OSDN Git Service

uclinux-h8/linux.git
4 years agoBluetooth: btusb: USB alternate setting 1 for WBS
Hilda Wu [Tue, 30 Jun 2020 13:09:40 +0000 (21:09 +0800)]
Bluetooth: btusb: USB alternate setting 1 for WBS

RTL8822CE supports transparent WBS to apply USB alternate setting 1.
Add a flag to the device match data to apply alternate setting 1 which
meet the transfer speed for WBS support.

Signed-off-by: Hilda Wu <hildawu@realtek.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: use configured default params for active scans
Alain Michaud [Mon, 29 Jun 2020 17:04:15 +0000 (17:04 +0000)]
Bluetooth: use configured default params for active scans

This patch fixes active scans to use the configured default parameters.

Signed-off-by: Alain Michaud <alainm@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: Adding a configurable autoconnect timeout
Alain Michaud [Mon, 29 Jun 2020 16:11:00 +0000 (16:11 +0000)]
Bluetooth: Adding a configurable autoconnect timeout

This patch adds a configurable LE autoconnect timeout.

Signed-off-by: Alain Michaud <alainm@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agodt-bindings: net: bluetooth: realtek: Fix uart-has-rtscts example
Martin Blumenstingl [Mon, 29 Jun 2020 18:05:45 +0000 (20:05 +0200)]
dt-bindings: net: bluetooth: realtek: Fix uart-has-rtscts example

uart-has-rtscts is a boolean property. These are defined as present
(which means that this property evaluates to "true") or absent (which
means that this property evaluates to "false"). Remove the numeric value
from the example to make it comply with the boolean property bindings.

Fixes: 1cc2d0e021f867 ("dt-bindings: net: bluetooth: Add rtl8723bs-bluetooth")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: btusb: Comment on unbalanced pm reference
Abhishek Pandit-Subedi [Thu, 25 Jun 2020 23:26:27 +0000 (16:26 -0700)]
Bluetooth: btusb: Comment on unbalanced pm reference

Add a comment clarifying that a PM reference in btusb_qca_cmd_timeout is
not unbalanced because it results in a device reset.

Signed-off-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: btusb: Refactor of firmware download flow for Intel conrollers
Kiran K [Wed, 24 Jun 2020 18:34:32 +0000 (00:04 +0530)]
Bluetooth: btusb: Refactor of firmware download flow for Intel conrollers

Address the scalability to support new generation Intel controller with
respect to readability and enhancement to new firmware download sequence

Signed-off-by: Kiran K <kiran.k@intel.com>
Reviewed-by: Chethan T N <chethan.tumkur.narayan@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: Don't restart scanning if paused
Abhishek Pandit-Subedi [Wed, 24 Jun 2020 18:34:19 +0000 (11:34 -0700)]
Bluetooth: Don't restart scanning if paused

When restarting LE scanning, check if it's currently paused before
enabling passive scanning.

Signed-off-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: btusb: BTUSB_WAKEUP_DISABLE prevents wake
Abhishek Pandit-Subedi [Wed, 24 Jun 2020 18:24:30 +0000 (11:24 -0700)]
Bluetooth: btusb: BTUSB_WAKEUP_DISABLE prevents wake

When the BTUSB_WAKEUP_DISABLE flag is set, always return true for
prevent wake. This tells the suspend notifier not to prepare the
controller for reconnections during suspend.

Signed-off-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: btusb: Reset port on cmd timeout
Abhishek Pandit-Subedi [Wed, 24 Jun 2020 18:11:44 +0000 (11:11 -0700)]
Bluetooth: btusb: Reset port on cmd timeout

QCA_ROME sometimes gets into a state where it is unresponsive to
commands. Since it doesn't have support for a reset gpio, reset the usb
port when this occurs instead.

Signed-off-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: btmtksdio: fix up firmware download sequence
Sean Wang [Fri, 19 Jun 2020 19:52:02 +0000 (19:52 +0000)]
Bluetooth: btmtksdio: fix up firmware download sequence

Data RAM on the device have to be powered on before starting to download
the firmware.

Fixes: 9aebfd4a2200 ("Bluetooth: mediatek: add support for MediaTek MT7663S and MT7668S SDIO devices")
Co-developed-by: Mark Chen <Mark-YW.Chen@mediatek.com>
Signed-off-by: Mark Chen <Mark-YW.Chen@mediatek.com>
Signed-off-by: Sean Wang <sean.wang@mediatek.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: btusb: fix up firmware download sequence
Sean Wang [Fri, 19 Jun 2020 19:52:01 +0000 (19:52 +0000)]
Bluetooth: btusb: fix up firmware download sequence

Data RAM on the device have to be powered on before starting to download
the firmware.

Fixes: a1c49c434e15 ("Bluetooth: btusb: Add protocol support for MediaTek MT7668U USB devices")
Co-developed-by: Mark Chen <Mark-YW.Chen@mediatek.com>
Signed-off-by: Mark Chen <Mark-YW.Chen@mediatek.com>
Signed-off-by: Sean Wang <sean.wang@mediatek.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: add a mutex lock to avoid UAF in do_enale_set
Lihong Kou [Tue, 23 Jun 2020 12:28:41 +0000 (20:28 +0800)]
Bluetooth: add a mutex lock to avoid UAF in do_enale_set

In the case we set or free the global value listen_chan in
different threads, we can encounter the UAF problems because
the method is not protected by any lock, add one to avoid
this bug.

BUG: KASAN: use-after-free in l2cap_chan_close+0x48/0x990
net/bluetooth/l2cap_core.c:730
Read of size 8 at addr ffff888096950000 by task kworker/1:102/2868

CPU: 1 PID: 2868 Comm: kworker/1:102 Not tainted 5.5.0-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine,
BIOS Google 01/01/2011
Workqueue: events do_enable_set
Call Trace:
 __dump_stack lib/dump_stack.c:77 [inline]
 dump_stack+0x1fb/0x318 lib/dump_stack.c:118
 print_address_description+0x74/0x5c0 mm/kasan/report.c:374
 __kasan_report+0x149/0x1c0 mm/kasan/report.c:506
 kasan_report+0x26/0x50 mm/kasan/common.c:641
 __asan_report_load8_noabort+0x14/0x20 mm/kasan/generic_report.c:135
 l2cap_chan_close+0x48/0x990 net/bluetooth/l2cap_core.c:730
 do_enable_set+0x660/0x900 net/bluetooth/6lowpan.c:1074
 process_one_work+0x7f5/0x10f0 kernel/workqueue.c:2264
 worker_thread+0xbbc/0x1630 kernel/workqueue.c:2410
 kthread+0x332/0x350 kernel/kthread.c:255
 ret_from_fork+0x24/0x30 arch/x86/entry/entry_64.S:352

Allocated by task 2870:
 save_stack mm/kasan/common.c:72 [inline]
 set_track mm/kasan/common.c:80 [inline]
 __kasan_kmalloc+0x118/0x1c0 mm/kasan/common.c:515
 kasan_kmalloc+0x9/0x10 mm/kasan/common.c:529
 kmem_cache_alloc_trace+0x221/0x2f0 mm/slab.c:3551
 kmalloc include/linux/slab.h:555 [inline]
 kzalloc include/linux/slab.h:669 [inline]
 l2cap_chan_create+0x50/0x320 net/bluetooth/l2cap_core.c:446
 chan_create net/bluetooth/6lowpan.c:640 [inline]
 bt_6lowpan_listen net/bluetooth/6lowpan.c:959 [inline]
 do_enable_set+0x6a4/0x900 net/bluetooth/6lowpan.c:1078
 process_one_work+0x7f5/0x10f0 kernel/workqueue.c:2264
 worker_thread+0xbbc/0x1630 kernel/workqueue.c:2410
 kthread+0x332/0x350 kernel/kthread.c:255
 ret_from_fork+0x24/0x30 arch/x86/entry/entry_64.S:352

Freed by task 2870:
 save_stack mm/kasan/common.c:72 [inline]
 set_track mm/kasan/common.c:80 [inline]
 kasan_set_free_info mm/kasan/common.c:337 [inline]
 __kasan_slab_free+0x12e/0x1e0 mm/kasan/common.c:476
 kasan_slab_free+0xe/0x10 mm/kasan/common.c:485
 __cache_free mm/slab.c:3426 [inline]
 kfree+0x10d/0x220 mm/slab.c:3757
 l2cap_chan_destroy net/bluetooth/l2cap_core.c:484 [inline]
 kref_put include/linux/kref.h:65 [inline]
 l2cap_chan_put+0x170/0x190 net/bluetooth/l2cap_core.c:498
 do_enable_set+0x66c/0x900 net/bluetooth/6lowpan.c:1075
 process_one_work+0x7f5/0x10f0 kernel/workqueue.c:2264
 worker_thread+0xbbc/0x1630 kernel/workqueue.c:2410
 kthread+0x332/0x350 kernel/kthread.c:255
 ret_from_fork+0x24/0x30 arch/x86/entry/entry_64.S:352

The buggy address belongs to the object at ffff888096950000
 which belongs to the cache kmalloc-2k of size 2048
The buggy address is located 0 bytes inside of
 2048-byte region [ffff888096950000ffff888096950800)
The buggy address belongs to the page:
page:ffffea00025a5400 refcount:1 mapcount:0 mapping:ffff8880aa400e00 index:0x0
flags: 0xfffe0000000200(slab)
raw: 00fffe0000000200 ffffea00027d1548 ffffea0002397808 ffff8880aa400e00
raw: 0000000000000000 ffff888096950000 0000000100000001 0000000000000000
page dumped because: kasan: bad access detected

Memory state around the buggy address:
 ffff88809694ff00: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 ffff88809694ff80: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
>ffff888096950000: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
                   ^
 ffff888096950080: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
 ffff888096950100: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
==================================================================

Reported-by: syzbot+96414aa0033c363d8458@syzkaller.appspotmail.com
Signed-off-by: Lihong Kou <koulihong@huawei.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: Disconnect if E0 is used for Level 4
Luiz Augusto von Dentz [Wed, 20 May 2020 21:20:14 +0000 (14:20 -0700)]
Bluetooth: Disconnect if E0 is used for Level 4

E0 is not allowed with Level 4:

BLUETOOTH CORE SPECIFICATION Version 5.2 | Vol 3, Part C page 1319:

  '128-bit equivalent strength for link and encryption keys
   required using FIPS approved algorithms (E0 not allowed,
   SAFER+ not allowed, and P-192 not allowed; encryption key
   not shortened'

SC enabled:

> HCI Event: Read Remote Extended Features (0x23) plen 13
        Status: Success (0x00)
        Handle: 256
        Page: 1/2
        Features: 0x0b 0x00 0x00 0x00 0x00 0x00 0x00 0x00
          Secure Simple Pairing (Host Support)
          LE Supported (Host)
          Secure Connections (Host Support)
> HCI Event: Encryption Change (0x08) plen 4
        Status: Success (0x00)
        Handle: 256
        Encryption: Enabled with AES-CCM (0x02)

SC disabled:

> HCI Event: Read Remote Extended Features (0x23) plen 13
        Status: Success (0x00)
        Handle: 256
        Page: 1/2
        Features: 0x03 0x00 0x00 0x00 0x00 0x00 0x00 0x00
          Secure Simple Pairing (Host Support)
          LE Supported (Host)
> HCI Event: Encryption Change (0x08) plen 4
        Status: Success (0x00)
        Handle: 256
        Encryption: Enabled with E0 (0x01)
[May 8 20:23] Bluetooth: hci0: Invalid security: expect AES but E0 was used
< HCI Command: Disconnect (0x01|0x0006) plen 3
        Handle: 256
        Reason: Authentication Failure (0x05)

Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: use configured params for ext adv
Alain Michaud [Mon, 22 Jun 2020 13:30:28 +0000 (13:30 +0000)]
Bluetooth: use configured params for ext adv

When the extended advertisement feature is enabled, a hardcoded min and
max interval of 0x8000 is used.  This patch fixes this issue by using
the configured min/max value.

This was validated by setting min/max in main.conf and making sure the
right setting is applied:

< HCI Command: LE Set Extended Advertising Parameters (0x08|0x0036) plen
25                                          #93 [hci0] 10.953011

Min advertising interval: 181.250 msec (0x0122)
Max advertising interval: 181.250 msec (0x0122)


Signed-off-by: Alain Michaud <alainm@chromium.org>
Reviewed-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Reviewed-by: Daniel Winkler <danielwinkler@google.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: Add hci_dev_lock to get/set device flags
Abhishek Pandit-Subedi [Sat, 20 Jun 2020 00:10:24 +0000 (17:10 -0700)]
Bluetooth: Add hci_dev_lock to get/set device flags

Adding hci_dev_lock since hci_conn_params_(lookup|add) require this
lock.

Suggested-by: Miao-chen Chou <mcchou@chromium.org>
Signed-off-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: mgmt: Use command complete on success for set system config
Marcel Holtmann [Wed, 17 Jun 2020 14:39:20 +0000 (16:39 +0200)]
Bluetooth: mgmt: Use command complete on success for set system config

The command status reply is only for failure. When completing set system
config command, the reply has to be command complete.

Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Reviewed-by: Alain Michaud <alainm@chromium.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
4 years agoBluetooth: Terminate the link if pairing is cancelled
Manish Mandlik [Wed, 17 Jun 2020 14:39:19 +0000 (16:39 +0200)]
Bluetooth: Terminate the link if pairing is cancelled

If user decides to cancel the ongoing pairing process (e.g. by clicking
the cancel button on pairing/passkey window), abort any ongoing pairing
and then terminate the link if it was created because of the pair
device action.

Signed-off-by: Manish Mandlik <mmandlik@google.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
4 years agoBluetooth: Update background scan and report device based on advertisement monitors
Miao-chen Chou [Wed, 17 Jun 2020 14:39:18 +0000 (16:39 +0200)]
Bluetooth: Update background scan and report device based on advertisement monitors

This calls hci_update_background_scan() when there is any update on the
advertisement monitors. If there is at least one advertisement monitor,
the filtering policy of scan parameters should be 0x00. This also reports
device found mgmt events if there is at least one monitor.

The following cases were tested with btmgmt advmon-* commands.
(1) add a ADV monitor and observe that the passive scanning is
triggered.
(2) remove the last ADV monitor and observe that the passive scanning is
terminated.
(3) with a LE peripheral paired, repeat (1) and observe the passive
scanning continues.
(4) with a LE peripheral paired, repeat (2) and observe the passive
scanning continues.
(5) with a ADV monitor, suspend/resume the host and observe the passive
scanning continues.

Signed-off-by: Miao-chen Chou <mcchou@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
4 years agoBluetooth: Notify adv monitor removed event
Miao-chen Chou [Wed, 17 Jun 2020 14:39:17 +0000 (16:39 +0200)]
Bluetooth: Notify adv monitor removed event

This notifies management sockets on MGMT_EV_ADV_MONITOR_REMOVED event.

The following test was performed.
- Start two btmgmt consoles, issue a btmgmt advmon-remove command on one
console and observe a MGMT_EV_ADV_MONITOR_REMOVED event on the other.

Signed-off-by: Miao-chen Chou <mcchou@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
4 years agoBluetooth: Notify adv monitor added event
Miao-chen Chou [Wed, 17 Jun 2020 14:39:16 +0000 (16:39 +0200)]
Bluetooth: Notify adv monitor added event

This notifies management sockets on MGMT_EV_ADV_MONITOR_ADDED event.

The following test was performed.
- Start two btmgmt consoles, issue a btmgmt advmon-add command on one
console and observe a MGMT_EV_ADV_MONITOR_ADDED event on the other

Signed-off-by: Miao-chen Chou <mcchou@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
4 years agoBluetooth: Add handler of MGMT_OP_REMOVE_ADV_MONITOR
Miao-chen Chou [Wed, 17 Jun 2020 14:39:15 +0000 (16:39 +0200)]
Bluetooth: Add handler of MGMT_OP_REMOVE_ADV_MONITOR

This adds the request handler of MGMT_OP_REMOVE_ADV_MONITOR command.
Note that the controller-based monitoring is not yet in place. This
removes the internal monitor(s) without sending HCI traffic, so the
request returns immediately.

The following test was performed.
- Issue btmgmt advmon-remove with valid and invalid handles.

Signed-off-by: Miao-chen Chou <mcchou@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
4 years agoBluetooth: Add handler of MGMT_OP_ADD_ADV_PATTERNS_MONITOR
Miao-chen Chou [Wed, 17 Jun 2020 14:39:14 +0000 (16:39 +0200)]
Bluetooth: Add handler of MGMT_OP_ADD_ADV_PATTERNS_MONITOR

This adds the request handler of MGMT_OP_ADD_ADV_PATTERNS_MONITOR command.
Note that the controller-based monitoring is not yet in place. This tracks
the content of the monitor without sending HCI traffic, so the request
returns immediately.

The following manual test was performed.
- Issue btmgmt advmon-add with valid and invalid inputs.
- Issue btmgmt advmon-add more the allowed number of monitors.

Signed-off-by: Miao-chen Chou <mcchou@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
4 years agoBluetooth: Add handler of MGMT_OP_READ_ADV_MONITOR_FEATURES
Miao-chen Chou [Wed, 17 Jun 2020 14:39:13 +0000 (16:39 +0200)]
Bluetooth: Add handler of MGMT_OP_READ_ADV_MONITOR_FEATURES

This adds the request handler of MGMT_OP_READ_ADV_MONITOR_FEATURES
command. Since the controller-based monitoring is not yet in place, this
report only the supported features but not the enabled features.

The following test was performed.
- Issuing btmgmt advmon-features.

Signed-off-by: Miao-chen Chou <mcchou@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
4 years agoBluetooth: Add definitions for advertisement monitor features
Miao-chen Chou [Wed, 17 Jun 2020 14:39:12 +0000 (16:39 +0200)]
Bluetooth: Add definitions for advertisement monitor features

This adds support for Advertisement Monitor API. Here are the commands
and events added.
- Read Advertisement Monitor Feature command
- Add Advertisement Pattern Monitor command
- Remove Advertisement Monitor command
- Advertisement Monitor Added event
- Advertisement Monitor Removed event

Signed-off-by: Miao-chen Chou <mcchou@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
4 years agoBluetooth: Add get/set device flags mgmt op
Abhishek Pandit-Subedi [Wed, 17 Jun 2020 14:39:11 +0000 (16:39 +0200)]
Bluetooth: Add get/set device flags mgmt op

Add the get device flags and set device flags mgmt ops and the device
flags changed event. Their behavior is described in detail in
mgmt-api.txt in bluez.

Sample btmon trace when a HID device is added (trimmed to 75 chars):

@ MGMT Command: Unknown (0x0050) plen 11        {0x0001} [hci0] 18:06:14.98
        90 c5 13 cd f3 cd 02 01 00 00 00                 ...........
@ MGMT Event: Unknown (0x002a) plen 15          {0x0004} [hci0] 18:06:14.98
        90 c5 13 cd f3 cd 02 01 00 00 00 01 00 00 00     ...............
@ MGMT Event: Unknown (0x002a) plen 15          {0x0003} [hci0] 18:06:14.98
        90 c5 13 cd f3 cd 02 01 00 00 00 01 00 00 00     ...............
@ MGMT Event: Unknown (0x002a) plen 15          {0x0002} [hci0] 18:06:14.98
        90 c5 13 cd f3 cd 02 01 00 00 00 01 00 00 00     ...............
@ MGMT Event: Command Compl.. (0x0001) plen 10  {0x0001} [hci0] 18:06:14.98
      Unknown (0x0050) plen 7
        Status: Success (0x00)
        90 c5 13 cd f3 cd 02                             .......
@ MGMT Command: Add Device (0x0033) plen 8      {0x0001} [hci0] 18:06:14.98
        LE Address: CD:F3:CD:13:C5:90 (Static)
        Action: Auto-connect remote device (0x02)
@ MGMT Event: Device Added (0x001a) plen 8      {0x0004} [hci0] 18:06:14.98
        LE Address: CD:F3:CD:13:C5:90 (Static)
        Action: Auto-connect remote device (0x02)
@ MGMT Event: Device Added (0x001a) plen 8      {0x0003} [hci0] 18:06:14.98
        LE Address: CD:F3:CD:13:C5:90 (Static)
        Action: Auto-connect remote device (0x02)
@ MGMT Event: Device Added (0x001a) plen 8      {0x0002} [hci0] 18:06:14.98
        LE Address: CD:F3:CD:13:C5:90 (Static)
        Action: Auto-connect remote device (0x02)
@ MGMT Event: Unknown (0x002a) plen 15          {0x0004} [hci0] 18:06:14.98
        90 c5 13 cd f3 cd 02 01 00 00 00 01 00 00 00     ...............
@ MGMT Event: Unknown (0x002a) plen 15          {0x0003} [hci0] 18:06:14.98
        90 c5 13 cd f3 cd 02 01 00 00 00 01 00 00 00     ...............
@ MGMT Event: Unknown (0x002a) plen 15          {0x0002} [hci0] 18:06:14.98
        90 c5 13 cd f3 cd 02 01 00 00 00 01 00 00 00     ...............
@ MGMT Event: Unknown (0x002a) plen 15          {0x0001} [hci0] 18:06:14.98
        90 c5 13 cd f3 cd 02 01 00 00 00 01 00 00 00     ...............

Signed-off-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Reviewed-by: Alain Michaud <alainm@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
4 years agoBluetooth: Replace wakeable in hci_conn_params
Abhishek Pandit-Subedi [Wed, 17 Jun 2020 14:39:10 +0000 (16:39 +0200)]
Bluetooth: Replace wakeable in hci_conn_params

Replace the wakeable boolean with flags in hci_conn_params and all users
of this boolean. This will be used by the get/set device flags mgmt op.

Signed-off-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Reviewed-by: Alain Michaud <alainm@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
4 years agoBluetooth: Replace wakeable list with flag
Abhishek Pandit-Subedi [Wed, 17 Jun 2020 14:39:09 +0000 (16:39 +0200)]
Bluetooth: Replace wakeable list with flag

Since the classic device list now supports flags, convert the wakeable
list into a flag on the existing device list.

Signed-off-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Reviewed-by: Alain Michaud <alainm@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
4 years agoBluetooth: Add bdaddr_list_with_flags for classic whitelist
Abhishek Pandit-Subedi [Wed, 17 Jun 2020 14:39:08 +0000 (16:39 +0200)]
Bluetooth: Add bdaddr_list_with_flags for classic whitelist

In order to more easily add device flags to classic devices, create
a new type of bdaddr_list that supports setting flags.

Signed-off-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Reviewed-by: Alain Michaud <alainm@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
4 years agoBluetooth: mgmt: Add commands for runtime configuration
Marcel Holtmann [Wed, 17 Jun 2020 14:39:07 +0000 (16:39 +0200)]
Bluetooth: mgmt: Add commands for runtime configuration

This adds the required read/set commands for runtime configuration. Even
while currently no parameters are specified, the commands are made
available.

Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Reviewed-by: Alain Michaud <alainm@chromium.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
4 years agoBluetooth: implement read/set default system parameters mgmt
Alain Michaud [Thu, 11 Jun 2020 02:01:57 +0000 (02:01 +0000)]
Bluetooth: implement read/set default system parameters mgmt

This patch implements the read default system parameters and the set
default system parameters mgmt commands.

Signed-off-by: Alain Michaud <alainm@chromium.org>
Reviewed-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: centralize default value initialization.
Alain Michaud [Thu, 11 Jun 2020 02:01:56 +0000 (02:01 +0000)]
Bluetooth: centralize default value initialization.

This patch centralized the initialization of default parameters.  This
is required to allow clients to more easily customize the default
system parameters.

Signed-off-by: Alain Michaud <alainm@chromium.org>
Reviewed-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: mgmt: read/set system parameter definitions
Alain Michaud [Thu, 11 Jun 2020 02:01:55 +0000 (02:01 +0000)]
Bluetooth: mgmt: read/set system parameter definitions

This patch submits the corresponding kernel definitions to mgmt.h.
This is submitted before the implementation to avoid any conflicts in
values allocations.

Signed-off-by: Alain Michaud <alainm@chromium.org>
Reviewed-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Reviewed-by: Yu Liu <yudiliu@google.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: hci_qca: Request Tx clock vote off only when Tx is pending
Balakrishna Godavarthi [Fri, 12 Jun 2020 12:15:17 +0000 (17:45 +0530)]
Bluetooth: hci_qca: Request Tx clock vote off only when Tx is pending

Tx pending flag is set to true when HOST IBS state is AWAKE or
AWAKEING. If IBS state is ASLEEP, then Tx clock is already voted
off. To optimize further directly calling serial_clock_vote()
instead of qca_wq_serial_tx_clock_vote_off(), at this point of
qca_suspend() already data is sent out. No need to wake up hci to
send data.

Signed-off-by: Balakrishna Godavarthi <bgodavar@codeaurora.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: hci_qca: Increase SoC idle timeout to 200ms
Balakrishna Godavarthi [Fri, 12 Jun 2020 12:21:31 +0000 (17:51 +0530)]
Bluetooth: hci_qca: Increase SoC idle timeout to 200ms

In some version of WCN399x, SoC idle timeout is configured
as 80ms instead of 20ms or 40ms. To honor all the SoC's
supported in the driver increasing SoC idle timeout to 200ms.

Fixes: 41d5b25fed0a0 ("Bluetooth: hci_qca: add PM support")
Signed-off-by: Balakrishna Godavarthi <bgodavar@codeaurora.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: hci_qca: Disable SoC debug logging for WCN3991
Balakrishna Godavarthi [Fri, 12 Jun 2020 12:20:28 +0000 (17:50 +0530)]
Bluetooth: hci_qca: Disable SoC debug logging for WCN3991

By default, WCN3991 sent debug packets to HOST via ACL packet
with header 0xDC2E. This logging is not required on commercial
devices. With this patch SoC logging is disabled post fw
download.

Signed-off-by: Balakrishna Godavarthi <bgodavar@codeaurora.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: Use only 8 bits for the HCI CMSG state flags
Alain Michaud [Thu, 11 Jun 2020 14:26:10 +0000 (14:26 +0000)]
Bluetooth: Use only 8 bits for the HCI CMSG state flags

This change implements suggestions from the code review of the SCO CMSG
state flag patch.

Signed-off-by: Alain Michaud <alainm@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: Add support for BT_PKT_STATUS CMSG data for SCO connections
Alain Michaud [Thu, 11 Jun 2020 19:50:41 +0000 (19:50 +0000)]
Bluetooth: Add support for BT_PKT_STATUS CMSG data for SCO connections

This change adds support for reporting the BT_PKT_STATUS to the socket
CMSG data to allow the implementation of a packet loss correction on
erroneous data received on the SCO socket.

The patch was partially developed by Marcel Holtmann and validated by
Hsin-yu Chao.

Signed-off-by: Alain Michaud <alainm@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: btmrvl_sdio: Refactor irq wakeup
Abhishek Pandit-Subedi [Thu, 11 Jun 2020 01:53:55 +0000 (18:53 -0700)]
Bluetooth: btmrvl_sdio: Refactor irq wakeup

Use device_init_wakeup to allow the Bluetooth dev to wake the system
from suspend. Currently, the device can wake the system but no
power/wakeup entry is created in sysfs to allow userspace to disable
wakeup.

Signed-off-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: btmrvl_sdio: Implement prevent_wake
Abhishek Pandit-Subedi [Thu, 11 Jun 2020 01:53:54 +0000 (18:53 -0700)]
Bluetooth: btmrvl_sdio: Implement prevent_wake

Use the parent device's power/wakeup to control whether we support
remote wake. If remote wakeup is disabled, Bluetooth will not enable
scanning for incoming connections.

Signed-off-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: btmrvl_sdio: Set parent dev to hdev
Abhishek Pandit-Subedi [Thu, 11 Jun 2020 01:53:53 +0000 (18:53 -0700)]
Bluetooth: btmrvl_sdio: Set parent dev to hdev

Set the correct parent dev when registering hdev. This allows userspace
tools to find the parent device (for example, to set the power/wakeup
property).

Before this change, the path was /sys/devices/virtual/bluetooth/hci0
and after this change, it looks more like:
/sys/bus/mmc/devices/mmc1:0001/mmc1:0001:2/bluetooth/hci0

Signed-off-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: btusb: Configure Intel debug feature based on available support
Chethan T N [Mon, 8 Jun 2020 12:27:47 +0000 (17:57 +0530)]
Bluetooth: btusb: Configure Intel debug feature based on available support

This patch shall enable the Intel telemetry exception format
based on the supported features

Signed-off-by: Chethan T N <chethan.tumkur.narayan@intel.com>
Signed-off-by: Ps AyappadasX <AyappadasX.Ps@intel.com>
Signed-off-by: Kiran K <kiran.k@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: btusb: Add support to read Intel debug feature
Chethan T N [Mon, 8 Jun 2020 12:27:46 +0000 (17:57 +0530)]
Bluetooth: btusb: Add support to read Intel debug feature

The command shall read the Intel controller supported
debug feature. Based on the supported features additional debug
configuration shall be enabled.

Signed-off-by: Chethan T N <chethan.tumkur.narayan@intel.com>
Signed-off-by: Ps AyappadasX <AyappadasX.Ps@intel.com>
Signed-off-by: Kiran K <kiran.k@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: hci_qca: Bug fix during SSR timeout
Venkata Lakshmi Narayana Gubba [Tue, 9 Jun 2020 15:57:08 +0000 (21:27 +0530)]
Bluetooth: hci_qca: Bug fix during SSR timeout

Due to race conditions between qca_hw_error and qca_controller_memdump
during SSR timeout,the same pointer is freed twice. This results in a
double free. Now a lock is acquired before checking the stauts of SSR
state.

Fixes: d841502c79e3 ("Bluetooth: hci_qca: Collect controller memory dump during SSR")
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: Allow suspend even when preparation has failed
Abhishek Pandit-Subedi [Fri, 5 Jun 2020 20:50:15 +0000 (13:50 -0700)]
Bluetooth: Allow suspend even when preparation has failed

It is preferable to allow suspend even when Bluetooth has problems
preparing for sleep. When Bluetooth fails to finish preparing for
suspend, log the error and allow the suspend notifier to continue
instead.

To also make it clearer why suspend failed, change bt_dev_dbg to
bt_dev_err when handling the suspend timeout.

Fixes: dd522a7429b07e ("Bluetooth: Handle LE devices during suspend")
Reported-by: Len Brown <len.brown@intel.com>
Signed-off-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: hci_qca: Refactor error handling in qca_suspend()
Matthias Kaehlcke [Fri, 5 Jun 2020 18:46:11 +0000 (11:46 -0700)]
Bluetooth: hci_qca: Refactor error handling in qca_suspend()

If waiting for IBS sleep times out jump to the error handler, this is
easier to read than multiple 'if' branches and a fall through to the
error handler.

Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Reviewed-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: hci_qca: Skip serdev wait when no transfer is pending
Matthias Kaehlcke [Fri, 5 Jun 2020 18:46:10 +0000 (11:46 -0700)]
Bluetooth: hci_qca: Skip serdev wait when no transfer is pending

qca_suspend() calls serdev_device_wait_until_sent() regardless of
whether a transfer is pending. While it does no active harm since
the function should return immediately it makes the code more
confusing. Add a flag to track whether a transfer is pending and
only call serdev_device_wait_until_sent() is needed.

Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Reviewed-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: hci_qca: Only remove TX clock vote after TX is completed
Matthias Kaehlcke [Fri, 5 Jun 2020 18:46:09 +0000 (11:46 -0700)]
Bluetooth: hci_qca: Only remove TX clock vote after TX is completed

qca_suspend() removes the vote for the UART TX clock after
writing an IBS sleep request to the serial buffer. This is
not a good idea since there is no guarantee that the request
has been sent at this point. Instead remove the vote after
successfully entering IBS sleep. This also fixes the issue
of the vote being removed in case of an aborted suspend due
to a failure of entering IBS sleep.

Fixes: 41d5b25fed0a0 ("Bluetooth: hci_qca: add PM support")
Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Reviewed-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: hci_qca: Simplify determination of serial clock on/off state from votes
Matthias Kaehlcke [Sat, 6 Jun 2020 16:53:06 +0000 (09:53 -0700)]
Bluetooth: hci_qca: Simplify determination of serial clock on/off state from votes

The serial clocks should be on when there is a vote for at least one
of the clocks (RX or TX), and off when there is no 'on' vote. The
current logic to determine the combined state is a bit redundant
in the code paths for different types of votes, use a single
statement in the common path instead.

Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: hci_qca: Fix an error pointer dereference
Dan Carpenter [Fri, 29 May 2020 09:59:48 +0000 (12:59 +0300)]
Bluetooth: hci_qca: Fix an error pointer dereference

When a function like devm_clk_get_optional() function returns both error
pointers on error and NULL then the NULL return means that the optional
feature is deliberately disabled.  It is a special sort of success and
should not trigger an error message.  The surrounding code should be
written to check for NULL and not crash.

On the other hand, if we encounter an error, then the probe from should
clean up and return a failure.

In this code, if devm_clk_get_optional() returns an error pointer then
the kernel will crash inside the call to:

clk_set_rate(qcadev->susclk, SUSCLK_RATE_32KHZ);

The error handling must be updated to prevent that.

Fixes: 77131dfec6af ("Bluetooth: hci_qca: Replace devm_gpiod_get() with devm_gpiod_get_optional()")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: Removing noisy dbg message
Alain Michaud [Mon, 1 Jun 2020 14:20:59 +0000 (14:20 +0000)]
Bluetooth: Removing noisy dbg message

This patch removes a particularly noisy dbg message.  The debug message
isn't particularly interesting for debuggability so it was simply
removed to reduce noise in dbg logs.

Signed-off-by: Alain Michaud <alainm@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoBluetooth: Check scan state before disabling during suspend
Manish Mandlik [Tue, 2 Jun 2020 01:42:51 +0000 (18:42 -0700)]
Bluetooth: Check scan state before disabling during suspend

Check current scan state by checking HCI_LE_SCAN flag and send scan
disable command only if scan is already enabled.

Signed-off-by: Manish Mandlik <mmandlik@google.com>
Reviewed-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Reviewed-by: Alain Michaud <alainm@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agobtmrvl: Fix firmware filename for sd8997 chipset
Pali Rohár [Wed, 3 Jun 2020 08:22:29 +0000 (10:22 +0200)]
btmrvl: Fix firmware filename for sd8997 chipset

Firmware for sd8997 chipset is distributed by Marvell package and also as
part of the linux-firmware repository in filename sdsd8997_combo_v4.bin.

This patch fixes mwifiex driver to load correct firmware file for sd8997.

Fixes: f0ef67485f591 ("Bluetooth: btmrvl: add sd8997 chipset support")
Signed-off-by: Pali Rohár <pali@kernel.org>
Acked-by: Ganapathi Bhat <ganapathi.bhat@nxp.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agobtmrvl: Fix firmware filename for sd8977 chipset
Pali Rohár [Wed, 3 Jun 2020 08:22:28 +0000 (10:22 +0200)]
btmrvl: Fix firmware filename for sd8977 chipset

Firmware for sd8977 chipset is distributed by Marvell package and also as
part of the linux-firmware repository in filename sdsd8977_combo_v2.bin.

This patch fixes mwifiex driver to load correct firmware file for sd8977.

Fixes: 8c57983bf7a79 ("Bluetooth: btmrvl: add support for sd8977 chipset")
Signed-off-by: Pali Rohár <pali@kernel.org>
Acked-by: Ganapathi Bhat <ganapathi.bhat@nxp.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agomwifiex: Fix firmware filename for sd8997 chipset
Pali Rohár [Wed, 3 Jun 2020 08:22:27 +0000 (10:22 +0200)]
mwifiex: Fix firmware filename for sd8997 chipset

Firmware for sd8997 chipset is distributed by Marvell package and also as
part of the linux-firmware repository in filename sdsd8997_combo_v4.bin.

This patch fixes mwifiex driver to load correct firmware file for sd8997.

Fixes: 6d85ef00d9dfe ("mwifiex: add support for 8997 chipset")
Signed-off-by: Pali Rohár <pali@kernel.org>
Acked-by: Ganapathi Bhat <ganapathi.bhat@nxp.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agomwifiex: Fix firmware filename for sd8977 chipset
Pali Rohár [Wed, 3 Jun 2020 08:22:26 +0000 (10:22 +0200)]
mwifiex: Fix firmware filename for sd8977 chipset

Firmware for sd8977 chipset is distributed by Marvell package and also as
part of the linux-firmware repository in filename sdsd8977_combo_v2.bin.

This patch fixes mwifiex driver to load correct firmware file for sd8977.

Fixes: 1a0f547831dce ("mwifiex: add support for sd8977 chipset")
Signed-off-by: Pali Rohár <pali@kernel.org>
Acked-by: Ganapathi Bhat <ganapathi.bhat@nxp.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
4 years agoselftests: net: ip_defrag: ignore EPERM
Thadeu Lima de Souza Cascardo [Tue, 2 Jun 2020 18:38:37 +0000 (15:38 -0300)]
selftests: net: ip_defrag: ignore EPERM

When running with conntrack rules, the dropped overlap fragments may cause
EPERM to be returned to sendto. Instead of completely failing, just ignore
those errors and continue. If this causes packets with overlap fragments to
be dropped as expected, that is okay. And if it causes packets that are
expected to be received to be dropped, which should not happen, it will be
detected as failure.

Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet_failover: fixed rollback in net_failover_open()
Vasily Averin [Tue, 2 Jun 2020 12:55:26 +0000 (15:55 +0300)]
net_failover: fixed rollback in net_failover_open()

found by smatch:
drivers/net/net_failover.c:65 net_failover_open() error:
 we previously assumed 'primary_dev' could be null (see line 43)

Fixes: cfc80d9a1163 ("net: Introduce net_failover driver")
Signed-off-by: Vasily Averin <vvs@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'tipc-revert-two-patches'
David S. Miller [Tue, 2 Jun 2020 22:13:47 +0000 (15:13 -0700)]
Merge branch 'tipc-revert-two-patches'

Tuong Lien says:

====================
tipc: revert two patches

We revert two patches:

tipc: Fix potential tipc_node refcnt leak in tipc_rcv
tipc: Fix potential tipc_aead refcnt leak in tipc_crypto_rcv

which prevented TIPC encryption from working properly and caused kernel
panic.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoRevert "tipc: Fix potential tipc_aead refcnt leak in tipc_crypto_rcv"
Tuong Lien [Tue, 2 Jun 2020 04:46:41 +0000 (11:46 +0700)]
Revert "tipc: Fix potential tipc_aead refcnt leak in tipc_crypto_rcv"

This reverts commit 441870ee4240cf67b5d3ab8e16216a9ff42eb5d6.

Like the previous patch in this series, we revert the above commit that
causes similar issues with the 'aead' object.

Acked-by: Jon Maloy <jmaloy@redhat.com>
Signed-off-by: Tuong Lien <tuong.t.lien@dektech.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoRevert "tipc: Fix potential tipc_node refcnt leak in tipc_rcv"
Tuong Lien [Tue, 2 Jun 2020 04:46:40 +0000 (11:46 +0700)]
Revert "tipc: Fix potential tipc_node refcnt leak in tipc_rcv"

This reverts commit de058420767df21e2b6b0f3bb36d1616fb962032.

There is no actual tipc_node refcnt leak as stated in the above commit.
The refcnt is hold carefully for the case of an asynchronous decryption
(i.e. -EINPROGRESS/-EBUSY and skb = NULL is returned), so that the node
object cannot be freed in the meantime. The counter will be re-balanced
when the operation's callback arrives with the decrypted buffer if any.
In other cases, e.g. a synchronous crypto the counter will be decreased
immediately when it is done.

Now with that commit, a kernel panic will occur when there is no node
found (i.e. n = NULL) in the 'tipc_rcv()' or a premature release of the
node object.

This commit solves the issues by reverting the said commit, but keeping
one valid case that the 'skb_linearize()' is failed.

Acked-by: Jon Maloy <jmaloy@redhat.com>
Signed-off-by: Tuong Lien <tuong.t.lien@dektech.com.au>
Tested-by: Hoang Le <hoang.h.le@dektech.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agovmxnet3: allow rx flow hash ops only when rss is enabled
Ronak Doshi [Tue, 2 Jun 2020 03:02:39 +0000 (20:02 -0700)]
vmxnet3: allow rx flow hash ops only when rss is enabled

It makes sense to allow changes to get/set rx flow hash callback only
when rss is enabled. This patch restricts get_rss_hash_opts and
set_rss_hash_opts methods to allow querying and configuring different
Rx flow hash configurations only when rss is enabled

Signed-off-by: Ronak Doshi <doshir@vmware.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agohinic: add set_channels ethtool_ops support
Luo bin [Tue, 2 Jun 2020 00:40:32 +0000 (08:40 +0800)]
hinic: add set_channels ethtool_ops support

add support to change TX/RX queue number with "ethtool -L combined".

V5 -> V6: remove check for carrier in hinic_xmit_frame
V4 -> V5: change time zone in patch header
V3 -> V4: update date in patch header
V2 -> V3: remove check for zero channels->combined_count
V1 -> V2: update commit message("ethtool -L" to "ethtool -L combined")
V0 -> V1: remove check for channels->tx_count/rx_count/other_count

Signed-off-by: Luo bin <luobin9@huawei.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 [Tue, 2 Jun 2020 22:09:39 +0000 (15:09 -0700)]
Merge git://git./linux/kernel/git/bpf/bpf-next

Daniel Borkmann says:

====================
pull-request: bpf-next 2020-06-02

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

We've added 10 non-merge commits during the last 1 day(s) which contain
a total of 15 files changed, 229 insertions(+), 74 deletions(-).

The main changes are:

1) Several fixes to s390 BPF JIT e.g. fixing kernel panic when BPF stack is
   not 8-byte aligned, from Ilya Leoshkevich.

2) Fix bpf_skb_adjust_room() helper's CHECKSUM_UNNECESSARY handling which
   was wrongly bypassing TCP checksum verification, from Daniel Borkmann.

3) Fix tools/bpf/ build under MAKEFLAGS=rR which causes built-in CXX and
   others vars to be undefined, also from Ilya Leoshkevich.

4) Fix BPF ringbuf's selftest shared sample_cnt variable to avoid compiler
   optimizations on it, from Andrii Nakryiko.

5) Fix up test_verifier selftest due to addition of rx_queue_mapping to
   the bpf_sock structure, from Alexei Starovoitov.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoselftests/bpf: Add a default $(CXX) value
Ilya Leoshkevich [Tue, 2 Jun 2020 17:56:49 +0000 (19:56 +0200)]
selftests/bpf: Add a default $(CXX) value

When using make kselftest TARGETS=bpf, tools/bpf is built with
MAKEFLAGS=rR, which causes $(CXX) to be undefined, which in turn causes
the build to fail with

  CXX      test_cpp
/bin/sh: 2: g: not found

Fix by adding a default $(CXX) value, like tools/build/feature/Makefile
already does.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20200602175649.2501580-3-iii@linux.ibm.com
4 years agotools/bpf: Don't use $(COMPILE.c)
Ilya Leoshkevich [Tue, 2 Jun 2020 17:56:48 +0000 (19:56 +0200)]
tools/bpf: Don't use $(COMPILE.c)

When using make kselftest TARGETS=bpf, tools/bpf is built with
MAKEFLAGS=rR, which causes $(COMPILE.c) to be undefined, which in turn
causes the build to fail with

  CC       kselftest/bpf/tools/build/bpftool/map_perf_ring.o
/bin/sh: 1: -MMD: not found

Fix by using $(CC) $(CFLAGS) -c instead of $(COMPILE.c).

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20200602175649.2501580-2-iii@linux.ibm.com
4 years agobpf, selftests: Use bpf_probe_read_kernel
Ilya Leoshkevich [Tue, 2 Jun 2020 17:44:48 +0000 (19:44 +0200)]
bpf, selftests: Use bpf_probe_read_kernel

Since commit 0ebeea8ca8a4 ("bpf: Restrict bpf_probe_read{, str}() only to
archs where they work") 44 verifier tests fail on s390 due to not having
bpf_probe_read anymore. Fix by using bpf_probe_read_kernel.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20200602174448.2501214-1-iii@linux.ibm.com
4 years agos390/bpf: Use bcr 0,%0 as tail call nop filler
Ilya Leoshkevich [Tue, 2 Jun 2020 17:45:55 +0000 (19:45 +0200)]
s390/bpf: Use bcr 0,%0 as tail call nop filler

Currently used 0x0000 filler confuses bfd disassembler, making bpftool
prog dump xlated output nearly useless. Fix by using a real instruction.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20200602174555.2501389-1-iii@linux.ibm.com
4 years agos390/bpf: Maintain 8-byte stack alignment
Ilya Leoshkevich [Tue, 2 Jun 2020 17:43:39 +0000 (19:43 +0200)]
s390/bpf: Maintain 8-byte stack alignment

Certain kernel functions (e.g. get_vtimer/set_vtimer) cause kernel
panic when the stack is not 8-byte aligned. Currently JITed BPF programs
may trigger this by allocating stack frames with non-rounded sizes and
then being interrupted. Fix by using rounded fp->aux->stack_depth.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20200602174339.2501066-1-iii@linux.ibm.com
4 years agoselftests/bpf: Fix verifier test
Alexei Starovoitov [Tue, 2 Jun 2020 18:57:43 +0000 (11:57 -0700)]
selftests/bpf: Fix verifier test

Adjust verifier test due to addition of new field.

Fixes: c3c16f2ea6d2 ("bpf: Add rx_queue_mapping to bpf_sock")
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
4 years agoselftests/bpf: Fix sample_cnt shared between two threads
Andrii Nakryiko [Tue, 2 Jun 2020 05:03:49 +0000 (22:03 -0700)]
selftests/bpf: Fix sample_cnt shared between two threads

Make sample_cnt volatile to fix possible selftests failure due to compiler
optimization preventing latest sample_cnt value to be visible to main thread.
sample_cnt is incremented in background thread, which is then joined into main
thread. So in terms of visibility sample_cnt update is ok. But because it's
not volatile, compiler might make optimizations that would prevent main thread
to see latest updated value. Fix this by marking global variable volatile.

Fixes: cb1c9ddd5525 ("selftests/bpf: Add BPF ringbuf selftests")
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/20200602050349.215037-1-andriin@fb.com
4 years agoMerge branch 'csum-fixes'
Alexei Starovoitov [Tue, 2 Jun 2020 18:50:23 +0000 (11:50 -0700)]
Merge branch 'csum-fixes'

Daniel Borkmann says:

====================
This series fixes an issue originally reported by Lorenz Bauer where using
the bpf_skb_adjust_room() helper hid a checksum bug since it wasn't adjusting
CHECKSUM_UNNECESSARY's skb->csum_level after decap. The fix is two-fold:
 i) We do a safe reset in bpf_skb_adjust_room() to CHECKSUM_NONE with an opt-
    out flag BPF_F_ADJ_ROOM_NO_CSUM_RESET.
ii) We add a new bpf_csum_level() for the latter in order to allow users to
    manually inc/dec the skb->csum_level when needed.
The series is rebased against latest bpf-next tree. It can be applied there,
or to bpf after the merge win sync from net-next.

Thanks!
====================

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
4 years agobpf, selftests: Adapt cls_redirect to call csum_level helper
Daniel Borkmann [Tue, 2 Jun 2020 14:58:34 +0000 (16:58 +0200)]
bpf, selftests: Adapt cls_redirect to call csum_level helper

Adapt bpf_skb_adjust_room() to pass in BPF_F_ADJ_ROOM_NO_CSUM_RESET flag and
use the new bpf_csum_level() helper to inc/dec the checksum level by one after
the encap/decap.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Reviewed-by: Lorenz Bauer <lmb@cloudflare.com>
Link: https://lore.kernel.org/bpf/e7458f10e3f3d795307cbc5ad870112671d9c6f7.1591108731.git.daniel@iogearbox.net
4 years agobpf: Add csum_level helper for fixing up csum levels
Daniel Borkmann [Tue, 2 Jun 2020 14:58:33 +0000 (16:58 +0200)]
bpf: Add csum_level helper for fixing up csum levels

Add a bpf_csum_level() helper which BPF programs can use in combination
with bpf_skb_adjust_room() when they pass in BPF_F_ADJ_ROOM_NO_CSUM_RESET
flag to the latter to avoid falling back to CHECKSUM_NONE.

The bpf_csum_level() allows to adjust CHECKSUM_UNNECESSARY skb->csum_levels
via BPF_CSUM_LEVEL_{INC,DEC} which calls __skb_{incr,decr}_checksum_unnecessary()
on the skb. The helper also allows a BPF_CSUM_LEVEL_RESET which sets the skb's
csum to CHECKSUM_NONE as well as a BPF_CSUM_LEVEL_QUERY to just return the
current level. Without this helper, there is no way to otherwise adjust the
skb->csum_level. I did not add an extra dummy flags as there is plenty of free
bitspace in level argument itself iff ever needed in future.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Reviewed-by: Alan Maguire <alan.maguire@oracle.com>
Acked-by: Lorenz Bauer <lmb@cloudflare.com>
Link: https://lore.kernel.org/bpf/279ae3717cb3d03c0ffeb511493c93c450a01e1a.1591108731.git.daniel@iogearbox.net
4 years agobpf: Fix up bpf_skb_adjust_room helper's skb csum setting
Daniel Borkmann [Tue, 2 Jun 2020 14:58:32 +0000 (16:58 +0200)]
bpf: Fix up bpf_skb_adjust_room helper's skb csum setting

Lorenz recently reported:

  In our TC classifier cls_redirect [0], we use the following sequence of
  helper calls to decapsulate a GUE (basically IP + UDP + custom header)
  encapsulated packet:

    bpf_skb_adjust_room(skb, -encap_len, BPF_ADJ_ROOM_MAC, BPF_F_ADJ_ROOM_FIXED_GSO)
    bpf_redirect(skb->ifindex, BPF_F_INGRESS)

  It seems like some checksums of the inner headers are not validated in
  this case. For example, a TCP SYN packet with invalid TCP checksum is
  still accepted by the network stack and elicits a SYN ACK. [...]

  That is, we receive the following packet from the driver:

    | ETH | IP | UDP | GUE | IP | TCP |
    skb->ip_summed == CHECKSUM_UNNECESSARY

  ip_summed is CHECKSUM_UNNECESSARY because our NICs do rx checksum offloading.
  On this packet we run skb_adjust_room_mac(-encap_len), and get the following:

    | ETH | IP | TCP |
    skb->ip_summed == CHECKSUM_UNNECESSARY

  Note that ip_summed is still CHECKSUM_UNNECESSARY. After bpf_redirect()'ing
  into the ingress, we end up in tcp_v4_rcv(). There, skb_checksum_init() is
  turned into a no-op due to CHECKSUM_UNNECESSARY.

The bpf_skb_adjust_room() helper is not aware of protocol specifics. Internally,
it handles the CHECKSUM_COMPLETE case via skb_postpull_rcsum(), but that does
not cover CHECKSUM_UNNECESSARY. In this case skb->csum_level of the original
skb prior to bpf_skb_adjust_room() call was 0, that is, covering UDP. Right now
there is no way to adjust the skb->csum_level. NICs that have checksum offload
disabled (CHECKSUM_NONE) or that support CHECKSUM_COMPLETE are not affected.

Use a safe default for CHECKSUM_UNNECESSARY by resetting to CHECKSUM_NONE and
add a flag to the helper called BPF_F_ADJ_ROOM_NO_CSUM_RESET that allows users
from opting out. Opting out is useful for the case where we don't remove/add
full protocol headers, or for the case where a user wants to adjust the csum
level manually e.g. through bpf_csum_level() helper that is added in subsequent
patch.

The bpf_skb_proto_{4_to_6,6_to_4}() for NAT64/46 translation from the BPF
bpf_skb_change_proto() helper uses bpf_skb_net_hdr_{push,pop}() pair internally
as well but doesn't change layers, only transitions between v4 to v6 and vice
versa, therefore no adoption is required there.

  [0] https://lore.kernel.org/bpf/20200424185556.7358-1-lmb@cloudflare.com/

Fixes: 2be7e212d541 ("bpf: add bpf_skb_adjust_room helper")
Reported-by: Lorenz Bauer <lmb@cloudflare.com>
Reported-by: Alan Maguire <alan.maguire@oracle.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Lorenz Bauer <lmb@cloudflare.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Reviewed-by: Alan Maguire <alan.maguire@oracle.com>
Link: https://lore.kernel.org/bpf/CACAyw9-uU_52esMd1JjuA80fRPHJv5vsSg8GnfW3t_qDU4aVKQ@mail.gmail.com/
Link: https://lore.kernel.org/bpf/11a90472e7cce83e76ddbfce81fdfce7bfc68808.1591108731.git.daniel@iogearbox.net
4 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next
David S. Miller [Mon, 1 Jun 2020 22:53:08 +0000 (15:53 -0700)]
Merge git://git./linux/kernel/git/bpf/bpf-next

Alexei Starovoitov says:

====================
pull-request: bpf-next 2020-06-01

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

We've added 55 non-merge commits during the last 1 day(s) which contain
a total of 91 files changed, 4986 insertions(+), 463 deletions(-).

The main changes are:

1) Add rx_queue_mapping to bpf_sock from Amritha.

2) Add BPF ring buffer, from Andrii.

3) Attach and run programs through devmap, from David.

4) Allow SO_BINDTODEVICE opt in bpf_setsockopt, from Ferenc.

5) link based flow_dissector, from Jakub.

6) Use tracing helpers for lsm programs, from Jiri.

7) Several sk_msg fixes and extensions, from John.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agosfc: add missing annotation for efx_ef10_try_update_nic_stats_vf()
Jules Irenge [Mon, 1 Jun 2020 18:45:52 +0000 (19:45 +0100)]
sfc: add missing annotation for efx_ef10_try_update_nic_stats_vf()

Sparse reports a warning at efx_ef10_try_update_nic_stats_vf()
warning: context imbalance in efx_ef10_try_update_nic_stats_vf()
- unexpected unlock
The root cause is the missing annotation at
efx_ef10_try_update_nic_stats_vf()
Add the missing _must_hold(&efx->stats_lock) annotation

Signed-off-by: Jules Irenge <jbi.octave@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agocrypto/chtls: IPv6 support for inline TLS
Vinay Kumar Yadav [Mon, 1 Jun 2020 18:37:05 +0000 (00:07 +0530)]
crypto/chtls: IPv6 support for inline TLS

Extends support to IPv6 for Inline TLS server.

Signed-off-by: Vinay Kumar Yadav <vinay.yadav@chelsio.com>
v1->v2:
- cc'd tcp folks.

v2->v3:
- changed EXPORT_SYMBOL() to EXPORT_SYMBOL_GPL()

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'chelsio-crypto-fixes'
David S. Miller [Mon, 1 Jun 2020 22:50:26 +0000 (15:50 -0700)]
Merge branch 'chelsio-crypto-fixes'

Ayush Sawal says:

====================
Fixing compilation warnings and errors

Patch 1: Fixes the warnings seen when compiling using sparse tool.

Patch 2: Fixes a cocci check error introduced after commit
567be3a5d227 ("crypto: chelsio -
Use multiple txq/rxq per tfm to process the requests").

V1->V2

patch1: Avoid type casting by using get_unaligned_be32() and
     put_unaligned_be16/32() functions.

patch2: Modified subject of the patch.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoCrypto/chcr: Fixes a coccinile check error
Ayush Sawal [Mon, 1 Jun 2020 17:41:59 +0000 (23:11 +0530)]
Crypto/chcr: Fixes a coccinile check error

This fixes an error observed after running coccinile check.
drivers/crypto/chelsio/chcr_algo.c:1462:5-8: Unneeded variable:
"err". Return "0" on line 1480

This line is missed in the commit 567be3a5d227 ("crypto:
chelsio - Use multiple txq/rxq per tfm to process the requests").

Fixes: 567be3a5d227 ("crypto:
chelsio - Use multiple txq/rxq per tfm to process the requests").

V1->V2
-Modified subject.

Signed-off-by: Ayush Sawal <ayush.sawal@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoCrypto/chcr: Fixes compilations warnings
Ayush Sawal [Mon, 1 Jun 2020 17:41:58 +0000 (23:11 +0530)]
Crypto/chcr: Fixes compilations warnings

This patch fixes the compilation warnings displayed by sparse tool for
chcr driver.

V1->V2

Avoid type casting by using get_unaligned_be32() and
put_unaligned_be16/32() functions.

The key which comes from stack is an u8 byte stream so we store it in
an unsigned char array(ablkctx->key). The function get_aes_decrypt_key()
is a used to calculate  the reverse round key for decryption, for this
operation the key has to be divided into 4 bytes, so to extract 4 bytes
from an u8 byte stream and store it in an u32 variable, get_aligned_be32()
is used. Similarly for copying back the key from u32 variable to the
original u8 key stream, put_aligned_be32() is used.

Signed-off-by: Ayush Sawal <ayush.sawal@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agocrypto/chcr: IPV6 code needs to be in CONFIG_IPV6
Rohit Maheshwari [Mon, 1 Jun 2020 14:03:32 +0000 (19:33 +0530)]
crypto/chcr: IPV6 code needs to be in CONFIG_IPV6

Error messages seen while building kernel with CONFIG_IPV6
disabled.

Signed-off-by: Rohit Maheshwari <rohitm@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agocxgb4/chcr: Enable ktls settings at run time
Rohit Maheshwari [Mon, 1 Jun 2020 07:38:29 +0000 (13:08 +0530)]
cxgb4/chcr: Enable ktls settings at run time

Current design enables ktls setting from start, which is not
efficient. Now the feature will be enabled when user demands
TLS offload on any interface.

v1->v2:
- taking ULD module refcount till any single connection exists.
- taking rtnl_lock() before clearing tls_devops.

v2->v3:
- cxgb4 is now registering to tlsdev_ops.
- module refcount inc/dec in chcr.
- refcount is only for connections.
- removed new code from cxgb_set_feature().

v3->v4:
- fixed warning message.

Signed-off-by: Rohit Maheshwari <rohitm@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoipv6: fix IPV6_ADDRFORM operation logic
Hangbin Liu [Mon, 1 Jun 2020 03:55:03 +0000 (11:55 +0800)]
ipv6: fix IPV6_ADDRFORM operation logic

Socket option IPV6_ADDRFORM supports UDP/UDPLITE and TCP at present.
Previously the checking logic looks like:
if (sk->sk_protocol == IPPROTO_UDP || sk->sk_protocol == IPPROTO_UDPLITE)
do_some_check;
else if (sk->sk_protocol != IPPROTO_TCP)
break;

After commit b6f6118901d1 ("ipv6: restrict IPV6_ADDRFORM operation"), TCP
was blocked as the logic changed to:
if (sk->sk_protocol == IPPROTO_UDP || sk->sk_protocol == IPPROTO_UDPLITE)
do_some_check;
else if (sk->sk_protocol == IPPROTO_TCP)
do_some_check;
break;
else
break;

Then after commit 82c9ae440857 ("ipv6: fix restrict IPV6_ADDRFORM operation")
UDP/UDPLITE were blocked as the logic changed to:
if (sk->sk_protocol == IPPROTO_UDP || sk->sk_protocol == IPPROTO_UDPLITE)
do_some_check;
if (sk->sk_protocol == IPPROTO_TCP)
do_some_check;

if (sk->sk_protocol != IPPROTO_TCP)
break;

Fix it by using Eric's code and simply remove the break in TCP check, which
looks like:
if (sk->sk_protocol == IPPROTO_UDP || sk->sk_protocol == IPPROTO_UDPLITE)
do_some_check;
else if (sk->sk_protocol == IPPROTO_TCP)
do_some_check;
else
break;

Fixes: 82c9ae440857 ("ipv6: fix restrict IPV6_ADDRFORM operation")
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agotipc: Fix NULL pointer dereference in __tipc_sendstream()
YueHaibing [Thu, 28 May 2020 14:34:07 +0000 (22:34 +0800)]
tipc: Fix NULL pointer dereference in __tipc_sendstream()

tipc_sendstream() may send zero length packet, then tipc_msg_append()
do not alloc skb, skb_peek_tail() will get NULL, msg_set_ack_required
will trigger NULL pointer dereference.

Reported-by: syzbot+8eac6d030e7807c21d32@syzkaller.appspotmail.com
Fixes: 0a3e060f340d ("tipc: add test for Nagle algorithm effectiveness")
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'Link-based-attach-to-netns'
Alexei Starovoitov [Mon, 1 Jun 2020 22:21:03 +0000 (15:21 -0700)]
Merge branch 'Link-based-attach-to-netns'

Jakub Sitnicki says:

====================
One of the pieces of feedback from recent review of BPF hooks for socket
lookup [0] was that new program types should use bpf_link-based
attachment.

This series introduces new bpf_link type for attaching to network
namespace. All link operations are supported. Errors returned from ops
follow cgroup example. Patch 4 description goes into error semantics.

The major change in v2 is a switch away from RCU to mutex-only
synchronization. Andrii pointed out that it is not needed, and it makes
sense to keep locking straightforward.

Also, there were a couple of bugs in update_prog and fill_info initial
implementation, one picked up by kbuild. Those are now fixed. Tests have
been extended to cover them. Full changelog below.

Series is organized as so:

Patches 1-3 prepare a space in struct net to keep state for attached BPF
programs, and massage the code in flow_dissector to make it attach type
agnostic, to finally move it under kernel/bpf/.

Patch 4, the most important one, introduces new bpf_link link type for
attaching to network namespace.

Patch 5 unifies the update error (ENOLINK) between BPF cgroup and netns.

Patches 6-8 make libbpf and bpftool aware of the new link type.

Patches 9-12 Add and extend tests to check that link low- and high-level
API for operating on links to netns works as intended.

Thanks to Alexei, Andrii, Lorenz, Marek, and Stanislav for feedback.

-jkbs

[0] https://lore.kernel.org/bpf/20200511185218.1422406-1-jakub@cloudflare.com/

Cc: Alexei Starovoitov <alexei.starovoitov@gmail.com>
Cc: Andrii Nakryiko <andrii.nakryiko@gmail.com>
Cc: Lorenz Bauer <lmb@cloudflare.com>
Cc: Marek Majkowski <marek@cloudflare.com>
Cc: Stanislav Fomichev <sdf@google.com>
v1 -> v2:

- Switch to mutex-only synchronization. Don't rely on RCU grace period
  guarantee when accessing struct net from link release / update /
  fill_info, and when accessing bpf_link from pernet pre_exit
  callback. (Andrii)
- Drop patch 1, no longer needed with mutex-only synchronization.
- Don't leak uninitialized variable contents from fill_info callback
  when link is in defunct state. (kbuild)
- Make fill_info treat the link as defunct (i.e. no attached netns) when
  struct net refcount is 0, but link has not been yet auto-detached.
- Add missing BPF_LINK_TYPE define in bpf_types.h for new link type.
- Fix link update_prog callback to update the prog that will run, and
  not just the link itself.
- Return EEXIST on prog attach when link already exists, and on link
  create when prog is already attached directly. (Andrii)
- Return EINVAL on prog detach when link is attached. (Andrii)
- Fold __netns_bpf_link_attach into its only caller. (Stanislav)
- Get rid of a wrapper around container_of() (Andrii)
- Use rcu_dereference_protected instead of rcu_access_pointer on
  update-side. (Stanislav)
- Make return-on-success from netns_bpf_link_create less
  confusing. (Andrii)
- Adapt bpf_link for cgroup to return ENOLINK when updating a defunct
  link. (Andrii, Alexei)
- Order new exported symbols in libbpf.map alphabetically (Andrii)
- Keep libbpf's "failed to attach link" warning message clear as to what
  we failed to attach to (cgroup vs netns). (Andrii)
- Extract helpers for printing link attach type. (bpftool, Andrii)
- Switch flow_dissector tests to BPF skeleton and extend them to
  exercise link-based flow dissector attachment. (Andrii)
- Harden flow dissector attachment tests with prog query checks after
  prog attach/detach, or link create/update/close.
- Extend flow dissector tests to cover fill_info for defunct links.
- Rebase onto recent bpf-next
====================

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
4 years agoselftests/bpf: Extend test_flow_dissector to cover link creation
Jakub Sitnicki [Sun, 31 May 2020 08:28:46 +0000 (10:28 +0200)]
selftests/bpf: Extend test_flow_dissector to cover link creation

Extend the existing flow_dissector test case to run tests once using direct
prog attachments, and then for the second time using indirect attachment
via link.

The intention is to exercises the newly added high-level API for attaching
programs to network namespace with links (bpf_program__attach_netns).

Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200531082846.2117903-13-jakub@cloudflare.com
4 years agoselftests/bpf: Convert test_flow_dissector to use BPF skeleton
Jakub Sitnicki [Sun, 31 May 2020 08:28:45 +0000 (10:28 +0200)]
selftests/bpf: Convert test_flow_dissector to use BPF skeleton

Switch flow dissector test setup from custom BPF object loader to BPF
skeleton to save boilerplate and prepare for testing higher-level API for
attaching flow dissector with bpf_link.

To avoid depending on program order in the BPF object when populating the
flow dissector PROG_ARRAY map, change the program section names to contain
the program index into the map. This follows the example set by tailcall
tests.

Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200531082846.2117903-12-jakub@cloudflare.com
4 years agoselftests/bpf, flow_dissector: Close TAP device FD after the test
Jakub Sitnicki [Sun, 31 May 2020 08:28:44 +0000 (10:28 +0200)]
selftests/bpf, flow_dissector: Close TAP device FD after the test

test_flow_dissector leaves a TAP device after it's finished, potentially
interfering with other tests that will run after it. Fix it by closing the
TAP descriptor on cleanup.

Fixes: 0905beec9f52 ("selftests/bpf: run flow dissector tests in skb-less mode")
Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200531082846.2117903-11-jakub@cloudflare.com
4 years agoselftests/bpf: Add tests for attaching bpf_link to netns
Jakub Sitnicki [Sun, 31 May 2020 08:28:43 +0000 (10:28 +0200)]
selftests/bpf: Add tests for attaching bpf_link to netns

Extend the existing test case for flow dissector attaching to cover:

 - link creation,
 - link updates,
 - link info querying,
 - mixing links with direct prog attachment.

Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200531082846.2117903-10-jakub@cloudflare.com
4 years agobpftool: Support link show for netns-attached links
Jakub Sitnicki [Sun, 31 May 2020 08:28:42 +0000 (10:28 +0200)]
bpftool: Support link show for netns-attached links

Make `bpf link show` aware of new link type, that is links attached to
netns. When listing netns-attached links, display netns inode number as its
identifier and link attach type.

Sample session:

  # readlink /proc/self/ns/net
  net:[4026532251]
  # bpftool prog show
  357: flow_dissector  tag a04f5eef06a7f555  gpl
          loaded_at 2020-05-30T16:53:51+0200  uid 0
          xlated 16B  jited 37B  memlock 4096B
  358: flow_dissector  tag a04f5eef06a7f555  gpl
          loaded_at 2020-05-30T16:53:51+0200  uid 0
          xlated 16B  jited 37B  memlock 4096B
  # bpftool link show
  108: netns  prog 357
          netns_ino 4026532251  attach_type flow_dissector
  # bpftool link -jp show
  [{
          "id": 108,
          "type": "netns",
          "prog_id": 357,
          "netns_ino": 4026532251,
          "attach_type": "flow_dissector"
      }
  ]

  (... after netns is gone ...)

  # bpftool link show
  108: netns  prog 357
          netns_ino 0  attach_type flow_dissector
  # bpftool link -jp show
  [{
          "id": 108,
          "type": "netns",
          "prog_id": 357,
          "netns_ino": 0,
          "attach_type": "flow_dissector"
      }
  ]

Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200531082846.2117903-9-jakub@cloudflare.com
4 years agobpftool: Extract helpers for showing link attach type
Jakub Sitnicki [Sun, 31 May 2020 08:28:41 +0000 (10:28 +0200)]
bpftool: Extract helpers for showing link attach type

Code for printing link attach_type is duplicated in a couple of places, and
likely will be duplicated for future link types as well. Create helpers to
prevent duplication.

Suggested-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200531082846.2117903-8-jakub@cloudflare.com
4 years agolibbpf: Add support for bpf_link-based netns attachment
Jakub Sitnicki [Sun, 31 May 2020 08:28:40 +0000 (10:28 +0200)]
libbpf: Add support for bpf_link-based netns attachment

Add bpf_program__attach_nets(), which uses LINK_CREATE subcommand to create
an FD-based kernel bpf_link, for attach types tied to network namespace,
that is BPF_FLOW_DISSECTOR for the moment.

Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200531082846.2117903-7-jakub@cloudflare.com
4 years agobpf, cgroup: Return ENOLINK for auto-detached links on update
Jakub Sitnicki [Sun, 31 May 2020 08:28:39 +0000 (10:28 +0200)]
bpf, cgroup: Return ENOLINK for auto-detached links on update

Failure to update a bpf_link because it has been auto-detached by a dying
cgroup currently results in EINVAL error, even though the arguments passed
to bpf() syscall are not wrong.

bpf_links attaching to netns in this case will return ENOLINK, which
carries the message that the link is no longer attached to anything.

Change cgroup bpf_links to do the same to keep the uAPI errors consistent.

Fixes: 0c991ebc8c69 ("bpf: Implement bpf_prog replacement for an active bpf_cgroup_link")
Suggested-by: Lorenz Bauer <lmb@cloudflare.com>
Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200531082846.2117903-6-jakub@cloudflare.com
4 years agobpf: Add link-based BPF program attachment to network namespace
Jakub Sitnicki [Sun, 31 May 2020 08:28:38 +0000 (10:28 +0200)]
bpf: Add link-based BPF program attachment to network namespace

Extend bpf() syscall subcommands that operate on bpf_link, that is
LINK_CREATE, LINK_UPDATE, OBJ_GET_INFO, to accept attach types tied to
network namespaces (only flow dissector at the moment).

Link-based and prog-based attachment can be used interchangeably, but only
one can exist at a time. Attempts to attach a link when a prog is already
attached directly, and the other way around, will be met with -EEXIST.
Attempts to detach a program when link exists result in -EINVAL.

Attachment of multiple links of same attach type to one netns is not
supported with the intention to lift the restriction when a use-case
presents itself. Because of that link create returns -E2BIG when trying to
create another netns link, when one already exists.

Link-based attachments to netns don't keep a netns alive by holding a ref
to it. Instead links get auto-detached from netns when the latter is being
destroyed, using a pernet pre_exit callback.

When auto-detached, link lives in defunct state as long there are open FDs
for it. -ENOLINK is returned if a user tries to update a defunct link.

Because bpf_link to netns doesn't hold a ref to struct net, special care is
taken when releasing, updating, or filling link info. The netns might be
getting torn down when any of these link operations are in progress. That
is why auto-detach and update/release/fill_info are synchronized by the
same mutex. Also, link ops have to always check if auto-detach has not
happened yet and if netns is still alive (refcnt > 0).

Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200531082846.2117903-5-jakub@cloudflare.com
4 years agoflow_dissector: Move out netns_bpf prog callbacks
Jakub Sitnicki [Sun, 31 May 2020 08:28:37 +0000 (10:28 +0200)]
flow_dissector: Move out netns_bpf prog callbacks

Move functions to manage BPF programs attached to netns that are not
specific to flow dissector to a dedicated module named
bpf/net_namespace.c.

The set of functions will grow with the addition of bpf_link support for
netns attached programs. This patch prepares ground by creating a place
for it.

This is a code move with no functional changes intended.

Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200531082846.2117903-4-jakub@cloudflare.com
4 years agonet: Introduce netns_bpf for BPF programs attached to netns
Jakub Sitnicki [Sun, 31 May 2020 08:28:36 +0000 (10:28 +0200)]
net: Introduce netns_bpf for BPF programs attached to netns

In order to:

 (1) attach more than one BPF program type to netns, or
 (2) support attaching BPF programs to netns with bpf_link, or
 (3) support multi-prog attach points for netns

we will need to keep more state per netns than a single pointer like we
have now for BPF flow dissector program.

Prepare for the above by extracting netns_bpf that is part of struct net,
for storing all state related to BPF programs attached to netns.

Turn flow dissector callbacks for querying/attaching/detaching a program
into generic ones that operate on netns_bpf. Next patch will move the
generic callbacks into their own module.

This is similar to how it is organized for cgroup with cgroup_bpf.

Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Cc: Stanislav Fomichev <sdf@google.com>
Link: https://lore.kernel.org/bpf/20200531082846.2117903-3-jakub@cloudflare.com
4 years agoflow_dissector: Pull locking up from prog attach callback
Jakub Sitnicki [Sun, 31 May 2020 08:28:35 +0000 (10:28 +0200)]
flow_dissector: Pull locking up from prog attach callback

Split out the part of attach callback that happens with attach/detach lock
acquired. This structures the prog attach callback in a way that opens up
doors for moving the locking out of flow_dissector and into generic
callbacks for attaching/detaching progs to netns in subsequent patches.

Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Reviewed-by: Stanislav Fomichev <sdf@google.com>
Link: https://lore.kernel.org/bpf/20200531082846.2117903-2-jakub@cloudflare.com
4 years agolibbpf: Add _GNU_SOURCE for reallocarray to ringbuf.c
Andrii Nakryiko [Mon, 1 Jun 2020 20:26:01 +0000 (13:26 -0700)]
libbpf: Add _GNU_SOURCE for reallocarray to ringbuf.c

On systems with recent enough glibc, reallocarray compat won't kick in, so
reallocarray() itself has to come from stdlib.h include. But _GNU_SOURCE is
necessary to enable it. So add it.

Fixes: bf99c936f947 ("libbpf: Add BPF ring buffer support")
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/20200601202601.2139477-1-andriin@fb.com
4 years agobpf: Use tracing helpers for lsm programs
Jiri Olsa [Sun, 31 May 2020 15:42:55 +0000 (17:42 +0200)]
bpf: Use tracing helpers for lsm programs

Currenty lsm uses bpf_tracing_func_proto helpers which do
not include stack trace or perf event output. It's useful
to have those for bpftrace lsm support [1].

Using tracing_prog_func_proto helpers for lsm programs.

[1] https://github.com/iovisor/bpftrace/pull/1347

Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Cc: KP Singh <kpsingh@google.com>
Link: https://lore.kernel.org/bpf/20200531154255.896551-1-jolsa@kernel.org
4 years agoxdp: Rename convert_to_xdp_frame in xdp_convert_buff_to_frame
Lorenzo Bianconi [Thu, 28 May 2020 20:47:29 +0000 (22:47 +0200)]
xdp: Rename convert_to_xdp_frame in xdp_convert_buff_to_frame

In order to use standard 'xdp' prefix, rename convert_to_xdp_frame
utility routine in xdp_convert_buff_to_frame and replace all the
occurrences

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Link: https://lore.kernel.org/bpf/6344f739be0d1a08ab2b9607584c4d5478c8c083.1590698295.git.lorenzo@kernel.org