TCP out-of-order handling.
- iscsi retransmit flow

- Fcoe
- Fixes:
- upload + cleanup flows

- Debug
- Better handling of extracting data during traffic
- ILT Dump -> dumping host memory used by chip
- MDUMP -> collect debug data on system crash and extract after
  reboot

Patches prefixed with FW 8.42.2.0 are required to work with binary
8.42.2.0 FW where as the rest are FW related but do not require the
binary.

Changes from V2
---------------
- Move FW version to the start of the series to maintain minimal compatibility
- Fix some kbuild errors:
- frame size larger than 1024 (Queue Manager patch - remove redundant
  field from struct)
- sparse warning on endianity (Dmae patch fix - wrong use of __le32 for field
  used only on host, should be u32)
- static should be used for some functions (Debug feature ilt and mdump)

Reported-by: kbuild test robot <lkp@intel.com>
Changes from V1
---------------
- Remove epoch + kernel version from device debug dump
- don't bump driver version
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoqed: FW 8.42.2.0 debug features
Michal Kalderon [Mon, 27 Jan 2020 13:26:19 +0000 (15:26 +0200)]
qed: FW 8.42.2.0 debug features

Add to debug dump more information on the platform it was collected
from (pci func, path id).
Provide human readable reg fifo erros.

Removed static debug arrays from HSI Functions, and move them to
the hwfn.

Some structures were slightly changed (removing reserved chip id
for example) which lead to many long initializations being modified
with one parameter less during initialization. This leads to
some long diffs that don't really change anything.

Signed-off-by: Ariel Elior <ariel.elior@marvell.com>
Signed-off-by: Michal Kalderon <michal.kalderon@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoqed: rt init valid initialization changed
Michal Kalderon [Mon, 27 Jan 2020 13:26:18 +0000 (15:26 +0200)]
qed: rt init valid initialization changed

The QM phase init tool can be invoked multiple times during
the driver lifetime. Part of the init comes from the runtime array.
The logic for setting the values did not init all values, basically
assuming the runtime array was all zeroes. But if it was invoked
multiple times, nobody was zeroing it after the first time.

In this change we zero the runtime array right after using it.

Signed-off-by: Ariel Elior <ariel.elior@marvell.com>
Signed-off-by: Michal Kalderon <michal.kalderon@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoqed: Debug feature: ilt and mdump
Michal Kalderon [Mon, 27 Jan 2020 13:26:17 +0000 (15:26 +0200)]
qed: Debug feature: ilt and mdump

Part of the FW drop includes new debug capabilities implemented in the
qed_debug file. This patch dumps additional information during ethtool -d
for better debugging. The data dumped is the ilt (internal logical table)
and information gathered by the management firmware incase there was a
crash and driver was not able to extract the information (mdump).

Signed-off-by: Ariel Elior <ariel.elior@marvell.com>
Signed-off-by: Michal Kalderon <michal.kalderon@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoqed: FW 8.42.2.0 Add fw overlay feature
Michal Kalderon [Mon, 27 Jan 2020 13:26:16 +0000 (15:26 +0200)]
qed: FW 8.42.2.0 Add fw overlay feature

This feature enables the FW to page out FW code when required

Signed-off-by: Ariel Elior <ariel.elior@marvell.com>
Signed-off-by: Michal Kalderon <michal.kalderon@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoqed: FW 8.42.2.0 HSI changes
Michal Kalderon [Mon, 27 Jan 2020 13:26:15 +0000 (15:26 +0200)]
qed: FW 8.42.2.0 HSI changes

This patch contains several HSI changes. The changes are part of
features like RDMA VF and OVS, the patch also contains a fix to
how the init code determines if the dmae is ready to be used.

Signed-off-by: Ariel Elior <ariel.elior@marvell.com>
Signed-off-by: Michal Kalderon <michal.kalderon@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoqed: FW 8.42.2.0 iscsi/fcoe changes
Michal Kalderon [Mon, 27 Jan 2020 13:26:14 +0000 (15:26 +0200)]
qed: FW 8.42.2.0 iscsi/fcoe changes

- Remove struct iscsi_slow_path_hdr and field fw_cid from several structs
- Remove struct iscsi_spe_func_dstry
- Remove fields pbe_page_size_log and pbl_page_size_log from struct
  iscsi_conn_offload_param

Signed-off-by: Manish Rangankar <manish.rangankar@marvell.com>
Signed-off-by: Saurav Kashyap <saurav.kashyap@marvell.com>
Signed-off-by: Ariel Elior <ariel.elior@marvell.com>
Signed-off-by: Michal Kalderon <michal.kalderon@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoqed: Add abstraction for different hsi values per chip
Michal Kalderon [Mon, 27 Jan 2020 13:26:13 +0000 (15:26 +0200)]
qed: Add abstraction for different hsi values per chip

The number of BTB blocks was modified to be different between the two chip
flavors supported (BB/K2) as a result, this lead to a re-write of selecting
the default hsi value based on the chip.
This patch creates a lookup table for hsi values per chip rather than
ask again and again for every value.

Signed-off-by: Ariel Elior <ariel.elior@marvell.com>
Signed-off-by: Michal Kalderon <michal.kalderon@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoqed: FW 8.42.2.0 Additional ll2 type
Michal Kalderon [Mon, 27 Jan 2020 13:26:12 +0000 (15:26 +0200)]
qed: FW 8.42.2.0 Additional ll2 type

LL2 queues were a limited resource due to FW constraints.
This FW introduced a new resource which is a context based ll2 queue
(memory on host). The additional ll2 queues are required for RDMA SRIOV.
The code refers to the previous ll2 queues as ram-based or legacy, and the
new queues as ctx-based.
This change decreased the "legacy" ram-based queues therefore the first ll2
queue used for iWARP was converted to the ctx-based ll2 queue.
This feature also exposed a bug in the DIRECT_REG_WR64 macro implementation
which didn't have an effect in other use cases.

Signed-off-by: Ariel Elior <ariel.elior@marvell.com>
Signed-off-by: Michal Kalderon <michal.kalderon@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoqed: Use dmae to write to widebus registers in fw_funcs
Michal Kalderon [Mon, 27 Jan 2020 13:26:11 +0000 (15:26 +0200)]
qed: Use dmae to write to widebus registers in fw_funcs

There are several wide-bus registers written to by the fw_funcs
that require using the dmae for atomicity. Therefore using the dmae
channel functionality was added to the fw_funcs file, since the code
is very similar to the previously used code, the structures used were
moved to qed_hsi. Due to FW conventions, the names of the flags in the
struct changed. Since this required slight modification in the places
that set the flags the code was modified to use GET/SET FIELD macros.

Signed-off-by: Ariel Elior <ariel.elior@marvell.com>
Signed-off-by: Michal Kalderon <michal.kalderon@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoqed: FW 8.42.2.0 Parser offsets modified
Michal Kalderon [Mon, 27 Jan 2020 13:26:10 +0000 (15:26 +0200)]
qed: FW 8.42.2.0 Parser offsets modified

Convert storm ram line to regpair rather than two distinct u32
to better represent the u64 width of the ram.
Convert some defines to be hex instead of negative values
these values also changed by FW from previous value.

Signed-off-by: Ariel Elior <ariel.elior@marvell.com>
Signed-off-by: Michal Kalderon <michal.kalderon@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoqed: FW 8.42.2.0 Queue Manager changes
Michal Kalderon [Mon, 27 Jan 2020 13:26:09 +0000 (15:26 +0200)]
qed: FW 8.42.2.0 Queue Manager changes

This patch contains changes in initialization and usage of the QM blocks.
Instead of setting a rate limiter per vport the rate limiters are now a
global resource and set independentaly.

The patch also contains a field name change:
vport_wfq which is part of vport_params was renamed to wfq as the vport
prefix is redundant.

Signed-off-by: Ariel Elior <ariel.elior@marvell.com>
Signed-off-by: Michal Kalderon <michal.kalderon@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoqed: FW 8.42.2.0 Expose new registers and change windows
Michal Kalderon [Mon, 27 Jan 2020 13:26:08 +0000 (15:26 +0200)]
qed: FW 8.42.2.0 Expose new registers and change windows

This patch contains register initialization related changes.
- Modifications to the runtime offsets - these are defines used
  by the driver or firmware functions to set values that are used
  by the initialization functions to set device register values.
- Global window values changes to provide different device register
  ranges.
- Additional device registers addresses were added to the register file,
  used in later stages.

Signed-off-by: Ariel Elior <ariel.elior@marvell.com>
Signed-off-by: Michal Kalderon <michal.kalderon@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoqed: FW 8.42.2.0 Internal ram offsets modifications
Michal Kalderon [Mon, 27 Jan 2020 13:26:07 +0000 (15:26 +0200)]
qed: FW 8.42.2.0 Internal ram offsets modifications

IRO stands for internal RAM offsets. Updating the FW binary produces
different iro offsets. This file contains the different values,
and a new representation of the values.
Update the FW version

Signed-off-by: Ariel Elior <ariel.elior@marvell.com>
Signed-off-by: Michal Kalderon <michal.kalderon@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'octeontx2-pf-Add-network-driver-for-physical-function'
David S. Miller [Mon, 27 Jan 2020 13:33:40 +0000 (14:33 +0100)]
Merge branch 'octeontx2-pf-Add-network-driver-for-physical-function'

Sunil Goutham says:

====================
octeontx2-pf: Add network driver for physical function

OcteonTX2 SOC's resource virtualization unit (RVU) supports
multiple physical and virtual functions. Each of the PF's
functionality is determined by what kind of resources are attached
to it. If NPA and NIX blocks are attached to a PF it can function
as a highly capable network device.

This patch series add a network driver for the PF. Initial set of
patches adds mailbox communication with admin function (RVU AF)
and configuration of queues. Followed by Rx and tx pkts NAPI
handler and then support for HW offloads like RSS, TSO, Rxhash etc.
Ethtool support to extract stats, config RSS, queue sizes, queue
count is also added.

Added documentation to give a high level overview of HW and
different drivers which will be upstreamed and how they interact.

Changes from v5:
   * Fixed otx2_atomic64_add() non ARM64 fallback definition.
     - Suggested by David Miller

Changes from v4:
   * Replaced pci_set_dma_mask and pci_set_consistent_dma_mask
     fn()s with dma_set_mask_and_coherent().
   * Some additonal code cleanup.
   * Fixed receive buffer segmnetation logic in otx2_alloc_rbuf()
   * Removed all unused BIG_ENDIAN structure definitions.
   * Removed unnecessary memory barriers
     - Sugested by Jakub Kicinski
   * Fixed mailbox initalization failure handling
   * Removed unused function parameter in otx2_skb_add_frag()
     - Suggested by Maciej Fijalkowski

Changes from v3:
   * Fixed receive side scaling reinitialization during interface
     DOWN and UP to retain user configured settings, if any.
   * Removed driver version from ethtool.
   * Fixed otx2_set_rss_hash_opts() to return error incase RSS is
     not enabled.
     - Sugested by Jakub Kicinski

Changes from v2:
   * Removed frames, bytes, dropped packet stats from ethtool to avoid
     duplication of same stats in netlink and ethtool.
     - Sugested by Jakub Kicinski
   * Removed number of channels and ringparam upper bound checking
     in ethtool support.
   * Fixed RSS hash option setting to reject unsupported config.
     - Suggested by Michal Kubecek

Changes from v1:
   * Made driver dependent on 64bit, to fix build errors related to
     non availability of writeq/readq APIs for 32bit platforms.
     - Reported by kbuild test robot
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMAINTAINERS: Add entry for Marvell OcteonTX2 Physical Function driver
Sunil Goutham [Mon, 27 Jan 2020 13:05:31 +0000 (18:35 +0530)]
MAINTAINERS: Add entry for Marvell OcteonTX2 Physical Function driver

Added maintainers entry for Marvell OcteonTX2 SOC's physical
function NIC driver.

Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoDocumentation: net: octeontx2: Add RVU HW and drivers overview
Sunil Goutham [Mon, 27 Jan 2020 13:05:30 +0000 (18:35 +0530)]
Documentation: net: octeontx2: Add RVU HW and drivers overview

Added high level overview of OcteonTx2 RVU HW and functionality of
various drivers which will be upstreamed.

Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-pf: ethtool RSS config support
Sunil Goutham [Mon, 27 Jan 2020 13:05:29 +0000 (18:35 +0530)]
octeontx2-pf: ethtool RSS config support

Added support to show or configure RSS hash key, indirection table,
2,4 tuple via ethtool. Also added debug msg_level support
to dump messages when HW reports errors in packet received
or transmitted.

Signed-off-by: Prakash Brahmajyosyula <bprakash@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-pf: Add basic ethtool support
Christina Jacob [Mon, 27 Jan 2020 13:05:28 +0000 (18:35 +0530)]
octeontx2-pf: Add basic ethtool support

This patch adds ethtool support for
 - Driver stats, Tx/Rx perqueue and CGX LMAC stats
 - Set/show Rx/Tx queue count
 - Set/show Rx/Tx ring sizes
 - Set/show IRQ coalescing parameters

Signed-off-by: Christina Jacob <cjacob@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-pf: Add ndo_get_stats64
Geetha sowjanya [Mon, 27 Jan 2020 13:05:27 +0000 (18:35 +0530)]
octeontx2-pf: Add ndo_get_stats64

Added ndo_get_stats64 which returns stats maintained by HW.

Signed-off-by: Geetha sowjanya <gakula@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-pf: TCP segmentation offload support
Sunil Goutham [Mon, 27 Jan 2020 13:05:26 +0000 (18:35 +0530)]
octeontx2-pf: TCP segmentation offload support

Adds TCP segmentation offload (TSO) support. First version
of the silicon didn't support TSO offload, for this driver
level TSO support is added.

Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-pf: Receive side scaling support
Sunil Goutham [Mon, 27 Jan 2020 13:05:25 +0000 (18:35 +0530)]
octeontx2-pf: Receive side scaling support

Adds receive side scaling (RSS) support to distribute
pkts/flows across multiple queues. Sets up key, indirection
table etc. Also added extraction of HW calculated rxhash and
adding to same to SKB ie NETIF_F_RXHASH offload support.

Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-pf: Error handling support
Geetha sowjanya [Mon, 27 Jan 2020 13:05:24 +0000 (18:35 +0530)]
octeontx2-pf: Error handling support

HW reports many errors on the receive and transmit paths.
Such as incorrect queue configuration, pkt transmission errors,
LMTST instruction errors, transmit queue full etc. These are reported
via QINT interrupt. Most of the errors are fatal and needs
reinitialization.

Also added support to allocate receive buffers in non-atomic context
when allocation fails in NAPI context.

Signed-off-by: Geetha sowjanya <gakula@marvell.com>
Signed-off-by: Aleksey Makarov <amakarov@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-pf: MTU, MAC and RX mode config support
Sunil Goutham [Mon, 27 Jan 2020 13:05:23 +0000 (18:35 +0530)]
octeontx2-pf: MTU, MAC and RX mode config support

This patch addes support to change interface MTU, MAC address
retrieval and config, RX mode ie unicast, multicast and promiscuous.
Also added link loopback support

Signed-off-by: Tomasz Duszynski <tduszynski@marvell.com>
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-pf: Register and handle link notifications
Linu Cherian [Mon, 27 Jan 2020 13:05:22 +0000 (18:35 +0530)]
octeontx2-pf: Register and handle link notifications

PF and AF (admin function) shares 64KB of reserved memory region for
communication. This region is shared for
 - Messages sent by PF and responses sent by AF.
 - Notifications sent by AF and ACKs sent by PF.

This patch adds infrastructure to handle notifications sent
by AF and adds handlers to process them.

One of the main usecase of notifications from AF is physical
link changes. So this patch adds registration of PF with AF
to receive link status change notifications and also adds
the handler for that notification.

Signed-off-by: Linu Cherian <lcherian@marvell.com>
Signed-off-by: Tomasz Duszynski <tduszynski@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-pf: Add packet transmission support
Sunil Goutham [Mon, 27 Jan 2020 13:05:21 +0000 (18:35 +0530)]
octeontx2-pf: Add packet transmission support

This patch adds the packet transmission support.
For a given skb prepares send queue descriptors (SQEs) and pushes them
to HW. Here driver doesn't maintain it's own SQ rings, SQEs are pushed
to HW using a silicon specific operations called LMTST. From the
instuction HW derives the transmit queue number and queues the SQE to
that queue. These LMTST instructions are designed to avoid queue
maintenance in SW and lockless behavior ie when multiple cores are trying
to add SQEs to same queue then HW will takecare of serialization, no need
for SW to hold locks.

Also supports scatter/gather.

Co-developed-by: Geetha sowjanya <gakula@marvell.com>
Signed-off-by: Geetha sowjanya <gakula@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-pf: Receive packet handling support
Sunil Goutham [Mon, 27 Jan 2020 13:05:20 +0000 (18:35 +0530)]
octeontx2-pf: Receive packet handling support

Added receive packet handling (NAPI) support, error stats, RX_ALL
capability config option to passon error pkts to stack upon user request.

In subsequent patches these error stats will be added to ethttool.

Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-pf: Setup interrupts and NAPI handler
Sunil Goutham [Mon, 27 Jan 2020 13:05:19 +0000 (18:35 +0530)]
octeontx2-pf: Setup interrupts and NAPI handler

Completion queue (CQ) is the one with which HW notifies SW on a packet
reception or transmission. Each of the RQ and SQ are mapped to a unique
CQ and again both CQs are mapped to same interrupt ie the CINT. So that
each core has one interrupt source in whose handler both Rx and Tx
notifications are processed.

Also
- Registered a NAPI handler for the CINT.
- Setup coalescing parameters.
- IRQ affinity hints etc

Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-pf: Initialize and config queues
Sunil Goutham [Mon, 27 Jan 2020 13:05:18 +0000 (18:35 +0530)]
octeontx2-pf: Initialize and config queues

This patch does the initialization of all queues ie the
receive buffer pools, receive and transmit queues, completion
or notification queues etc. Allocates all required resources
(eg transmit schedulers, receive buffers etc) and configures
them for proper functioning of queues. Also sets up receive
queue's RED dropping levels.

Co-developed-by: Geetha sowjanya <gakula@marvell.com>
Signed-off-by: Geetha sowjanya <gakula@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-pf: Attach NIX and NPA block LFs
Sunil Goutham [Mon, 27 Jan 2020 13:05:17 +0000 (18:35 +0530)]
octeontx2-pf: Attach NIX and NPA block LFs

For a PF to function as a NIC, NPA (for Rx buffers, Tx descriptors etc)
and NIX (for rcv, send and completion queues) are the minimum resources
needed. So request admin function (AF) to attach one each of NIX and NPA
block LFs (local functions).

Only AF can configure a LF's contexts, so request AF to allocate memory
for NPA aura/pool and NIX RQ/SQ/CQ HW contexts. Upon receiving response,
save some of the HW constants like number of pointers per stack page,
size of send queue buffer (SQBs, where SQEs are queued by HW) e.t.c which
are later used to initialize queues.

A HW context here is like a state machine maintained for a descriptor
queue. eg size, head/tail pointers, irq etc etc. HW maintains this in
memory.

Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-pf: Mailbox communication with AF
Sunil Goutham [Mon, 27 Jan 2020 13:05:16 +0000 (18:35 +0530)]
octeontx2-pf: Mailbox communication with AF

In the resource virtualization unit (RVU) each of the PF and AF
(admin function) share a 64KB of reserved memory region for
communication. This patch initializes PF <=> AF mailbox IRQs,
registers handlers for processing these communication messages.
Also adds support to process these messages in both directions
ie responses to PF initiated DOWN (PF => AF) messages and AF
initiated UP messages (AF => PF).

Mbox communication APIs and message formats are defined in AF driver
(drivers/net/ethernet/marvell/octeontx2/af), mbox.h from AF driver is
included here to avoid duplication.

Signed-off-by: Geetha sowjanya <gakula@marvell.com>
Signed-off-by: Christina Jacob <cjacob@marvell.com>
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Aleksey Makarov <amakarov@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-pf: Add Marvell OcteonTX2 NIC driver
Sunil Goutham [Mon, 27 Jan 2020 13:05:15 +0000 (18:35 +0530)]
octeontx2-pf: Add Marvell OcteonTX2 NIC driver

This patch adds template for the Marvell's OcteonTX2 network
controller's physical function driver. Just the probe, PCI
specific initialization and netdev registration.

Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next
David S. Miller [Mon, 27 Jan 2020 13:31:40 +0000 (14:31 +0100)]
Merge git://git./linux/kernel/git/bpf/bpf-next

Daniel Borkmann says:

====================
pull-request: bpf-next 2020-01-27

The following pull-request contains BPF updates for your *net-next* tree.

We've added 20 non-merge commits during the last 5 day(s) which contain
a total of 24 files changed, 433 insertions(+), 104 deletions(-).

The main changes are:

1) Make BPF trampolines and dispatcher aware for the stack unwinder, from Jiri Olsa.

2) Improve handling of failed CO-RE relocations in libbpf, from Andrii Nakryiko.

3) Several fixes to BPF sockmap and reuseport selftests, from Lorenz Bauer.

4) Various cleanups in BPF devmap's XDP flush code, from John Fastabend.

5) Fix BPF flow dissector when used with port ranges, from Yoshiki Komachi.

6) Fix bpffs' map_seq_next callback to always inc position index, from Vasily Averin.

7) Allow overriding LLVM tooling for runqslower utility, from Andrey Ignatov.

8) Silence false-positive lockdep splats in devmap hash lookup, from Amol Grover.

9) Fix fentry/fexit selftests to initialize a variable before use, from John Sperbeck.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoRevert "pktgen: Allow configuration of IPv6 source address range"
David S. Miller [Mon, 27 Jan 2020 12:49:33 +0000 (13:49 +0100)]
Revert "pktgen: Allow configuration of IPv6 source address range"

This reverts commit 7786a1af2a6bceb07860ec720e74714004438834.

It causes build failures on 32-bit, for example:

   net/core/pktgen.o: In function `mod_cur_headers':
>> pktgen.c:(.text.mod_cur_headers+0xba0): undefined reference to `__umoddi3'

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet/core: Replace driver version to be kernel version
Leon Romanovsky [Mon, 27 Jan 2020 07:20:28 +0000 (09:20 +0200)]
net/core: Replace driver version to be kernel version

In order to stop useless driver version bumps and unify output
presented by ethtool -i, let's set default version string.

As Linus said in [1]: "Things are supposed to be backwards and
forwards compatible, because we don't accept breakage in user
space anyway. So versioning is pointless, and only causes
problems."

They cause problems when users start to see version changes
and expect specific set of features which will be different
for stable@, vanilla and distribution kernels.

Distribution kernels are based on some kernel version with extra
patches on top, for example, in RedHat world this "extra" is a lot
and for them your driver version say nothing. Users who run vanilla
kernels won't use driver version information too, because running
such kernels requires knowledge and understanding.

Another set of problems are related to difference in versioning scheme
and such doesn't allow to write meaningful automation which will work
sanely on all ethtool capable devices.

Before this change:
[leonro@erver ~]$ ethtool -i eth0
driver: virtio_net
version: 1.0.0
After this change and once ->version assignment will be deleted
from virtio_net:
[leonro@server ~]$ ethtool -i eth0
driver: virtio_net
version: 5.5.0-rc6+

Link: https://lore.kernel.org/ksummit-discuss/CA+55aFx9A=5cc0QZ7CySC4F2K7eYaEfzkdYEc9JaNgCcV25=rg@mail.gmail.com/
Link: https://lore.kernel.org/linux-rdma/20200122152627.14903-1-michal.kalderon@marvell.com/T/#md460ff8f976c532a89d6860411c3c50bb811038b
Link: https://lore.kernel.org/linux-rdma/20200127060835.GA570@unicorn.suse.cz
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Acked-by: Shannon Nelson <snelson@pensando.io>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'sfc-refactor-mcdi-filtering-code'
David S. Miller [Mon, 27 Jan 2020 12:05:43 +0000 (13:05 +0100)]
Merge branch 'sfc-refactor-mcdi-filtering-code'

Alex Maftei says:

====================
sfc: refactor mcdi filtering code

Splitting final bits of the driver code into different files, which
will later be used in another driver for a new product.

This is a continuation to my previous patch series. (three of them)
Refactoring will be concluded with this series, for now.

As instructed, split the renaming and moving into different patches.
Removed stray spaces before tabs... twice.
Minor refactoring was done with the renaming, as explained in the
first patch.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agosfc: move mcdi filtering code
Alex Maftei (amaftei) [Mon, 27 Jan 2020 11:13:55 +0000 (11:13 +0000)]
sfc: move mcdi filtering code

Signed-off-by: Alexandru-Mihai Maftei <amaftei@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agosfc: create header for mcdi filtering code
Alex Maftei (amaftei) [Mon, 27 Jan 2020 11:13:40 +0000 (11:13 +0000)]
sfc: create header for mcdi filtering code

Moved structs, enums, and added function prototypes.

The affected functions are no longer static.

Signed-off-by: Alexandru-Mihai Maftei <amaftei@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agosfc: rename mcdi filtering functions/structs
Alex Maftei (amaftei) [Mon, 27 Jan 2020 11:13:27 +0000 (11:13 +0000)]
sfc: rename mcdi filtering functions/structs

Minor style fixes included due to name lengths changing.

Signed-off-by: Alexandru-Mihai Maftei <amaftei@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'bnxt_en-next'
David S. Miller [Mon, 27 Jan 2020 10:33:29 +0000 (11:33 +0100)]
Merge branch 'bnxt_en-next'

Michael Chan says:

====================
bnxt_en: Updates for net-next.

This patch-set includes link up and link initialization improvements,
RSS and aRFS improvements, devlink refactoring and registration
improvements, devlink info support including documentation.

v2: Removed the TC ingress rate limiting patch. The developer Harsha needs
to rework some code.
    Use fw.psid suggested by Jakub Kicinski.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agodevlink: document devlink info versions reported by bnxt_en driver
Vasundhara Volam [Mon, 27 Jan 2020 09:56:27 +0000 (04:56 -0500)]
devlink: document devlink info versions reported by bnxt_en driver

Add the set of info versions reported by bnxt_en driver, including
a description of what the version represents, and what modes (fixed,
running, stored) it reports.

v2: Use fw.psid.

Cc: Jiri Pirko <jiri@mellanox.com>
Cc: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agobnxt_en: Add support for devlink info command
Vasundhara Volam [Mon, 27 Jan 2020 09:56:26 +0000 (04:56 -0500)]
bnxt_en: Add support for devlink info command

Display the following information via devlink info command:
  - Driver name
  - Board id
  - Broad revision
  - Board Serial number
  - Board FW version
  - FW parameter set version
  - FW App version
  - FW management version
  - FW RoCE version

  Standard output example:
  $ devlink dev info pci/0000:3b:00.0
  pci/0000:3b:00.0:
  driver bnxt_en
  serial_number 00-10-18-FF-FE-AD-05-00
  versions:
      fixed:
        asic.id D802
        asic.rev 1
      running:
        fw 216.1.124.0
        fw.psid 0.0.0
        fw.app 216.1.122.0
        fw.mgmt 864.0.32.0
        fw.roce 216.1.15.0

[ This version has incorporated changes suggested by Jakub Kicinski to
  use generic devlink version tags. ]

v2: Use fw.psid

Cc: Jiri Pirko <jiri@mellanox.com>
Cc: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agodevlink: add macro for "fw.roce"
Vasundhara Volam [Mon, 27 Jan 2020 09:56:25 +0000 (04:56 -0500)]
devlink: add macro for "fw.roce"

Add definition and documentation for the new generic info "fw.roce".

v2: Remove board.nvm_cfg since fw.psid is similar.

Cc: Jiri Pirko <jiri@mellanox.com>
Cc: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agobnxt_en: Rename switch_id to dsn
Vasundhara Volam [Mon, 27 Jan 2020 09:56:24 +0000 (04:56 -0500)]
bnxt_en: Rename switch_id to dsn

Instead of switch_id, renaming it to dsn will be more meaningful
so that it can be used to display device serial number in follow up
patch via devlink_info command.

Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agobnxt_en: Add support to update progress of flash update
Vasundhara Volam [Mon, 27 Jan 2020 09:56:23 +0000 (04:56 -0500)]
bnxt_en: Add support to update progress of flash update

This patch adds status notification to devlink flash update
while flashing is in progress.

$ devlink dev flash pci/0000:05:00.0 file 103.pkg
Preparing to flash
Flashing done

Cc: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agobnxt_en: Move devlink_register before registering netdev
Vasundhara Volam [Mon, 27 Jan 2020 09:56:22 +0000 (04:56 -0500)]
bnxt_en: Move devlink_register before registering netdev

Latest kernels get the phys_port_name via devlink, if
ndo_get_phys_port_name is not defined. To provide the phys_port_name
correctly, register devlink before registering netdev.

Also call devlink_port_type_eth_set() after registering netdev as
devlink port updates the netdev structure and notifies user.

Cc: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agobnxt_en: Register devlink irrespective of firmware spec version
Vasundhara Volam [Mon, 27 Jan 2020 09:56:21 +0000 (04:56 -0500)]
bnxt_en: Register devlink irrespective of firmware spec version

This will allow to register for devlink port and use port features.
Also register params only if firmware spec version is at least 0x10600
which will support reading/setting numbered variables in NVRAM.

Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agobnxt_en: Refactor bnxt_dl_register()
Vasundhara Volam [Mon, 27 Jan 2020 09:56:20 +0000 (04:56 -0500)]
bnxt_en: Refactor bnxt_dl_register()

Define bnxt_dl_params_register() and bnxt_dl_params_unregister()
functions and move params register/unregister code to these newly
defined functions. This patch is in preparation to register
devlink irrespective of firmware spec. version in the next patch.

Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agobnxt_en: Disable workaround for lost interrupts on 575XX B0 and newer chips.
Michael Chan [Mon, 27 Jan 2020 09:56:19 +0000 (04:56 -0500)]
bnxt_en: Disable workaround for lost interrupts on 575XX B0 and newer chips.

The hardware bug has been fixed on B0 and newer chips, so disable the
workaround on these chips.

Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agobnxt_en: Periodically check and remove aged-out ntuple filters
Pavan Chebbi [Mon, 27 Jan 2020 09:56:18 +0000 (04:56 -0500)]
bnxt_en: Periodically check and remove aged-out ntuple filters

Currently the only time we check and remove expired filters is
when we are inserting new filters.
Improving the aRFS expiry handling by adding code to do the above
work periodically.

Signed-off-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agobnxt_en: Do not accept fragments for aRFS flow steering.
Michael Chan [Mon, 27 Jan 2020 09:56:17 +0000 (04:56 -0500)]
bnxt_en: Do not accept fragments for aRFS flow steering.

In bnxt_rx_flow_steer(), if the dissected packet is a fragment, do not
proceed to create the ntuple filter and return error instead.  Otherwise
we would create a filter with 0 source and destination ports because
the dissected ports would not be available for fragments.

Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agobnxt_en: Support UDP RSS hashing on 575XX chips.
Michael Chan [Mon, 27 Jan 2020 09:56:16 +0000 (04:56 -0500)]
bnxt_en: Support UDP RSS hashing on 575XX chips.

575XX (P5) chips have the same UDP RSS hashing capability as P4 chips,
so we can enable it on P5 chips.

Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agobnxt_en: Remove the setting of dev_port.
Michael Chan [Mon, 27 Jan 2020 09:56:15 +0000 (04:56 -0500)]
bnxt_en: Remove the setting of dev_port.

The dev_port is meant to distinguish the network ports belonging to
the same PCI function.  Our devices only have one network port
associated with each PCI function and so we should not set it for
correctness.

Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agobnxt_en: Improve bnxt_probe_phy().
Michael Chan [Mon, 27 Jan 2020 09:56:14 +0000 (04:56 -0500)]
bnxt_en: Improve bnxt_probe_phy().

If the 2nd parameter fw_dflt is not set, we are calling bnxt_probe_phy()
after the firmware has reset.  There is no need to query the current
PHY settings from firmware as these settings may be different from
the ethtool settings that the driver will re-establish later.  So
return earlier in bnxt_probe_phy() to save one firmware call.

Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agobnxt_en: Improve link up detection.
Michael Chan [Mon, 27 Jan 2020 09:56:13 +0000 (04:56 -0500)]
bnxt_en: Improve link up detection.

In bnxt_update_phy_setting(), ethtool_get_link_ksettings() and
bnxt_disable_an_for_lpbk(), we inconsistently use netif_carrier_ok()
to determine link.  Instead, we should use bp->link_info.link_up
which has the true link state.  The netif_carrier state may be off
during self-test and while the device is being reset and may not always
reflect the true link state.

By always using bp->link_info.link_up, the code is now more
consistent and more correct.  Some unnecessary link toggles are
now prevented with this patch.

Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'ethtool-netlink-interface-part-2'
David S. Miller [Mon, 27 Jan 2020 10:31:36 +0000 (11:31 +0100)]
Merge branch 'ethtool-netlink-interface-part-2'

Michal Kubecek says:

====================
ethtool netlink interface, part 2

This shorter series adds support for getting and setting of wake-on-lan
settings and message mask (originally message level). Together with the
code already in net-next, this will allow full implementation of
"ethtool <dev>" and "ethtool -s <dev> ...".

Older versions of the ethtool netlink series allowed getting WoL settings
by unprivileged users and only filtered out the password but this was
a source of controversy so for now, ETHTOOL_MSG_WOL_GET request always
requires CAP_NET_ADMIN as ETHTOOL_GWOL ioctl request does.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: add WOL_NTF notification
Michal Kubecek [Sun, 26 Jan 2020 22:11:19 +0000 (23:11 +0100)]
ethtool: add WOL_NTF notification

Send ETHTOOL_MSG_WOL_NTF notification whenever wake-on-lan settings of
a device are modified using ETHTOOL_MSG_WOL_SET netlink message or
ETHTOOL_SWOL ioctl request.

As notifications can be received by anyone, do not include SecureOn(tm)
password in notification messages.

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: set wake-on-lan settings with WOL_SET request
Michal Kubecek [Sun, 26 Jan 2020 22:11:16 +0000 (23:11 +0100)]
ethtool: set wake-on-lan settings with WOL_SET request

Implement WOL_SET netlink request to set wake-on-lan settings. This is
equivalent to ETHTOOL_SWOL ioctl request.

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: provide WoL settings with WOL_GET request
Michal Kubecek [Sun, 26 Jan 2020 22:11:13 +0000 (23:11 +0100)]
ethtool: provide WoL settings with WOL_GET request

Implement WOL_GET request to get wake-on-lan settings for a device,
traditionally available via ETHTOOL_GWOL ioctl request.

As part of the implementation, provide symbolic names for wake-on-line
modes as ETH_SS_WOL_MODES string set.

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: add DEBUG_NTF notification
Michal Kubecek [Sun, 26 Jan 2020 22:11:10 +0000 (23:11 +0100)]
ethtool: add DEBUG_NTF notification

Send ETHTOOL_MSG_DEBUG_NTF notification message whenever debugging message
mask for a device are modified using ETHTOOL_MSG_DEBUG_SET netlink message
or ETHTOOL_SMSGLVL ioctl request.

The notification message has the same format as reply to DEBUG_GET request.
As with other ethtool notifications, netlink requests only trigger the
notification if the mask is actually changed while ioctl request trigger it
whenever the request results in calling the ethtool_ops handler.

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: set message mask with DEBUG_SET request
Michal Kubecek [Sun, 26 Jan 2020 22:11:07 +0000 (23:11 +0100)]
ethtool: set message mask with DEBUG_SET request

Implement DEBUG_SET netlink request to set debugging settings for a device.
At the moment, only message mask corresponding to message level as set by
ETHTOOL_SMSGLVL ioctl request can be set. (It is called message level in
ioctl interface but almost all drivers interpret it as a bit mask.)

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: provide message mask with DEBUG_GET request
Michal Kubecek [Sun, 26 Jan 2020 22:11:04 +0000 (23:11 +0100)]
ethtool: provide message mask with DEBUG_GET request

Implement DEBUG_GET request to get debugging settings for a device. At the
moment, only message mask corresponding to message level as reported by
ETHTOOL_GMSGLVL ioctl request is provided. (It is called message level in
ioctl interface but almost all drivers interpret it as a bit mask.)

As part of the implementation, provide symbolic names for message mask bits
as ETH_SS_MSG_CLASSES string set.

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoethtool: fix kernel-doc descriptions
Michal Kubecek [Sun, 26 Jan 2020 22:11:01 +0000 (23:11 +0100)]
ethtool: fix kernel-doc descriptions

Fix missing or incorrect function argument and struct member descriptions.

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge tag 'wireless-drivers-next-2020-01-26' of git://git.kernel.org/pub/scm/linux...
David S. Miller [Mon, 27 Jan 2020 10:25:36 +0000 (11:25 +0100)]
Merge tag 'wireless-drivers-next-2020-01-26' of git://git./linux/kernel/git/kvalo/wireless-drivers-next

Kalle Valo says:

====================
wireless-drivers-next patches for v5.6

Second set of patches for v5.6. Nothing special standing out, smaller
new features and fixes allover.

Major changes:

ar5523

* add support for SMCWUSBT-G2 USB device

iwlwifi

* support new versions of the FTM FW APIs

* support new version of the beacon template FW API

* print some extra information when the driver is loaded

rtw88

* support wowlan feature for 8822c

* add support for WIPHY_WOWLAN_NET_DETECT

brcmfmac

* add initial support for monitor mode

qtnfmac

* add module parameter to enable DFS offloading in firmware

* add support for STA HE rates

* add support for TWT responder and spatial reuse
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'bpf-flow-dissector-fix-port-ranges'
Daniel Borkmann [Mon, 27 Jan 2020 10:25:07 +0000 (11:25 +0100)]
Merge branch 'bpf-flow-dissector-fix-port-ranges'

Yoshiki Komachi says:

====================
When I tried a test based on the selftest program for BPF flow dissector
(test_flow_dissector.sh), I observed unexpected result as below:

$ tc filter add dev lo parent ffff: protocol ip pref 1337 flower ip_proto \
udp src_port 8-10 action drop
$ tools/testing/selftests/bpf/test_flow_dissector -i 4 -f 9 -F
inner.dest4: 127.0.0.1
inner.source4: 127.0.0.3
pkts: tx=10 rx=10

The last rx means the number of received packets. I expected rx=0 in this
test (i.e., all received packets should have been dropped), but it resulted
in acceptance.

Although the previous commit 8ffb055beae5 ("cls_flower: Fix the behavior
using port ranges with hw-offload") added new flag and field toward filtering
based on port ranges with hw-offload, it missed applying for BPF flow dissector
then. As a result, BPF flow dissector currently stores data extracted from
packets in incorrect field used for exact match whenever packets are classified
by filters based on port ranges. Thus, they never match rules in such cases
because flow dissector gives rise to generating incorrect flow keys.

This series fixes the issue by replacing incorrect flag and field with new
ones in BPF flow dissector, and adds a test for filtering based on specified
port ranges to the existing selftest program.

Changes in v2:
 - set key_ports to NULL at the top of __skb_flow_bpf_to_target()
====================

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
4 years agoselftests/bpf: Add test based on port range for BPF flow dissector
Yoshiki Komachi [Fri, 17 Jan 2020 07:05:33 +0000 (16:05 +0900)]
selftests/bpf: Add test based on port range for BPF flow dissector

Add a simple test to make sure that a filter based on specified port
range classifies packets correctly.

Signed-off-by: Yoshiki Komachi <komachi.yoshiki@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Petar Penkov <ppenkov@google.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20200117070533.402240-3-komachi.yoshiki@gmail.com
4 years agoflow_dissector: Fix to use new variables for port ranges in bpf hook
Yoshiki Komachi [Fri, 17 Jan 2020 07:05:32 +0000 (16:05 +0900)]
flow_dissector: Fix to use new variables for port ranges in bpf hook

This patch applies new flag (FLOW_DISSECTOR_KEY_PORTS_RANGE) and
field (tp_range) to BPF flow dissector to generate appropriate flow
keys when classified by specified port ranges.

Fixes: 8ffb055beae5 ("cls_flower: Fix the behavior using port ranges with hw-offload")
Signed-off-by: Yoshiki Komachi <komachi.yoshiki@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Petar Penkov <ppenkov@google.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20200117070533.402240-2-komachi.yoshiki@gmail.com
4 years agoMerge branch 'for-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetoot...
David S. Miller [Mon, 27 Jan 2020 10:24:46 +0000 (11:24 +0100)]
Merge branch 'for-upstream' of git://git./linux/kernel/git/bluetooth/bluetooth-next

Johan Hedberg says:

====================
pull request: bluetooth-next 2020-01-26

Here's (probably) the last bluetooth-next pull request for the 5.6 kernel.

 - Initial pieces of Bluetooth 5.2 Isochronous Channels support
 - mgmt: Various cleanups and a new Set Blocked Keys command
 - btusb: Added support for 04ca:3021 QCA_ROME device
 - hci_qca: Multiple fixes & cleanups
 - hci_bcm: Fixes & improved device tree support
 - Fixed attempts to create duplicate debugfs entries

Please let me know if there are any issues pulling. Thanks.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agodrivers: net: xgene: Fix the order of the arguments of 'alloc_etherdev_mqs()'
Christophe JAILLET [Sun, 26 Jan 2020 10:44:29 +0000 (11:44 +0100)]
drivers: net: xgene: Fix the order of the arguments of 'alloc_etherdev_mqs()'

'alloc_etherdev_mqs()' expects first 'tx', then 'rx'. The semantic here
looks reversed.

Reorder the arguments passed to 'alloc_etherdev_mqs()' in order to keep
the correct semantic.

In fact, this is a no-op because both XGENE_NUM_[RT]X_RING are 8.

Fixes: 107dec2749fe ("drivers: net: xgene: Add support for multiple queues")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agobpf, xdp: Remove no longer required rcu_read_{un}lock()
John Fastabend [Mon, 27 Jan 2020 00:14:02 +0000 (16:14 -0800)]
bpf, xdp: Remove no longer required rcu_read_{un}lock()

Now that we depend on rcu_call() and synchronize_rcu() to also wait
for preempt_disabled region to complete the rcu read critical section
in __dev_map_flush() is no longer required. Except in a few special
cases in drivers that need it for other reasons.

These originally ensured the map reference was safe while a map was
also being free'd. And additionally that bpf program updates via
ndo_bpf did not happen while flush updates were in flight. But flush
by new rules can only be called from preempt-disabled NAPI context.
The synchronize_rcu from the map free path and the rcu_call from the
delete path will ensure the reference there is safe. So lets remove
the rcu_read_lock and rcu_read_unlock pair to avoid any confusion
around how this is being protected.

If the rcu_read_lock was required it would mean errors in the above
logic and the original patch would also be wrong.

Now that we have done above we put the rcu_read_lock in the driver
code where it is needed in a driver dependent way. I think this
helps readability of the code so we know where and why we are
taking read locks. Most drivers will not need rcu_read_locks here
and further XDP drivers already have rcu_read_locks in their code
paths for reading xdp programs on RX side so this makes it symmetric
where we don't have half of rcu critical sections define in driver
and the other half in devmap.

Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Link: https://lore.kernel.org/bpf/1580084042-11598-4-git-send-email-john.fastabend@gmail.com
4 years agobpf, xdp: virtio_net use access ptr macro for xdp enable check
John Fastabend [Mon, 27 Jan 2020 00:14:01 +0000 (16:14 -0800)]
bpf, xdp: virtio_net use access ptr macro for xdp enable check

virtio_net currently relies on rcu critical section to access the xdp
program in its xdp_xmit handler. However, the pointer to the xdp program
is only used to do a NULL pointer comparison to determine if xdp is
enabled or not.

Use rcu_access_pointer() instead of rcu_dereference() to reflect this.
Then later when we drop rcu_read critical section virtio_net will not
need in special handling.

Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Link: https://lore.kernel.org/bpf/1580084042-11598-3-git-send-email-john.fastabend@gmail.com
4 years agobpf, xdp: Update devmap comments to reflect napi/rcu usage
John Fastabend [Mon, 27 Jan 2020 00:14:00 +0000 (16:14 -0800)]
bpf, xdp: Update devmap comments to reflect napi/rcu usage

Now that we rely on synchronize_rcu and call_rcu waiting to
exit perempt-disable regions (NAPI) lets update the comments
to reflect this.

Fixes: 0536b85239b84 ("xdp: Simplify devmap cleanup")
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Björn Töpel <bjorn.topel@intel.com>
Acked-by: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/1580084042-11598-2-git-send-email-john.fastabend@gmail.com
4 years agor8169: don't set min_mtu/max_mtu if not needed
Heiner Kallweit [Sun, 26 Jan 2020 09:40:44 +0000 (10:40 +0100)]
r8169: don't set min_mtu/max_mtu if not needed

Defaults for min_mtu and max_mtu are set by ether_setup(), which is
called from devm_alloc_etherdev(). Let rtl_jumbo_max() only return
a positive value if actually jumbo packets are supported. This also
allows to remove constant Jumbo_1K which is a little misleading anyway.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>