Change Reno to properly handle stretch ACKs in additive increase mode
by passing in the count of ACKed packets to tcp_cong_avoid_ai().
In addition, if snd_cwnd crosses snd_ssthresh during slow start
processing, and we then exit slow start mode, we need to carry over
any remaining "credit" for packets ACKed and apply that to additive
increase by passing this remaining "acked" count to
tcp_cong_avoid_ai().
Reported-by: Eyal Perry <eyalpe@mellanox.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
tcp_cong_avoid_ai() was too timid (snd_cwnd increased too slowly) on
"stretch ACKs" -- cases where the receiver ACKed more than 1 packet in
a single ACK. For example, suppose w is 10 and we get a stretch ACK
for 20 packets, so acked is 20. We ought to increase snd_cwnd by 2
(since acked/w = 20/10 = 2), but instead we were only increasing cwnd
by 1. This patch fixes that behavior.
Reported-by: Eyal Perry <eyalpe@mellanox.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
LRO, GRO, delayed ACKs, and middleboxes can cause "stretch ACKs" that
cover more than the RFC-specified maximum of 2 packets. These stretch
ACKs can cause serious performance shortfalls in common congestion
control algorithms that were designed and tuned years ago with
receiver hosts that were not using LRO or GRO, and were instead
politely ACKing every other packet.
This patch series fixes Reno and CUBIC to handle stretch ACKs.
This patch prepares for the upcoming stretch ACK bug fix patches. It
adds an "acked" parameter to tcp_cong_avoid_ai() to allow for future
fixes to tcp_cong_avoid_ai() to correctly handle stretch ACKs, and
changes all congestion control algorithms to pass in 1 for the ACKed
count. It also changes tcp_slow_start() to return the number of packet
ACK "credits" that were not processed in slow start mode, and can be
processed by the congestion control module in additive increase mode.
In future patches we will fix tcp_cong_avoid_ai() to handle stretch
ACKs, and fix Reno and CUBIC handling of stretch ACKs in slow start
and additive increase mode.
Reported-by: Eyal Perry <eyalpe@mellanox.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This was "toshiba_acpi: Change sci_open function return value"
Some Toshiba laptops have "poorly implemented" SCI calls on their
BIOSes and are not checking for sci_{open, close} calls, therefore,
the sci_open function is failing and making some of the supported
features unavailable (kbd backlight, touchpad, illumination, etc.).
This patch checks whether we receive TOS_NOT_SUPPORTED and returns
1, making the supported features work on such laptops.
In the case that some laptops really do not support the SCI, all the
SCI dependent functions check for TOS_NOT_SUPPORTED, and thus, not
registering support for the queried feature.
Signed-off-by: Azael Avalos <coproscefalo@gmail.com>
Signed-off-by: Darren Hart <dvhart@linux.intel.com>
The smp_unregister() function needs to be called every time the
controller is powered down. There are multiple entry points when
this can happen. One is "hciconfig hci0 reset" which will throw
a WARN_ON when LE support has been enabled.
[ 78.564620] WARNING: CPU: 0 PID: 148 at net/bluetooth/smp.c:3075 smp_register+0xf1/0x170()
[ 78.564622] Modules linked in:
[ 78.564628] CPU: 0 PID: 148 Comm: kworker/u3:1 Not tainted 3.19.0-rc4-devel+ #404
[ 78.564629] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS
[ 78.564635] Workqueue: hci0 hci_rx_work
[ 78.564638] ffffffff81b4a7a2 ffff88001cb2fb38 ffffffff8161d881 0000000080000000
[ 78.564642] 0000000000000000 ffff88001cb2fb78 ffffffff8103b870 696e55206e6f6f6d
[ 78.564645] ffff88001d965000 0000000000000000 0000000000000000 ffff88001d965000
[ 78.564648] Call Trace:
[ 78.564655] [<ffffffff8161d881>] dump_stack+0x4f/0x7b
[ 78.564662] [<ffffffff8103b870>] warn_slowpath_common+0x80/0xc0
[ 78.564667] [<ffffffff81544b00>] ? add_uuid+0x1f0/0x1f0
[ 78.564671] [<ffffffff8103b955>] warn_slowpath_null+0x15/0x20
[ 78.564674] [<ffffffff81562d81>] smp_register+0xf1/0x170
[ 78.564680] [<ffffffff81081236>] ? lock_timer_base.isra.30+0x26/0x50
[ 78.564683] [<ffffffff81544bf0>] powered_complete+0xf0/0x120
[ 78.564688] [<ffffffff8152e622>] hci_req_cmd_complete+0x82/0x260
[ 78.564692] [<ffffffff8153554f>] hci_cmd_complete_evt+0x6cf/0x2e20
[ 78.564697] [<ffffffff81623e43>] ? _raw_spin_unlock_irqrestore+0x13/0x30
[ 78.564701] [<ffffffff8106b0af>] ? __wake_up_sync_key+0x4f/0x60
[ 78.564705] [<ffffffff8153a2ab>] hci_event_packet+0xbcb/0x2e70
[ 78.564709] [<ffffffff814094d3>] ? skb_release_all+0x23/0x30
[ 78.564711] [<ffffffff81409529>] ? kfree_skb+0x29/0x40
[ 78.564715] [<ffffffff815296c8>] hci_rx_work+0x1c8/0x3f0
[ 78.564719] [<ffffffff8105bd91>] ? get_parent_ip+0x11/0x50
[ 78.564722] [<ffffffff8105be25>] ? preempt_count_add+0x55/0xb0
[ 78.564727] [<ffffffff8104f65f>] process_one_work+0x12f/0x360
[ 78.564731] [<ffffffff8104ff9b>] worker_thread+0x6b/0x4b0
[ 78.564735] [<ffffffff8104ff30>] ? cancel_delayed_work_sync+0x10/0x10
[ 78.564738] [<ffffffff810542fa>] kthread+0xea/0x100
[ 78.564742] [<ffffffff81620000>] ? __schedule+0x3e0/0x980
[ 78.564745] [<ffffffff81054210>] ? kthread_create_on_node+0x180/0x180
[ 78.564749] [<ffffffff816246ec>] ret_from_fork+0x7c/0xb0
[ 78.564752] [<ffffffff81054210>] ? kthread_create_on_node+0x180/0x180
[ 78.564755] ---[ end trace 8b0d943af76d3736 ]---
This warning is not critical and has only been placed in the code to
actually catch this exact situation. To avoid triggering it move
the smp_unregister() into hci_dev_do_close() which will now also
take care of remove the SMP channel. It is safe to call this function
since it only remove the channel if it has been previously registered.
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Use DEVICE_ATTR_{RO,WO,RW} macros to simplify sysfs attributes
declaration.
To declare a "foo" attribute, DEVICE_ATTR_RW() requires foo_show() and
foo_store(), so rename a few functions to satisfy this requirement.
Also put the macro below each related show/store functions for clarity.
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Read the resources from PCI BAR0 instead of using hardcoded values.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Darren Hart <dvhart@linux.intel.com>
This is small performance optimization of the busy_loop().
While here, use BIT() macro instead of plain integers when check the status.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Darren Hart <dvhart@linux.intel.com>
While here, do couple of amendments:
- move platform variable to the function where it's used
- define intel_scu_ipc_check_status() static
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Simple helpers that pass an arbitrary iov_iter to filesystems.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Added Intel Braswell CPU id for SOC DTS. Since this doesn't support
APIC IRQ, the driver is modified to have capability to not register
any modifiable trips.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Change agp_free_page_array to use kvfree function,
remove the duplicated code.
Signed-off-by: Yalin Wang <yalin.wang@sonymobile.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This backmerges drm-fixes into drm-next mainly for the amdkfd
stuff, I'm not 100% confident, but it builds and the amdkfd
folks can fix anything up.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Conflicts:
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
This contains the long-awaited drm_bridge series that makes Chromebooks
work for people. I had thought this would've been perfect by now, but
then I go and build test it and the first thing it does is yell about a
recursive dependency. I fixed that up because I was feeling bad for not
getting around to look at this earlier.
Biseds that there is new support for two more panels, a couple of fixup
patches to the Sharp LQ101R1SX01 dual-channel DSI panel driver and a
potential NULL pointer dereference fix.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2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=5PkY
-----END PGP SIGNATURE-----
Merge tag 'drm/panel/for-3.20-rc1' of git://anongit.freedesktop.org/tegra/linux into drm-next
drm/panel: Changes for v3.20-rc1
This contains the long-awaited drm_bridge series that makes Chromebooks
work for people. I had thought this would've been perfect by now, but
then I go and build test it and the first thing it does is yell about a
recursive dependency. I fixed that up because I was feeling bad for not
getting around to look at this earlier.
Biseds that there is new support for two more panels, a couple of fixup
patches to the Sharp LQ101R1SX01 dual-channel DSI panel driver and a
potential NULL pointer dereference fix.
* tag 'drm/panel/for-3.20-rc1' of git://anongit.freedesktop.org/tegra/linux: (23 commits)
drm/bridge: dw-hdmi: Adapt to bridge API change
drm/sti: fixup for bridge interface
drm/bridge: dw-hdmi: Fix return error path
drm: Check the right variable when setting formats
Documentation: bridge: Add documentation for ps8622 DT properties
Documentation: devicetree: Add vendor prefix for parade
Documentation: drm: bridge: move to video/bridge
drm/bridge: ptn3460: use gpiod interface
drm/bridge: ptn3460: probe connector at the end of bridge attach
drm/bridge: ptn3460: support drm_panel
drm/exynos: dp: support drm_bridge
drm/bridge: ptn3460: Convert to I2C driver model
drm/bridge: make bridge registration independent of drm flow
drm/bridge: do not pass drm_bridge_funcs to drm_bridge_init
drm/bridge: ptn3460: Few trivial cleanups
drm/panel: simple: Add AVIC TM070DDH03 panel support
of: Add vendor prefix for Shanghai AVIC Optoelectronics Co., Ltd.
drm/panel: sharp: lq101r1sx01: Remove unneeded include
drm/panel: sharp: lq101r1sx01: Respect power timings
drm/panel: sharp: lq101r1sx01: Add delay after display on
...
As of commit 9a1091ef00 ("irqchip: gic: Support hierarchy irq
domain."), the APE6EVM legacy board support is known to be broken.
The IRQ numbers of the GIC are now virtual, and no longer match the
hardcoded hardware IRQ numbers in the legacy platform board code.
To fix this issue specific to non-muliplatform r8a73a4 and APE6EVM:
1) Instantiate the GIC from platform board code and also
2) Skip over the DT arch timer as well as
3) Force delay setup based on DT CPU frequency
With these 3 fixes in place interrupts on APE6EVM are now unbroken.
Partially based on legacy GIC fix by Geert Uytterhoeven, thanks to
him for the initial work.
Signed-off-by: Magnus Damm <damm+renesas@opensource.se>
Acked-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Simon Horman <horms+renesas@verge.net.au>
The DISP1 power domain on Exynos5420 SoC includes the FIMD1, MIXER
and HDMI modules. Add a device node for this power domain and mark
these modules as consumer of the DISP1 power domain.
When a power domain is powered on and off, the input clocks of the
devices attached to it are reparented. So a reference to the input
and parent clocks of the devices are needed to manage that.
Signed-off-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Signed-off-by: Kukjin Kim <kgene@kernel.org>
When a power domain is powered off on Exynos5420 SoC, the input clocks of
the devices attached to this power domain are re-parented to oscclk and
restored to the original parent after powering on the power domain.
So a reference to the input and parent clocks for the devices attached to
a power domain are needed to be able to do the re-parenting. The DISP1 pd
includes modules which uses the following clocks:
ACLK_200_DISP1 (MIXER and HDMILINK)
ACLK_300_DISP1 (FIMD1)
ACLK_400_DISP1 (Internal Buses)
Each of these clocks are generated as the output of a clock mux so add an
ID for all of these clock muxes and their parents to be referenced in the
DISP1 power domain device node.
Signed-off-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Acked-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Acked-by: Michael Turquette <mturquette@linaro.org>
Signed-off-by: Kukjin Kim <kgene@kernel.org>
The owner module reference of the pata_of_platform's scsi_host is
initialized to pata_platform's one, because pata_of_platform driver
use a scsi_host_template defined in pata_platform. So this drivers
can be unloaded even if the scsi device is being accessed.
This fixes it by propagating the scsi_host_template to pata_of_platform
driver. The scsi_host_template is passed through a new
argument of __pata_platform_probe().
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Hans de Goede <hdegoede@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: "James E.J. Bottomley" <JBottomley@parallels.com>
Cc: linux-ide@vger.kernel.org
Cc: linux-scsi@vger.kernel.org
The owner module reference of the ahci platform's scsi_host is
initialized to libahci_platform's one, because these drivers use a
scsi_host_template defined in libahci_platform. So these drivers can
be unloaded even if the scsi device is being accessed.
This fixes it by pushing the scsi_host_template from libahci_platform
to all leaf drivers. The scsi_host_template is passed through a new
argument of ahci_platform_init_host().
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Hans de Goede <hdegoede@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: "James E.J. Bottomley" <JBottomley@parallels.com>
Cc: linux-ide@vger.kernel.org
Cc: linux-scsi@vger.kernel.org
In the function of_pci_get_host_bridge_resources() if the parsing of ranges
fails, previously allocated resources inclusive of bus_range are not freed
and are not expected to be freed by the function caller on error return.
This patch fixes the issues by adding code that properly frees resources
and bus_range before exiting the function with an error return value.
Fixes: cbe4097f8a ("of/pci: Add support for parsing PCI host bridge resources from DT")
Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Liviu Dudau <liviu.dudau@arm.com>
CC: Arnd Bergmann <arnd@arndb.de>
CC: Rob Herring <robh+dt@kernel.org>
375/38x. Only switch the PL310 to I/O coherent mode if I/O coherency
is enabled.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=qzYQ
-----END PGP SIGNATURE-----
Merge tag 'mvebu-fixes-3.19-6' of git://git.infradead.org/linux-mvebu into fixes
Merge "mvebu-fixes-6" from Andrew Lunn:
The previous fix for Armada XP, disabling I/O coherency, broke Armada
375/38x. Only switch the PL310 to I/O coherent mode if I/O coherency
is enabled.
* tag 'mvebu-fixes-3.19-6' of git://git.infradead.org/linux-mvebu:
ARM: mvebu: don't set the PL310 in I/O coherency mode when I/O coherency is disabled
Signed-off-by: Olof Johansson <olof@lixom.net>
The Exynos5420 cluster power management support allowing Exynos
5420/5422/5800 machines to power up and down the secondary CPUs.
Without this option enabled, the secondary CPUs are not brought
up on boot and the following error is shown:
CPU0: thread -1, cpu 0, socket 0, mpidr 80000000
Setting up static identity map for 0x40938e90 - 0x40938f28
ARM CCI driver probed
CPU1: failed to boot: -38
CPU2: failed to boot: -38
CPU3: failed to boot: -38
CPU4: failed to boot: -38
CPU5: failed to boot: -38
CPU6: failed to boot: -38
CPU7: failed to boot: -38
Brought up 1 CPUs
Signed-off-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Signed-off-by: Olof Johansson <olof@lixom.net>
Add binding documentation for the PCI controller found on Versatile PB
boards.
Signed-off-by: Rob Herring <robh@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
CC: Pawel Moll <pawel.moll@arm.com>
CC: Mark Rutland <mark.rutland@arm.com>
CC: Ian Campbell <ijc+devicetree@hellion.org.uk>
CC: Kumar Gala <galak@codeaurora.org>
This patch fix multiple words such as "the the" and "which which"
in Documentation/devicetree.
Signed-off-by: Masanari Iida <standby24x7@gmail.com>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
This patch fix a spelling typo in Documentation/sysctl/vm.txt
Signed-off-by: Masanari Iida <standby24x7@gmail.com>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Daniel Wagner <daniel.wagner@bmw-carit.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
Protect the call with a mutex, as this may be called in parallel
(either from the PCM rate change and the clock change).
Acked-by: Jaroslav Kysela <perex@perex.cz>
Tested-by: Pavel Hofman <pavel.hofman@ivitera.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Define snd_ak4114_suspend() and snd_ak4114_resume() functions to
handle PM properly, stopping and restarting the work at PM.
Currently only ice1712/juli.c deals with the PM and ak4114, so fix the
calls there appropriately.
The same PM functions are defined in ak4113.c, too, although they
aren't currently called yet (ice1712/quartet.c may be enhanced to
support PM later).
Acked-by: Jaroslav Kysela <perex@perex.cz>
Tested-by: Pavel Hofman <pavel.hofman@ivitera.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
... just to follow the standard coding style.
Acked-by: Jaroslav Kysela <perex@perex.cz>
Tested-by: Pavel Hofman <pavel.hofman@ivitera.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
When ak4114 work calls its callback and the callback invokes
ak4114_reinit(), it stalls due to flush_delayed_work(). For avoiding
this, control the reentrance by introducing a refcount. Also
flush_delayed_work() is replaced with cancel_delayed_work_sync().
The exactly same bug is present in ak4113.c and fixed as well.
Reported-by: Pavel Hofman <pavel.hofman@ivitera.com>
Acked-by: Jaroslav Kysela <perex@perex.cz>
Tested-by: Pavel Hofman <pavel.hofman@ivitera.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The Intel Bluetooth controllers can provide an additional exception
info string when a hardware error event occurs. The core will now
call hdev->hw_error to let the driver read out this information.
This change will cause a reset of the hardware to bring it back
into functional state and then read the Intel exception info
string and print it along with the error information.
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
When receiving a HCI Hardware Error event, the controller should be
assumed to be non-functional until issuing a HCI Reset command.
The Bluetooth hardware errors are vendor specific and so add a
new hdev->hw_error callback that drivers can provide to run extra
code to handle the hardware error.
After completing the vendor specific error handling perform a full
reset of the Bluetooth stack by closing and re-opening the transport.
Based-on-patch-by: Johan Hedberg <johan.hedberg@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Split the hci_dev_reset ioctl handling into using hci_dev_do_reset
helper function. Similar to what has been done with hci_dev_do_open
and hci_dev_do_close.
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
The discovery state should be set to stopped when the HCI device is
powered off. This patch adds the appropriate call to the
hci_discovery_set_state() function from hci_dev_do_close() which is
responsible for the power-off procedure.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
When HCI_Reset is issued the discovery state is assumed to be stopped.
The hci_cc_reset() handler was trying to set the state but it was doing
it without using the hci_discovery_set_state() function. Because of this
e.g. the mgmt Discovering event could go without being sent. This patch
fixes the code to use the hci_discovery_set_state() function instead of
just blindly setting the state value.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
There's a check in set_secure_conn() that's supposed to ensure that SSP
is enabled before we try to request the controller to enable SC (since
SSP is a pre-requisite for it). However, this check only makes sense for
controllers actually supporting BR/EDR SC. If we have a 4.0 controller
we're only interested in the LE part of SC and should therefore not be
requiring SSP to be enabled. This patch adds an additional condition to
check for lmp_sc_capable(hdev) before requiring SSP to be enabled.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
The btusb_disconnect() callback calls hci_unregister_dev() which in turn
calls btusb_close() if the HCI device is powered. The btusb_close()
function in turn will call btusb_free_frags(). It's therefore
unnecessary to have another call to btusb_free_frags() in the
btusb_disconnect() function. Besides the redundancy the second call
seems to also cause some strange stability issues which this patch then
also fixes.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
When loading the Intel firmware it can happen that the firmware loading
complete vendor event arrives before the command complete event for the
last firmware fragment.
< HCI Command: Vendor (0x3f|0x0009) plen 7
01 02 fc 03 00 00 00
> HCI Event: Vendor (0xff) plen 5
06 00 00 00 00
> HCI Event: Command Complete (0x0e) plen 4
Vendor (0x3f|0x0009) ncmd 31
Status: Success (0x00)
This is mainly caused by the fact that the vendor command and its
command complete event are transported over the bulk endpoints. The
firmware loading complete event however is send over the interrupt
endpoint. So with just bad timing one event arrives before the other.
Currently the code does not account for it. There are precautions for
receiving firmware loading complete event quickly, but not for receiving
it before the command complete.
Introduce an extra flag that tracks when the firmware sending has
completed from the driver point of view and track the completion of
the firmware loading procedure with a different flag. That way the
wakeup can be handled properly.
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
If Secure Connections Only mode has been enabled, the it is important
to check that OOB data for P-256 values is provided. In case it is not,
then tell the remote side that no OOB data is present.
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
When replying to the IO capability request for Secure Simple Pairing and
Secure Connections, the OOB data present fields needs to set. Instead of
making the calculation inline, split this into a separate helper
function.
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
When Secure Connections Only mode has been enabled and remote OOB data
is requested, then only provide P-256 hash and randomizer vaulues. The
fields for P-192 hash and randomizer should be set to zero.
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Until legacy SMP OOB pairing is implemented user space should be given a
clear error when trying to use it. This patch adds a corresponding check
to the Add Remote OOB Data handler function which returns "invalid
parameters" if non-zero Rand192 or Hash192 parameters were given for an
LE address.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
The core only supports up to 32 slaves, and the chipselect function
expects the same.
Signed-off-by: Ricardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
