Now that SMP has been converted to use fixed channels we've got a bit of
a problem with the hci_conn reference counting. So far the L2CAP code
has kept a reference for each L2CAP channel that was notified of the
connection. With SMP however this would mean that the connection is
never dropped even though there are no other users of it. Furthermore,
SMP already does its own hci_conn reference counting internally,
starting from a security or pairing request and ending with the key
distribution.
This patch makes L2CAP fixed channels default to the L2CAP core not
keeping a hci_conn reference for them. A new FLAG_HOLD_HCI_CONN flag is
added so that L2CAP users can declare an exception to this rule and hold
a reference even for their fixed channels. One such exception is the
L2CAP socket layer which does want a reference for each socket (e.g. an
ATT socket which uses a fixed channel).
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
The l2cap_chan_add() function doesn't require the channel to be
unlocked. It only requires the l2cap_conn to be unlocked. Therefore,
it's unnecessary to unlock a channel before calling l2cap_chan_add().
This patch removes such unnecessary unlocking from the
l2cap_chan_connect() function.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
The l2cap_create_le_flowctl_pdu() function that l2cap_segment_le_sdu()
calls is perfectly capable of doing packet fragmentation if given bigger
PDUs than the HCI buffers allow. Forcing the PDU length based on the HCI
MTU (conn->mtu) would therefore needlessly strict operation on hardware
with limited LE buffers (e.g. both Intel and Broadcom seem to have this
set to just 27 bytes).
This patch removes the restriction and makes it possible to send PDUs of
the full length that the remote MPS value allows.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Cc: stable@vger.kernel.org
The VFIO driver routes LSI interrupts by capturing, masking, and then
delivering. When passing though Mellanox adapters from host to guest,
interrupt storm are reported from host and guest. That's because the PCI
command register INTx Disable bit doesn't work on Mellanox devices.
# lspci | grep Mellanox
0001:05:00.0 Ethernet controller: Mellanox Technologies MT27500 Family [ConnectX-3]
0005:01:00.0 Ethernet controller: Mellanox Technologies MT26448 [ConnectX EN 10GigE, PCIe 2.0 5GT/s] (rev b0)
Amir Vadai confirmed that all Mellanox devices have same problem.
The patch marks broken INTx masking for all Mellanox adapters.
Suggested-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-By: Amir Vadai <amirv@mellanox.com>
The idea here is to translate a value of 0 received from
the firmware to the lowest rssi figure. As rx_status->chain_signal
is a signed byte the lowest possible value is -128 and not -256.
-256 was causing 0 to get stored in the signed byte.
Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
This change does the following:
1) Add a new 7265 series PCI ID
2) Add two new 3160 series PCI IDs
3) Add the new 3165 series PCI IDs and configurations
Signed-off-by: Oren Givon <oren.givon@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
With consolidated code, now we can add the required hooks for
OMAP5 to enable power management.
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
[nm@ti.com: minor rebase updates]
Signed-off-by: Nishanth Menon <nm@ti.com>
Reviewed-by: Kevin Hilman <khilman@linaro.org>
Tested-by: Kevin Hilman <khilman@linaro.org>
On OMAP5 / DRA7, prevent a CPU powerdomain OFF and resulting MPU OSWR
and instead attempt a CPU RET and side effect, MPU RET in suspend.
NOTE: the hardware was originally designed to be capable of achieving
deep power states such as OFF and OSWR, however due to various issues
and risks, deepest valid state was determined to be CSWR - hence we use
the errata framework to handle this case.
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
[nm@ti.com: updates]
Signed-off-by: Nishanth Menon <nm@ti.com>
Reviewed-by: Kevin Hilman <khilman@linaro.org>
Tested-by: Kevin Hilman <khilman@linaro.org>
Dont assume that all OMAP4+ code will be able to use OMAP4 hotplug
logic. On OMAP5, DRA7, we do not need this in place yet, also,
currently the CPU startup pointer is located in omap4_cpu_pm_info
instead of cpu_pm_ops.
So, isolate the function to hotplug_restart pointer in cpu_pm_ops
where it should have belonged, initalize them as per valid startup
pointers for OMAP4430/60 as in current logic, however provide
dummy_cpu_resume to be the startup location as well.
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
[nm@ti.com: split this out of original code and isolate it]
Signed-off-by: Nishanth Menon <nm@ti.com>
Reviewed-by: Kevin Hilman <khilman@linaro.org>
Tested-by: Kevin Hilman <khilman@linaro.org>
Get rid of all assumptions about always having a sar base on *all*
OMAP4+ platforms. We dont need one on DRA7 and it is not necessary at
this point for OMAP5 either.
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
[nm@ti.com: Split and optimize]
Signed-off-by: Nishanth Menon <nm@ti.com>
Reviewed-by: Kevin Hilman <khilman@linaro.org>
Tested-by: Kevin Hilman <khilman@linaro.org>
In addition to the standard power-management technique, the OMAP5 / DRA7
MPU subsystem also employs an SR3-APG (mercury) power management
technology to reduce leakage.
It allows for full logic and memories retention on MPU_C0 and MPU_C1 and
is controlled by the PRCM_MPU. Only "Fast-mode" is supported on the
OMAP5 and DRA7 family of processors.
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
[nm@ti.com: minor consolidation]
Signed-off-by: Nishanth Menon <nm@ti.com>
Reviewed-by: Kevin Hilman <khilman@linaro.org>
Tested-by: Kevin Hilman <khilman@linaro.org>
Enables MPUSS ES2 power management mode using ES2_PM_MODE in
AMBA_IF_MODE register.
0x0: OMAP5 ES1 behavior, CPU cores would enter and exit OFF mode together.
Broken! Fortunately, we do not support this anymore.
0x1: OMAP5 ES2, DRA7 behavior, CPU cores are allowed to enter/exit OFF mode
independently.
This is one time settings thanks to always ON domain.
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
[nm@ti.com: minor conflict resolutions, consolidation for DRA7]
Signed-off-by: Nishanth Menon <nm@ti.com>
Reviewed-by: Kevin Hilman <khilman@linaro.org>
Tested-by: Kevin Hilman <khilman@linaro.org>
With EMIF clock-domain put under hardware supervised control, memory
corruption and untraceable crashes are observed on OMAP5. Further
investigation revealed that there is a weakness in the PRCM on this
specific dynamic depedency.
The recommendation is to set MPUSS static dependency towards EMIF
clock-domain to avoid issues. This recommendation holds good for DRA7
family of devices as well.
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
[rnayak@ti.com: DRA7]
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
[nm@ti.com: conflict resolution, dra7]
Signed-off-by: Nishanth Menon <nm@ti.com>
Reviewed-by: Kevin Hilman <khilman@linaro.org>
Tested-by: Kevin Hilman <khilman@linaro.org>
On OMAP5, RM_CPUi_CPUi_CONTEXT offset has changed. Update the code
so that same code works for OMAP4+ devices. DRA7 and OMAP5 have the same
context offset as well.
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
[rnayak@ti.com: for DRA7]
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
[nm@ti.com: rebase, split/merge etc..]
Signed-off-by: Nishanth Menon <nm@ti.com>
Reviewed-by: Kevin Hilman <khilman@linaro.org>
Tested-by: Kevin Hilman <khilman@linaro.org>
Not all SoCs support OFF mode - for example DRA74/72. So, use valid
power state during CPU hotplug.
Signed-off-by: Nishanth Menon <nm@ti.com>
Reviewed-by: Kevin Hilman <khilman@linaro.org>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
We are using power domain state as RET and logic state as OFF. This
state is OSWR. This may not always be supported on ALL power domains. In
fact, on certain power domains, this might result in a hang on certain
platforms. Instead, depend on powerdomain data to provide accurate
information about the supported powerdomain states and use the
appropriate function to query and use it as part of suspend path.
Signed-off-by: Nishanth Menon <nm@ti.com>
Reviewed-by: Kevin Hilman <khilman@linaro.org>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Move the logic state as different for each power domain. This allows us
to customize the deepest power state we should target over all for each
powerdomain in the follow on patches.
Signed-off-by: Nishanth Menon <nm@ti.com>
Reviewed-by: Kevin Hilman <khilman@linaro.org>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
powerdomain configuration in OMAP is done using PWRSTCTRL register for
each power domain. However, PRCM lets us write any value we'd like to
the logic and power domain target states, however the SoC integration
tends to actually function only at a few discrete states. These valid
states are already in our powerdomains_xxx_data.c file.
So, provide a function to easily query valid low power state that the
power domain is allowed to go to.
Based on work originally done by Jean Pihet <j-pihet@ti.com>
https://patchwork.kernel.org/patch/1325091/ . There is no attempt to
create a new powerdomain solution here, except fixing issues seen
attempting invalid programming attempts. Future consolidation to the
generic powerdomain framework should consider this requirement as
well.
Similar solutions have been done in product kernels in the past such
as:
https://android.googlesource.com/kernel/omap.git/+blame/android-omap-panda-3.0/arch/arm/mach-omap2/pm44xx.c
Signed-off-by: Nishanth Menon <nm@ti.com>
Reviewed-by: Kevin Hilman <khilman@linaro.org>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
No need to invoke callback when the clkdm pointer is NULL.
Signed-off-by: Nishanth Menon <nm@ti.com>
Reviewed-by: Kevin Hilman <khilman@linaro.org>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Update the power domain power states for final production chip
capability. OFF mode, OSWR etc have been descoped for various domains.
Signed-off-by: Nishanth Menon <nm@ti.com>
Reviewed-by: Kevin Hilman <khilman@linaro.org>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Commit 3b29970909 "nfsd4: enforce rd_dircount" totally misunderstood
rd_dircount; it refers to total non-attribute bytes returned, not number
of directory entries returned.
Bring the code into agreement with RFC 3530 section 14.2.24.
Cc: stable@vger.kernel.org
Fixes: 3b29970909 "nfsd4: enforce rd_dircount"
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
A block_device may be attached to different gendisks and thus
different bdis over time. bdev_inode_switch_bdi() is used to switch
the associated bdi. The function assumes that the inode could be
dirty and transfers it between bdis if so. This is a bit nasty in
that it reaches into bdi internals.
This patch reimplements the function so that it writes out the inode
if dirty. This is a lot simpler and can be implemented without
exposing bdi internals.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jens Axboe <axboe@fb.com>
bdi_destroy() has code to transfer the remaining dirty inodes to the
default_backing_dev_info; however, given the shutdown sequence, it
isn't clear how such condition would happen. Also, it isn't a full
solution as the transferred inodes stlil point to the bdi which is
being destroyed. Operations on those inodes can end up accessing
already released fields such as the percpu stat fields.
Digging through the history, it seems that the code was added as a
quick workaround for a bug report without fully root-causing the
issue. We probably want to remove the code in time but for now let's
add a comment noting that it is a quick workaround.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@fb.com>
Canceling of bdi->wb.dwork is currently a bit mushy.
bdi_wb_shutdown() performs cancel_delayed_work_sync() at the end after
shutting down and flushing the delayed_work and bdi_destroy() tries
yet again after bdi_unregister().
bdi->wb.dwork is queued only after checking BDI_registered while
holding bdi->wb_lock and bdi_wb_shutdown() clears the flag while
holding the same lock and then flushes the delayed_work. There's no
way the delayed_work can be queued again after that.
Replace the two unnecessary cancel_delayed_work_sync() invocations
with WARNs on pending. This simplifies and clarifies the code a bit
and will help future changes in further isolating bdi_writeback
handling.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@fb.com>
The only places where NULL test on bdi->dev is used are
bdi_[un]register(). The functions can't be called in parallel anyway
and there's no point in protecting bdi->dev clearing with a lock.
Remove bdi->wb_lock grabbing around bdi->dev clearing and move it
after device_unregister() call so that bdi->dev doesn't have to be
cached in a local variable.
This patch shouldn't introduce any behavior difference.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@fb.com>
Two flags and one bdi_writeback field are no longer used. Remove
them.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@fb.com>
bdev_get_queue() returns the request_queue associated with the
specified block_device. blk_get_backing_dev_info() makes use of
bdev_get_queue() to determine the associated bdi given a block_device.
All the callers of bdev_get_queue() including
blk_get_backing_dev_info() assume that bdev_get_queue() may return
NULL and implement NULL handling; however, bdev_get_queue() requires
the passed in block_device is opened and attached to its gendisk.
Because an active gendisk always has a valid request_queue associated
with it, bdev_get_queue() can never return NULL and neither can
blk_get_backing_dev_info().
Make it clear that neither of the two functions can return NULL and
remove NULL handling from all the callers.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Chris Mason <clm@fb.com>
Cc: Dave Chinner <david@fromorbit.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
blkcg->id is a unique id given to each blkcg; however, the
cgroup_subsys_state which each blkcg embeds already has ->serial_nr
which can be used for the same purpose. Drop blkcg->id and replace
its uses with blkcg->css.serial_nr. Rename cfq_cgroup->blkcg_id to
->blkcg_serial_nr and @id in check_blkcg_changed() to @serial_nr for
consistency.
Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
Allow the PRM interrupt information to be picked up from device tree.
OMAP3 may use legacy boot and needs to be compatible with old dtbs
(without interrupt populated), for these, we use the value which is
pre-populated.
Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
OMAP5 and DRA7 can now use pinctrl based I/O daisychain wakeup
capability. So, enable the support.
Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
"wkup" event at bit offset 0 exists only on OMAP3.
OMAP4430/60 PRM_IRQSTATUS_A9, OMAP5/DRA7 PRM_IRQSTATUS_MPU
register bit 0 is DPLL_CORE_RECAL_ST not wakeup event like OMAP3.
The same applies to AM437x as well.
Remove the wrong definition.
Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Allow the PRM interrupt information to be picked up from device tree.
the only exception is for OMAP4 which uses values pre-populated and allows
compatibility with older dtb.
Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
use the generic function to pick up the prm_instance for a generic logic
which can be reused from OMAP4+
Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
PRM device instance can vary depending on SoC. We already handle the
same during reset of the device, However, this is also needed
for other logic instances. So, first abstract this out to a generic
function.
Signed-off-by: Nishanth Menon <nm@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Pull s390 fixes from Martin Schwidefsky:
"A bug fix for the vdso code, the loadparm for booting from SCSI is
added and the access permissions for the dasd module parameters are
corrected"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
s390/vdso: remove NULL pointer check from clock_gettime
s390/ipl: Add missing SCSI loadparm attributes to /sys/firmware
s390/dasd: Make module parameter visible in sysfs
Add quirks for XMOS based DACs for native DSD playback support using the new
DSD_U32_LE sample format.
This version adds native DSD support for:
- iFi Audio micro iDSD/nano iDSD (they use the same prod. id)
- DIYINHK USB to I2S/DSD converter
Changes from v2:
- fix and simplify switch statement
Changes from v1:
- use specific product id and alt setting per XMOS based device
[fixed a misc coding style issue by tiwai]
Signed-off-by: Jurgen Kramer <gtmkramer@xs4all.nl>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
XMOS based USB DACs with native DSD support expose this feature via a USB
alternate setting. The audio format is either 32-bit raw or a 32-bit PCM format.
To utilize this feature on linux this patch introduces a new 32-bit DSD
sampleformat DSD_U32_LE.
A follow up patch will add a quirk for XMOS based devices to utilize the new format.
Further patches will add support to alsa-lib.
Signed-off-by: Jurgen Kramer <gtmkramer@xs4all.nl>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The vblank waits in intel_tv_detect_type() are timing out for some
reason. This is a regression caused removing seemingly useless vblank
waits from the modeset seqeuence in:
commit 56ef52cad5
Author: Ville Syrjälä <ville.syrjala@linux.intel.com>
Date: Thu May 8 19:23:15 2014 +0300
drm/i915: Kill vblank waits after pipe enable on gmch platforms
So it turns out they weren't all entirely useless. Apparently the pipe
has to go through one full frame before we enable the TV port. Add a
vblank wait to intel_enable_tv() to make sure that happens.
Another approach was attempted by placing the vblank wait just after
enabling the port. The theory behind that attempt was that we need to
let the port stay enabled for one full frame before disabling it again
during load detection. But that didn't work, and we definitely must
have the vblank wait before enabling the port.
Cc: stable@vger.kernel.org
Cc: Alan Bartlett <ajb@elrepo.org>
Tested-by: Alan Bartlett <ajb@elrepo.org>
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=79311
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
If a memory block is not aligned to PAGE_SIZE, its base address must be
rounded up, not down, and its size must be reduced.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Grant Likely <grant.likely@linaro.org>
Simply swap of_alias and of_chosen initialization so
that of_alias ends up read first. This must be done
because it is accessed couple of lines below when
trying to initialize the of_stdout using the alias
based legacy method.
[Fixes a752ee5 - tty: Update hypervisor tty drivers to
use core stdout parsing code]
Signed-off-by: Laurentiu Tudor <Laurentiu.Tudor@freescale.com>
[glikely: Don't move the 'if (!of_aliases)' test]
Signed-off-by: Grant Likely <grant.likely@linaro.org>
This patch add a document that explains how the selftest test data is
dynamically attached into the live device tree irrespective of the
machine's architecture.
Signed-off-by: Gaurav Minocha <gaurav.minocha.os@gmail.com>
Signed-off-by: Grant Likely <grant.likely@linaro.org>
This patch is to the fix the recent runtime bug in kernel reported by
<fengguang.wu@intel.com>. The bug was exposed by commit b951f9dc,
"Enabling OF selftest to run without machine's devicetree" and is
exposed when CONFIG_OF_SELFTEST is enabled and CONFIG_SYSFS is
disabled.
Mail Subject: [OF test] BUG: unable to handle kernel NULL pointer
dereference at 00000038
Tested on x86 and arm architecture
Signed-off-by: Gaurav Minocha <gaurav.minocha.os@gmail.com>
Signed-off-by: Grant Likely <grant.likely@linaro.org>
The comments above of_console_check() say that it will return TRUE if it
registers a preferred console, but add_preferred_console() uses a
0-equals-success convention, so this leaves of_console_check() with an
inconsistent policy for its return values.
Fortunately, nobody was actually checking the return value of
of_console_check(), so this isn't significant at the moment.
But let's match the comments, so we're doing what we say.
Fixes: 3482f2c52b ('of: Create of_console_check() for selecting a console specified in /chosen')
Signed-off-by: Brian Norris <computersforpeace@gmail.com>
Signed-off-by: Grant Likely <grant.likely@linaro.org>
Raising the current maximum limit to 64. This is needed for Cavium's
Thunder systems that will have at least 48 cores per die.
The change keeps the current memory footprint in cpu mask structures.
It does not break existing code. Setting the maximum to 64 cpus still
boots systems with less cpus.
Mark's Juno happily booted with a NR_CPUS=64 kernel.
Tested on our Thunder system with 48 cores. We could see interrupts to
all cores.
Cc: Radha Mohan Chintakuntla <rchintakuntla@cavium.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Robert Richter <rrichter@cavium.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
