The fsafe value in the pingroup data tables is only used to implement
tegra_pinctrl_disable(). The only reason this function is called is when
dynamically switching between pinmux states, i.e. when disabling the old
state before programming the new state. It's simpler to have the new
target state define the expected value of each pin (and all current DTs
do that). This also gives more flexibility, since it allows individual
boards explicit control over the "inactive" mux function for each pin,
rather than requiring it to be an SoC-specific value. Assuming this, we
can get rid of the fsafe value from the driver completely, thus saving
some more space in the driver tables.
While re-writing the content of tegra124_pingroups[], fix the indentation
to use a TAB instead of spaces.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Laxman Dewangan <ldewangan@nvidia.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Any SoC which supports the einput, odrain, lock, ioreset, or rcv_sel
options has the relevant HW register fields in the same register as the
mux function selection. Similarly, the drvtype option is always in the
drive register, if it is supported at all. Hence, we don't need to have
struct *_reg fields in the pin group table to define which register and
bank to use for those options. Delete this to save space in the driver's
data tables.
However, many of those options are not supported on all SoCs, or not
supported on some pingroups. We need a way to detect when they are
supported. Previously, this was indicated by setting the struct *_reg
field to -1. With the struct *_reg fields removed, we use the struct
*_bit fields for this purpose instead. The struct *_bit fields need to
be expanded from 5 to 6 bits in order to store a value outside the valid
HW bit range of 0..31.
Even without removing the struct *_reg fields, we still need to add code
to validate the struct *_bit fields, since some struct *_bit fields were
already being set to -1, without an option-specific struct *_reg field to
"guard" them. In other words, before this change, the pinmux driver might
allow some unsupported options to be written to HW.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
The patch "mac80211: implement SMPS for AP" has caused kernel
oops at mesh STA if the peer mesh STA operates in sleep mode
and then becomes active mode. It can be easily reproduced by
setting the following commands at peer mesh STA:
iw mesh0 station set aa:bb:cc:dd:ee:ff mesh_power_mode deep
iw mesh0 station set aa:bb:cc:dd:ee:ff mesh_power_mode active
Kernel oops will happen at mesh STA aa:bb:cc:dd:ee:ff.
Fix this by avoiding SMPS for mesh mode.
Signed-off-by: Chun-Yeow Yeoh <yeohchunyeow@gmail.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
-------------------------
Linux does not use the ACC and F bits of the storage key. Newer Linux
versions also do not use the storage keys for dirty and reference
tracking. We can optimize the guest handling for those guests for faults
as well as page-in and page-out by simply not caring about the guest
visible storage key. We trap guest storage key instruction to enable
those keys only on demand.
Migration bitmap
Until now s390 never provided a proper dirty bitmap. Let's provide a
proper migration bitmap for s390. We also change the user dirty tracking
to a fault based mechanism. This makes the host completely independent
from the storage keys. Long term this will allow us to back guest memory
with large pages.
per-VM device attributes
------------------------
To avoid the introduction of new ioctls, let's provide the
attribute semanantic also on the VM-"device".
Userspace controlled CMMA
-------------------------
The CMMA assist is changed from "always on" to "on if requested" via
per-VM device attributes. In addition a callback to reset all usage
states is provided.
Proper guest DAT handling for intercepts
----------------------------------------
While instructions handled by SIE take care of all addressing aspects,
KVM/s390 currently does not care about guest address translation of
intercepts. This worked out fine, because
- the s390 Linux kernel has a 1:1 mapping between kernel virtual<->real
for all pages up to memory size
- intercepts happen only for a small amount of cases
- all of these intercepts happen to be in the kernel text for current
distros
Of course we need to be better for other intercepts, kernel modules etc.
We provide the infrastructure and rework all in-kernel intercepts to work
on logical addresses (paging etc) instead of real ones. The code has
been running internally for several months now, so it is time for going
public.
GDB support
-----------
We provide breakpoints, single stepping and watchpoints.
Fixes/Cleanups
--------------
- Improve program check delivery
- Factor out the handling of transactional memory on program checks
- Use the existing define __LC_PGM_TDB
- Several cleanups in the lowcore structure
- Documentation
NOTES
-----
- All patches touching base s390 are either ACKed or written by the s390
maintainers
- One base KVM patch "KVM: add kvm_is_error_gpa() helper"
- One patch introduces the notion of VM device attributes
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)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=jxBp
-----END PGP SIGNATURE-----
Merge tag 'kvm-s390-20140422' of git://git.kernel.org/pub/scm/linux/kernel/git/kvms390/linux into queue
Lazy storage key handling
-------------------------
Linux does not use the ACC and F bits of the storage key. Newer Linux
versions also do not use the storage keys for dirty and reference
tracking. We can optimize the guest handling for those guests for faults
as well as page-in and page-out by simply not caring about the guest
visible storage key. We trap guest storage key instruction to enable
those keys only on demand.
Migration bitmap
Until now s390 never provided a proper dirty bitmap. Let's provide a
proper migration bitmap for s390. We also change the user dirty tracking
to a fault based mechanism. This makes the host completely independent
from the storage keys. Long term this will allow us to back guest memory
with large pages.
per-VM device attributes
------------------------
To avoid the introduction of new ioctls, let's provide the
attribute semanantic also on the VM-"device".
Userspace controlled CMMA
-------------------------
The CMMA assist is changed from "always on" to "on if requested" via
per-VM device attributes. In addition a callback to reset all usage
states is provided.
Proper guest DAT handling for intercepts
----------------------------------------
While instructions handled by SIE take care of all addressing aspects,
KVM/s390 currently does not care about guest address translation of
intercepts. This worked out fine, because
- the s390 Linux kernel has a 1:1 mapping between kernel virtual<->real
for all pages up to memory size
- intercepts happen only for a small amount of cases
- all of these intercepts happen to be in the kernel text for current
distros
Of course we need to be better for other intercepts, kernel modules etc.
We provide the infrastructure and rework all in-kernel intercepts to work
on logical addresses (paging etc) instead of real ones. The code has
been running internally for several months now, so it is time for going
public.
GDB support
-----------
We provide breakpoints, single stepping and watchpoints.
Fixes/Cleanups
--------------
- Improve program check delivery
- Factor out the handling of transactional memory on program checks
- Use the existing define __LC_PGM_TDB
- Several cleanups in the lowcore structure
- Documentation
NOTES
-----
- All patches touching base s390 are either ACKed or written by the s390
maintainers
- One base KVM patch "KVM: add kvm_is_error_gpa() helper"
- One patch introduces the notion of VM device attributes
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Conflicts:
include/uapi/linux/kvm.h
We need to check whether drm_ht_create succeed and clean up
if not.
Spotted by coverity.
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
This is C standard hair-splitting, but afaict
- sum will be promoted to signed int in computation since
uint8_t fits
- signed overflow is undefined.
No we need to add up an awful lot of bytes to actually make it
overflow. But I guess the real risk is gcc spotting this and going
bananas. Fix this by simply using unsigned in to force all computations
to use the well-defined unsigned behaviour.
Spotted by coverity.
v2: Simplify the entire computation as suggested by Jean.
Cc: Russell King <rmk+kernel@arm.linux.org.uk>
Cc: Rob Clark <robdclark@gmail.com>
Cc: Jean-Francois Moine <moinejf@free.fr>
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The ->gem_free_object never gets called with a NULL pointer, the check
is redundant. Also checking after the upcast allows compilers to elide
it anyway.
Noticed while chasing coverity reports, somehow this one here was not
flagged.
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
ttm_bo_unref unconditionally calls kref_put on it's argument, so the
thing can't be NULL without already causing Oopses.
Spotted by coverity.
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We need to set it to -ENODEV when we don't recognize the device.
Otherwise we return/print stack garbage.
Spotted by coverity.
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The context_dtor callback is only called once we've successfully loaded
the driver, which means dev->dev_private is set up. The check is hence
pointless.
Also dev->dev_private is deref already above, so compilers are free
to elide it anyway.
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The ->gem_free_object never gets called with a NULL pointer, the check
is redundant. Also checking after the upcast allows compilers to elide
it anyway.
Spotted by coverity.
v2: Fix patch subject.
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
ttm_bo_unref unconditionally calls kref_put on it's argument, so the
thing can't be NULL without already causing Oopses.
Spotted by coverity.
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The ->gem_free_object never gets called with a NULL pointer, the check
is redundant. Also checking after the upcast allows compilers to elide
it anyway.
Spotted by coverity.
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
ttm_bo_unref unconditionally calls kref_put on it's argument, so the
thing can't be NULL without already causing Oopses.
Spotted by coverity.
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The ->gem_free_object never gets called with a NULL pointer, the check
is redundant. Also checking after the upcast allows compilers to elide
it anyway.
Spotted by coverity.
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
ttm_bo_unref unconditionally calls kref_put on it's argument, so the
thing can't be NULL without already causing Oopses.
Spotted by coverity.
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Add the muxing for the last missing i2c rcar core. Fix the sorting for
SH_PFC_PIN_NAMED while we are here.
Signed-off-by: Wolfram Sang <wsa@sang-engineering.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
With debug enabled we get better readability dumps of the mux_mode register if
we use hexadecimal format instead:
imx6sl-pinctrl 20e0000.iomuxc: MX6SL_PAD_FEC_REF_CLK: 0x10 0x0001b0a8
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: Shawn Guo <shawn.guo@freescale.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
In the TB10x pin database, a port index of -1 is used to indicate
unmuxed GPIO pin groups. This bug fixes a 'cast to unsigned' bug of
this value.
Thanks to Dan Carpenter for highlighting this.
CC: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Christian Ruppert <christian.ruppert@abilis.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Should be 5 rather than 4.
Noticed-by: Mathias Fröhlich <Mathias.Froehlich@gmx.net>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Christian König <christian.koenig@amd.com>
After moving the IO layer inside ASoC to the component level we can now easily
move the standard control helpers also to the component level. This allows to
reuse the same standard helper control implementations for other components.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@linaro.org>
The ASoC framework is in the process of migrating all IO operations to regmap.
regmap has its own more sophisticated tracing infrastructure for IO operations,
which means that the ASoC level IO tracing becomes redundant, hence this patch
removes them. There are still a handful of ASoC drivers left that do not use
regmap yet, but hopefully the removal of the ASoC IO tracing will be an
additional incentive to switch to regmap.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@linaro.org>
There is no unlocked version of soc_widget_update_bits_locked() and there is no
plan to introduce it in the near future, so drop the _locked suffix.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@linaro.org>
There are no users of snd_soc_update_bits_locked() left and it is identical to
snd_soc_update_bits(). So it can be removed.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@linaro.org>
File-private locks have been merged into Linux for v3.15, and *now*
people are commenting that the name and macro definitions for the new
file-private locks suck.
...and I can't even disagree. The names and command macros do suck.
We're going to have to live with these for a long time, so it's
important that we be happy with the names before we're stuck with them.
The consensus on the lists so far is that they should be rechristened as
"open file description locks".
The name isn't a big deal for the kernel, but the command macros are not
visually distinct enough from the traditional POSIX lock macros. The
glibc and documentation folks are recommending that we change them to
look like F_OFD_{GETLK|SETLK|SETLKW}. That lessens the chance that a
programmer will typo one of the commands wrong, and also makes it easier
to spot this difference when reading code.
This patch makes the following changes that I think are necessary before
v3.15 ships:
1) rename the command macros to their new names. These end up in the uapi
headers and so are part of the external-facing API. It turns out that
glibc doesn't actually use the fcntl.h uapi header, but it's hard to
be sure that something else won't. Changing it now is safest.
2) make the the /proc/locks output display these as type "OFDLCK"
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Carlos O'Donell <carlos@redhat.com>
Cc: Stefan Metzmacher <metze@samba.org>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Frank Filz <ffilzlnx@mindspring.com>
Cc: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
We currently have two very similar IO abstractions in ASoC, one for CODECs, the
other for platforms. Moving this to the component level will allow us to unify
those two. It will also enable us to move the standard kcontrol helpers as well
as DAPM support to the component level.
The new component level abstraction layer is primarily build around regmap.
There is a per component pointer for the regmap instance for the underlying
device. There are four new function snd_soc_component_read(),
snd_soc_component_write(), snd_soc_component_update_bits() and
snd_soc_component_update_bits_async(). They have the same signature as their
regmap counter-part and will internally forward the call one-to-one to regmap.
If the component it not using regmap it will fallback to using the custom IO
callbacks. This is done to be able to support drivers that haven't been
converted to regmap yet, but it is expected that this will eventually be removed
in the future once all component drivers have been converted to regmap.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@linaro.org>
Some I2C adapters are only compatible with the SMBus protocol and do not
support standard I2C transfers.
Fallback to SMBus transfers if we encounter such kind of adapters.
The transfer type is chosen according to the val_bits field in the regmap
config.
Signed-off-by: Boris BREZILLON <boris.brezillon@free-electrons.com>
Signed-off-by: Mark Brown <broonie@linaro.org>
The Allwinner A31 has support for four high speed timers. Apart for the
number of timers (4 vs 2), it's basically the same logic than the high
speed timers found in the sun5i chips.
Now that we have a driver to support it, we can enable them in the
device tree.
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
The Allwinner A31 that uses this timer has the timer IP asserted in reset.
Add an optional reset property to the DT, and deassert the timer from reset if
it's there.
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
We have been using rt5640.c codec driver with RT5642 codec chip before commit
022d21f004 ("ASoC: rt5640: add rt5639 support"). That commits starts using
device ID reading in reset register for adding device specific controls and
routes runtime.
Now since device ID appears to be different between RT5640 and RT5642 the
driver doesn't add those controls and routes that are valid also on RT5642.
Fix this by adding a device ID found by debugging and minimal code for
supporting RT5642.
Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Signed-off-by: Bard Liao <bardliao@realtek.com>
Signed-off-by: Mark Brown <broonie@linaro.org>
Remove the cleanup code related to the platform from the DAI drivers at the
same time to avoid breakage.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Mark Brown <broonie@linaro.org>
The A31 SoC has PL and PM banks and thus increase the default ARCH_NR_GPIO.
Signed-off-by: Boris BREZILLON <boris.brezillon@free-electrons.com>
Acked-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
The A31 SoC define a reset line for the R_PIO block which needs to be
deasserted.
Try to retrieve a reset control and deassert if one was found.
Signed-off-by: Boris BREZILLON <boris.brezillon@free-electrons.com>
Acked-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
The A31 SoC provides both PL and PM pio bank through the R_PIO block.
These pins all support gpio function and can bbe assigned to system
peripherals (like TWI, P2WI, JTAG, ...)
Signed-off-by: Boris BREZILLON <boris.brezillon@free-electrons.com>
Acked-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Add support for multiple pin controller instances.
First remove the static definition of the sunxi gpio chip struct and fill
the dynamically struct instead.
Then define a new pin_base field in the sunxi_pinctrl_desc which will be
used to specify the gpiochip base pin.
Signed-off-by: Boris BREZILLON <boris.brezillon@free-electrons.com>
Acked-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Disable the clk when failing to probe the pin controller device.
Signed-off-by: Boris BREZILLON <boris.brezillon@free-electrons.com>
Acked-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Check the clk_prepare_enable return value to avoid false positive probe.
Signed-off-by: Boris BREZILLON <boris.brezillon@free-electrons.com>
Acked-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
In commit
commit 6375b768a9
Author: Ville Syrjälä <ville.syrjala@linux.intel.com>
Date: Mon Mar 3 11:33:36 2014 +0200
drm/i915: Reject >165MHz modes w/ DVI monitors
the driver started to filter out display modes which exceed the
single-link DVI 165Mz dotclock limits when the monitor doesn't report
itself as being HDMI compliant. The intent was to filter out all
EDID derived modes that require dual-link DVI to operate since we
don't support dual-link.
However the patch went a bit too far and also causes the driver to reject
such modes even when specified by the user. Normally we don't check the
sink limitations when setting a mode from the user. This allows the user
to specify any mode whether the sink reports to support it or not. This
can be useful since often the sinks support more modes than they report
in the EDID.
So relax the checks a bit, and apply the single-link DVI dotclock limit
only when filtering the mode list, and ignore the limit when setting
a user specified mode.
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=72961
Tested-by: Nicholas Vinson <nvinson@comcast.net>
Cc: stable@vger.kernel.org [3.14]
Reviewed-by: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Factor out the new function handle_itdb(), which copies the ITDB into
guest lowcore to fully handle a TX abort.
Signed-off-by: Michael Mueller <mimu@linux.vnet.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
The generically assembled low core labels already contain the
address for the TDB.
Signed-off-by: Michael Mueller <mimu@linux.vnet.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
On hard exits (abort, sigkill) we have have some kvm_s390_interrupt_info
structures hanging around. Delete those on exit to avoid memory leaks.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
CC: stable@vger.kernel.org
Reviewed-by: Thomas Huth <thuth@linux.vnet.ibm.com>
