Fix typo where logical AND was used instead of bitwise AND.
Change-Id: Ic184559541646f87052b78d54efedcd1b0a2ce40
Reported-by: David Binderman <linuxdev.baldrick@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 9ebf1945d757433a089ab3ee940673503e3e11ec
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
- The global lock was used in cec_get_device when it should have
used the devnode lock.
- cec_put_device also took the global lock, but since the release
function takes that lock as well this could lead to a deadlock.
Just don't take the lock here since there is no reason for it.
- cec_devnode_register() should take the global lock when clearing
the bit in the global bitmap.
- In cec_devnode_unregister() place the devnode->(un)register tests
and assignments under the devnode lock as well: this has to be
in a critical block.
Change-Id: I229e6136937003f99153028d12780c0162620bef
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 2ab25d35a91098ef0f42d478cc37f6a5591a4ab0
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
This lock will be used to protect more than just the fhs list.
So rename it to just 'lock'.
Change-Id: I6a2623b1231154be2a4d764ad2086fc73015a3af
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 62148f0930a8e9bd5c5614f8387222f0220d7d47
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
The unused bytes of the features array should be zeroed, but the start index was one
byte too early. This caused the device features byte to be overwritten by 0.
The compliance test for the CEC_S_LOG_ADDRS ioctl didn't catch this because it tested
byte continuation with the second device features byte being 0 :-(
Change-Id: Ibe07ad9376fa7bebed57c99892ce0e5520cfa170
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 292eaf50c7df4ae2ae8aaa9e1ce3f1240a353ee8
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
Even when the adapter is configured for CEC 1.4, we still check and
use the CEC 2.0 parts of struct cec_log_addrs. Although these aren't
used in CEC messages, the information contained in them is still of
use in the CEC framework itself, so keep this information.
Also zero the unused trailing features[] data and unused logical address
data so the contents isn't random data.
Change-Id: Id8f226b730188b72abe0e81a50d540beee9418e2
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 009a620848218d521f008141c62f56bf19294dd9
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
For POLLOUT poll only checked if the adapter was configured, not
if there was room in the transmit queue. Add that check.
Change-Id: I56fc0585999f4c5ac313305fe43c54a2a1592879
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: b7cbc892532316f2abd880186301e57239a17f28
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
The main message loop checks if the physical address was valid, and if
not it is assumed that the adapter had been unconfigured.
However, this check is no longer correct, instead it should check
that both adap->is_configured and adap->is_configuring are false.
Change-Id: I178c46bf4e7f2324786a26e4c50d31899c866bc6
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 58ecc29b5e829c2072fe3a98a93ef163bdb36dd9
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
The size of the transmit queue was unlimited, which meant that
in non-blocking mode you could flood the CEC adapter with messages
to be transmitted.
Limit this to 18 messages.
Also print the number of pending transmits and the timeout value
in the status debugfs file.
Change-Id: I5cdadee7854e62d5a823b7dc53448860ba73228a
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 11065f8531ed33d5e847a36ad64231ee599d4996
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
Ensure that the unused part of the msg array is zeroed. This is
required by CEC 2.0 when receiving shorter messages than expected.
In that case the remaining bytes are assumed to be 0.
There are no such CEC messages yet, but it is required to be future proof.
And since we're doing it for received messages, do it for transmitted
messages as well. It's unambiguous this way.
Change-Id: I780888724448c24f432c5d9753d6c5289f725682
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 045344c3c1478f8d03a1a589d645fe10c3a23f77
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
The filehandle was set to NULL when in non-blocking mode or when
no reply is needed.
This is wrong: the filehandle is needed in non-blocking mode to ensure
that the result of the transmit can be obtained through CEC_RECEIVE.
And the 'reply' check was also incorrect since it should have checked the
timeout field (the reply can be 0). In any case, when in blocking mode
there is no need to set the fh to NULL either.
Change-Id: I8225959283d34f9a44439ab34ee291b87b5d4922
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 4eef404fd91543dc82be2bd05d3b29919ec6c0e6
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
Before transmitting a message clear all status fields and always fill
in the sequence number. Make sure the sequence number is never 0.
Change-Id: Ia8ad9b79935cad0e88371f00137a59326a336ffc
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 40df3a7e586df59190547fe0e7d9928da735eada
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
When CEC_RECEIVE returns a message the original timeout field
was overwritten. Restore the timeout field.
Change-Id: Ida65e37e1424745e3e4cf9d0108e10dc75e642ed
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: e883b4d02670ae83ab24e9af969de1af756959f8
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
The CEC documentation has been converted to sphinx, so this
TODO item can be dropped.
Change-Id: I7a0fa38382a5533fd9e3643613d63595881b2b78
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: bddb4952506d0e84c443a04840143ea98b610557
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
Attempts to send CEC messages to yourself are detected in the framework and
returned with a NACK error. However, the tx_ts was never filled in that case.
So just set it.
Change-Id: Ifd8be78030778d3d99b88b353905879c2079c2c5
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 9a3f14ea802fd0d8848d8e16566b769e4ce09d68
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
There is really no reason to zero the reply and timeout fields if an
error occurs. This is a left-over from earlier versions where that
was used to signal errors, but this is now handled by the rx/tx_status
fields.
Change-Id: I2064f6b2dcb3affe4097ebdf7270eb8c6bc7e21d
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 86e3577f11d0aa8a5d3c6199f9e3f23cdbd0c0ac
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
When transmitting a message and waiting for a reply it would be good
to know the time between when the message was transmitted and when
the reply arrived. With only one timestamp field it was set to when
the reply arrived and the original transmit time was overwritten.
Just taking the timestamp in userspace right before CEC_TRANSMIT is
called is not reliable, since the actual transmit can be delayed if
the CEC bus is busy. Only the driver can fill this in accurately.
So split up the ts field into an rx_ts and a tx_ts. Also move the
status fields to after the 'reply' field: they were placed in a
strange position and make much more sense when grouped with the
other status-related fields.
This patch also makes sure that the timestamp is taken as soon as
possible.
Change-Id: I643006cc32cbc6073b6ce8fa7ae03b63b3b0e266
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 980e0b36b55985dd0a37acd25d29c8f21482db32
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
Check (and warn) if the msg->len is too long or if it is 0.
Should never happen, but just in case...
Change-Id: I784aca0acdc74511b42b51ae01e57be0c04799b1
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 52d802d65b56a399aa67a9d3f26f232116d3e256
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
When the device is in monitor mode, then you should always be able to
call CEC_RECEIVE, even it the device is unconfigured.
Change-Id: I1d11a8cbb26d19197b7f3a4a77ed8d6ce00f1853
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: ca00033c9fc8b5f178573a76887c89296ed5b398
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
CEC USB dongles can be unplugged at any time, and at that point they will
be unregistered. Make sure that any attempt afterwards to set the physical
or logical addresses will be ignored.
Change-Id: If15d1e3f3447c6a844eaee27cfdfd65a3341062e
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: c000e5dad17d20c644bbf3ead52c334025d11ada
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
The cec_msg_request_short_audio_descriptor function was missing the reply
argument. That's needed if you want the framework to wait for the reply
message.
Change-Id: Id099aa3ec932beb7d49af56cd9f761ee66b2727e
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 9ad52b4db79d168867a2ca105eca00fb9cb28fe5
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
The cec_ops_* functions never touch cec_msg, so mark it as const.
This was done for some of the cec_ops_ functions, but not all.
Change-Id: Ibf4c85faac86588e937107bd1bc7648841060bab
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 0c1f8569942695b7a2735301efe45ecd69e2a41f
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
- Use IS_REACHABLE(RC_CORE) instead of IS_ENABLED: if cec is built-in and
RC_CORE is a module, then CEC can't reach the RC symbols.
- Both cec and cec-edid should be bool and use the same build 'mode' as
MEDIA_SUPPORT (just as is done for the media controller code).
- Add a note to staging that this should be changed once the cec framework
is moved out of staging.
Change-Id: If084eab2c2bbca7b9fa5b8959b27b1894b8ec2b7
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reported-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 5bb2399a4fe4a1703c1497b4838c5e8e869d0822
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
The MEDIA_CEC_EDID option is guarded by MEDIA_SUPPORT, so selecting
it from MEDIA_CEC produces a warning:
warning: (MEDIA_CEC) selects MEDIA_CEC_EDID which has unmet direct dependencies (MEDIA_SUPPORT)
The warning is harmless, but it's better to add an explicit
dependency to shut it up, to reduce the noise during randconfig
builds.
Fixes: ca684386e6e2 ("[media] cec: add HDMI CEC framework (api)")
Change-Id: I637ceabfcc8f311f8ca723671f954e1db060c965
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: ea8c535e30c1066a0bf8acb2d39bb3c4e643a034
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
CEC_MSG_REPORT_PHYSICAL_ADDR can theoretically be received from
an unregistered device, but in that case the code should not attempt
to write the received physical address to the phys_addrs array.
That would be pointless since there can be multiple unregistered
devices that report a physical address. We just ignore those.
While at it, improve the dprintk since it would attempt to read
from that array as well with the same out-of-bounds problem.
Change-Id: Ic4e969669cadd3e9f59a587d7d075e130f5cd869
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: f8db65fe4336baf818dde5d226eb3d35773e2371
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
Like the videodev2.h and other headers, explicitly allow these headers
to be used with the BSD license.
Change-Id: I8f9ae52ed90337ca1f697dbf310a774f50eb154d
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 12655445a2adc292db3352bbe80e43bd243517aa
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
When a reply to an earlier transmit is received, the tx_status of that
transmit needs to be restored. Otherwise the status that the transmit
was successful would be lost.
Change-Id: I53fac6d2346a88a8f93503952bd9e2c3c9bae5e6
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: 7b12b47f78e458ea90cc350bc7a4b9385eef897a
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
The added HDMI CEC framework provides a generic kernel interface for
HDMI CEC devices.
Note that the CEC framework is added to staging/media and that the
cec.h and cec-funcs.h headers are not exported yet. While the kABI
is mature, I would prefer to allow the uABI some more time before
it is mainlined in case it needs more tweaks.
This adds the cec-api.c source that deals with the public CEC API
and the Kconfig/Makefile plumbing.
The MAINTAINERS file is also updated.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
[k.debski@samsung.com: code cleanup and fixes]
Signed-off-by: Kamil Debski <kamil@wypas.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Git-commit: ca684386e6e21ba1511061f71577cdb6c3f2b3d3
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
Change-Id: Ie0ac63f18e51cab0e920f9c6c618563faf57a83c
Signed-off-by: Ray Zhang <rayz@codeaurora.org>
Specify short circuit (SC) interrupts for LAB/IBB so that the driver can
attempt re-enabling the module after PBS disables it for SC protection.
Interrupts are disabled by the driver if SC happens frequently so that
the module is kept disabled.
CRs-Fixed: 2002373
Change-Id: Idb6fdc4845aebb7133c210dd03ff77a167de5d69
Signed-off-by: Fenglin Wu <fenglinw@codeaurora.org>
Free already allocated buffer before allocating new one
to avoid memory leak when someone keep writing to this file
but never read it.
CRs-Fixed: 2051862
Change-Id: I29784fb30ea8eed315b7d073d35480972642bc55
Signed-off-by: Yuanyuan Liu <yuanliu@codeaurora.org>
Other than stats every debugfs entry should be under
CONFIG_ICNSS_DEBUG so that we don't enable these on
build which doesn't have debugging support.
CRs-Fixed: 2051871
Change-Id: I336f449313bb1f0f8af4e82f40869f65f58d9f4e
Signed-off-by: Yuanyuan Liu <yuanliu@codeaurora.org>
Currently vconn is getting disabled late after disabling vbus.
This is causing failure if type-C compliance TD.4.8.3(DRP Connect
Alternate Mode), as delay is disabling vconn goes above 35ms.
Fix the issue by moving vconn disable as first step upon receiving
disconnect notification. Also, making policy engine workqueue as high
priority workqueue to be able to run state machine work with lower
latency.
Change-Id: If9a979b4cc5bab3aa2ebf42397c36e5b434b50b9
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
Change debugfs permission to allow root only access to those files.
CRs-Fixed: 2051867
Change-Id: I4bef907f8cbe39169b38e4f19072a0628e199691
Signed-off-by: Yuanyuan Liu <yuanliu@codeaurora.org>
Fix a potential buffer overflow that could occur
during firmware download by storing buffer address
locally instead of inside MHI context.
CRs-Fixed: 1109397
Change-Id: If27ba602cdafdc8d25a94fdc0d74e8970bf0b0f4
Signed-off-by: Sujeev Dias <sdias@codeaurora.org>
Hardened user copy provides an extra layer of security checks when
copy_to_user/copy_from_user APIs are used to ensure data accesses are
valid. When invalid accesses are detected the kernel will forcibly
crash.
Change-Id: Ia0529c67da84017ea91a72427db008737d3c5dbd
Signed-off-by: Olav Haugan <ohaugan@codeaurora.org>
In transcode loopback session open, ALSA to ASM format conversion is
incorrect. Update the format based on ASM defined formats.
CRs-Fixed: 2049714
Change-Id: Ie5ddb6a58a30b0d6b78bfc68cbbbcd3d92f28fb1
Signed-off-by: Siddartha Shaik <sshaik@codeaurora.org>
Set voc path type according to stream type during routing
preparation for voice session to fix uninitialized voc
path type causing voc device port configuration failure.
CRs-fixed: 2045461
Change-Id: I4983ba55b1854879680b6d13c33904a522092eb0
Signed-off-by: Derek Chen <chenche@codeaurora.org>
There is a design limitation in host driver. Driver
sends regulatory_hint_user for both user-space and
11d country to kernel. So when wiphy re-registration
happens, kernel tells the source as USER even for 11d.
And per driver logic the 11d needs to be disabled
if INI gCountryCodePriority is set.
To mitigate this issue, storing the country source
in the platform driver.
CRs-Fixed: 2051651
Change-Id: I335a046a886ac3ce35cf96eb71231bfe75d33c60
Signed-off-by: Paul Zhang <paulz@codeaurora.org>
UHS-I SD cards support SDR104 mode which runs the SD card interface
clock upto 208 MHz. But we may see repeated CRC errors in SDR104
with some SDCC controllers. If this happens, commit <aafc430b36f3>
("mmc: sd: reduce the bus speed in case of multiple CRC errors") would
reinit the card to lower speed (SDR50) hoping that CRC error
rate would reduce at lower clock speed (100MHz for SDR50). As the error
may happen for any cmd, this change tries to fix several other paths -
clock scaling, mmc_rescan, non-data commands error path.
Change-Id: I8ccbbf7e330cde3862d9660d3be4d67fb00d97ca
Signed-off-by: Sahitya Tummala <stummala@codeaurora.org>
Signed-off-by: Subhash Jadavani <subhashj@codeaurora.org>
Signed-off-by: Veerabhadrarao Badiganti <vbadigan@codeaurora.org>
This reverts commit e6855ed3fe ("arm: cpu: read all
address cells in dt for cpuid"). The address cells read
issue is fixed in upstream commit c84cc7b906 ("ARM:
8616/1: dt: Respect property size when parsing CPUs"),
and both changes conflict and break the parsing of address
cells.
Change-Id: Iafd918ce8623ceb0c701df15bbdbccba165c89dd
Signed-off-by: Neeraj Upadhyay <neeraju@codeaurora.org>
Add support for required node for audio of HDK660/630 boards
with tasha codec.
Change-Id: I1b5acb70d4fb8e8e4958ffafb2b83c16df1d933c
CRs-Fixed: 2047107
Signed-off-by: Walter Yang <yandongy@codeaurora.org>
Currently PM QoS is not getting notifications if irq affinity
is changed by core isolation with irq_do_set_affinity.
Use irq_set_affinity_locked instead to get notifications when
updating affinity.
Change-Id: Iab745a23637d9353f730ca77ac7e92cf61b1bf67
Signed-off-by: Maulik Shah <mkshah@codeaurora.org>
ICL recovery mechanism is used by user space daemon
to recover from adapter collapse during voltage increment
event by reducing the USB ICL. There could be a situation
where nobody votes on usb_icl_votable and this could lead
to a negative ICL value.
Fix this by reading "HW_CURRENT_MAX" property to get ICL in
situations where usb_icl_votable is not voted by any client,
this ensures ICL reduction will always vote a valid ICL.
CRs-Fixed: 2048330
Change-Id: I8e53a945530c1685aaaba0537d16ac5fb11858df
Signed-off-by: Ashay Jaiswal <ashayj@codeaurora.org>
Current AICL re-run design for PM660 uses AICL restart mechanism
which starts AICL algorithm from Min ICL value to the configure ICL
value one step (25mA) at a time, this adds considerable amount of
delay for algorithm to reach a settled ICL value. Fix this by using
PMI8998 based AICL re-run design.
Change-Id: I84f612c275fa0d9785a6688a56c9da90df1b86d4
Signed-off-by: Ashay Jaiswal <ashayj@codeaurora.org>
Define ops_keep_radio_on_during_sleep as static as it is used
only in msm_11ad context.
Change-Id: Id95579961ce5cdd3fbdf063578bd4d9144f8a725
Signed-off-by: Maya Erez <merez@codeaurora.org>
Low latency use cases are failing because glink RX thread to handle
the TX Done command is not being scheduled during high system load.
These new APIs allow clients to specify if they need the RX glink
thread to be Realtime
CRs-Fixed: 2050701
Change-Id: I6bd4023394e9ee617797826687f34abaee3fe65d
Signed-off-by: Chris Lew <clew@codeaurora.org>
Configure quat mi2s in slave mode. Create quat mi2s node and
add the pincontrols to be used for data lines.
Change-Id: Ib0a06389941f09014c0f872d3643251859eb6a6b
Signed-off-by: Varun Balaraj <varunb@codeaurora.org>
Return mementry from kgsl_sharedmem_find only if pending_free
is not set for that mementry. This is necessary to avoid use
of a mementry after it is already marked for free.
Change-Id: I23111e9c82a88ccbda2ab259074c38d91f9ff5cb
Signed-off-by: Deepak Kumar <dkumar@codeaurora.org>
