CABC(Content Adaptive Backlight Control) is supported by
mako lcd panel. So this patch enables it to reduce power
consumption.
Change-Id: I433281bc5fe4e638cfeba1beb9cb61e1775f3896
Since the waiting time for pll lock is not enough,
hdmi_pll_enable() have never succeeded. So this patch is increase
waiting time for hdmi_pll_enable().
Change-Id: I9a6b6358b430027c74992f662f8565c32202befc
Some device has problems in packed write for eMMC.
So we can set packed write feature in platform data.
Change-Id: I7e76c78fd076fa5cb0d540c1312fd6ae69aae1b4
msm: iommu: Split rotator/mdp iommu domains
commit ef95ae33a328426b45f502eeb86f2edc120faf30
Finer granularity is needed when mapping into the IOMMU's for
rotator and mdp for secure playback. During secure playback
HLOS will only have access to map into one context bank. The
other context bank is secured and programmed by TrustZone.
Divide DISPLAY_DOMAIN into DISPLAY_WRITE_DOMAIN and
DISPLAY_READ_DOMAIN. Divide ROTATOR_DOMAIN into ROTATOR_SRC_DOMAIN
and ROTATOR_DST_DOMAIN
Change-Id: Iab8f72705b2fd9f3f01b3f42a9fc0193be8ce69a
Use ARRAYED_TOUCH_FW_BIN feature to support both new(PLG137/0.9T)
and old(TM2000/0.55T) panels with one binary.
Firmware Verison: New - E003, Old - E059 (no change)
Change-Id: Ia96eba137f8a02054eb8067d47dc887edffb9970
These changes are reguired to compile sucessfully with gcc 4.6
It's temporary solutions and better to remove the warning from codes.
Change-Id: I7fb7456134bc44bf5ce49edf5f00dae10ae5e778
earsense gpio is set to wake irq.
if it is not added to MPM irq wakeup source,
it blocks to enter the XO shutdown state.
Change-Id: I46a406c1a2a38668ed952ef368bbbb1066975577
update lcd initial code and change power on sequence for
new initial code.
Change-Id: Ic876d5dbef3528591ca374ba0f67f073f4584534
Signed-off-by: Iliyan Malchev <malchev@google.com>
git clone git://codeaurora.org/external/wlan/prima.git
SHA: 04dd8a8d8eb61b4e09387f952c241560a361935f
Incorporate Driver Releases 3.1.7.10 and 3.1.7.11
Release 3.1.7.10
Integrate the following CLs from Perforce:
633566: Do not drop beacons or probe response ever when in scan mode
If beacons or probe response is dropped for an AP, it can take upto 10
seconds to connect to an AP as the AP will not be discovered until
next scan cycle. There was logic to throttle/drop incomming frames
even during scan and not overhwhelm PE and VOSS Message wrappers which
was done in Gen6 days. With the new MSMs, faster processing power and
and removal of SDIO, host should be able to handle and process all
frames fast enough. Therefore do not ever drop beacons or probe
response frames when we are in scan mode. (CR 368500)
633567: Fix BMPS issues which prevent disconnect after missed beacon
1. In missed beacon scenario, BMPS request from PMC must not be
allowed by PE. This is because DUT has exited BMPS to probe the AP and
if probe has not completed, BMPS must not be allowed.
2. When ENTER BMPS is rejected or ENTER BMPS fails, make sure
heartbeat timer is always running. This is just a fail safe check to
ensure we do not ever land in a situation where DUT is out of BMPS and
also out of AP's range but heartbeat timer is deactivated for some
reason. If that happens, link will never be torn down and device will
always keep trying to enter BMPS unsuccessfully.
3. Introduce a new eSIR_SME_BMPS_REQ_REJECT which implies that request
is rejected by PE. eSIR_SME_BMPS_REQ_FAILED now implies that request
was denied by Riva. This will help us in debugging. (CR 367659)
633809: Change value of MAX_SCAN_SSID
While doing cfg80211 interface registeration with kernel, we are
filling wiphy->max_scan_ssids parameter with MAX_SCAN_SSID value
(MAX_SCAN_SSID =2). So During combo scan wpa_supplicant is not able to
send more then 2 profile to host driver. As host driver supporting 9
max ssid. MAX_SCAN_SSID has value set to 9 while registering the
cfg80211 interface. (CR 365219)
634012, 634018, 634021, 634163: Legacy Fast Roaming Feature (CR 362782)
634148: Fix SSR race condition in cfg80211 APIs and BTAMP
cfg80211 APIs are blocked when SSR is going on; there could be various
race conditions in WLAN driver if these APIs are not blocked during
SSR. Some of them are already SSR protected, adding protection to the
missing ones. Also fixing a BTAMP race condition during SSR in softap
mode. (CR 370064)
634185: Fuzzing Fix - Vos Packet is Freed Twice
Fix issue seen in fuzzing test case #48518 for malformed data packet -
There is a bug where if WLANTL_GetEtherType fails, the vos packet is
returned twice. (CR 369987)
634238: Fix for Timer Leak
While unloading host driver it is showing timer leak. In rrmClose
function, it is not destroying neighborRspWaitTimer. Same way it is
not destroying txSleepTimer in hddDevTmUnregisterNotifyCallback
function. this change is going to fix both timer leak issue. (CR
369951)
634272: Fix of kernel Panic in P2P mode after including DXE TX ring empty fi
DXE TX ring is taking more than 3 seconds in some secarios to get
cleared and becasue of this after getting DEL BSS response sessionId
is getting changed and as a side effect of this dev is comign as Null
and any access to it is leading to Kernel Panic. (CR 370441)
634397: SME Close Session wait timeout should be more than WDI timeout
SME Close Session is timing out before getting the call back from
WDI. Increased the WLAN_WAIT_TIME_SESSIONOPENCLOSE to 15 seconds which
is more than WDI timeout (10 seconds). (CR 368821)
634422: Port config80211 softap interfaces to Kernel 3.4
The softap config80211 interfaces are changed in Kernel 3.4, some of
the interfaces are renamed; and also a new data structures is
introduced for softap interface. So the existing WLAN driver APIs are
modified to accomadate these changes. (CR 371516)
634858: Fix typo in sirWrapper.h Copyright
Fix typo that was uncovered during Open Source legal scan
Release 3.1.7.11
Integrate the following CLs from Perforce:
634522: Make changeCountryCode ioctl completly synchronous
Implemented completion routine and callback function to make
changeCountryCode Ioctl completely Synchronous (CR 365094)
634543: Fix typo in WCNSS version log message
634791: Do not allow the second interface for the same type.
Currently two concurrent sessions of same type are not supported. The
new additional session is causing problems to existing session. Seeing
errors like sme_OpenSession() failure, disconnecting and reconnect the
existing session, but data transfer failure etc., As two P2P sessions
are currently not supported, restrict this, at HDD level. This can be
removed once concurrent P2P CLI sessions have to be supported.
634824: Set ERP protection bit when SoftAP overlaps with 11b AP.
SoftAP is not setting ERP protection bit when 11b AP operates in same
channel as that of SoftAP. Added this 11b OBSS overlap support. After
turning off 11b AP, ERP protection bit is cleared but not "OBSS non-HT
STA" persent bit. HT protection also remained "01". We need to disable
this HT protection also while disabling ERP protection. (CR 368189)
634825: Fix for Memory Leak
While sending scan response from LIM to SME, it allocate memory of
8000 byte length for pSirSmeScanRsp. In case where scan results are
empty (bssCount=0) LIM send eWNI_SME_SCAN_RSP message with
statusCode. After posting message it is not freeing memroy which is
allocated previously which result into memory leak for 8000 byte when
we unload driver. Now After posting message we will check if
pSirSmeScanRsp is not NULL then free allocate memory. (CR 371901)
634850: Fix for setband and country commands which occasionally enable dualb
Added a fix to check the INI settings for band (CR 368987)
634876: Max tx power value from ini is not reflected during initial assoc
check is added to pick up the lowest of regmax, ap and ini value (CR 367780
635070: Fix issue with iwconfig for p2p session when concurrency is enabled
635072: Fix for leting device go to IMPS when p2p_client is disconnected
In case where p2p_client is the last session to be
deleted/disconnected, what happens is that csrScanTriggerIdleScan()
fails at vos_concurrent_sessions_running(). We add further check to
see if any session is in connect state. In case there are no sessions
in connect state, it is okie to go into IMPS.
635216: Fix SoftP and P2P porting issues on Kernel 3.4
When registering with config80211 interface, Kernel 3.4 expects to
announce the capability for "remain on channel" interface. So add this
flag when WLAN driver register with cfg80211. For SoftAP when cfg80211
new station interface is invoked, Kernel 3.4 expects an additional
flag in station info, to indicate that it has filled association
request IES.
- Firmware: E052 -> E059 (noise tunning & resolution change)
- Resolution: 1100x1900 -> 1536x2560
Need to reboot the system when abnormal touch work after upgrade
Change-Id: I476180c2a1fa5528fc90d2933072ac01a7e1fd31
disable unnecessary feature to reduce the kernel image size.
- support only GZIP for initrd
- disable unused modules
- disable unused ethernet drivers
- disable remote control from media
- disable printk dynamic debug ( < +2KB)
Change-Id: I5443b7c90191a60aa19cb6c7ba1948f0858a8253
msm: board: enable HS200 bus speed mode support
commit cb6f9ce6659e52b5adcdbac834e632286671f32f
eMMC4.5 specification supports HS200 bus speed mode which allows the
card interface clock to run at more than 52MHz and maximum up to 200MHz.
On most of our existing chipsets we interface eMMC cards on SDC1 slot
as it give 8-bit data interface but timing closure for this slot is not
done for >100MHz clock on any of the existing chipsets (MSM8960, MSM8930
& APQ8064).
Since the timing closure is done till 100 MHz, we can enable HS200 mode
that runs at max. 100MHz. Recent performance measurements also indicate
that HS200 @96MHz (Clock plan supports max. 96MHz) gives better
performance than DDR@48MHz. So enable HS200 @96 MHz. On future targets
which get timing closed at ~200MHz, HS200 at that frequency will be
enabled.
Change-Id: I581c64ad4a4dcff43f8cbbc14255a6f156f2c15e
cherry-picking from kernel v3.5
There is just no point mapping up to 512MB for a serial port.
Using a single 1MB entry is way sufficient for all users.
This will create less interference for the following debugging patch.
Change-Id: Ifc06ec5dbf1c1f25053adb08b9d0a3dccfe46f3e
Signed-off-by: Nicolas Pitre <nico@linaro.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
There is no point using vmalloc for small temporary allocations. kmalloc will
allocate more quickly and will not fragment the vmalloc space.
Signed-off-by: Iliyan Malchev <malchev@google.com>
There is no point using vmalloc for small temporary allocations. kmalloc will
allocate more quickly and will not fragment the vmalloc space.
Signed-off-by: Iliyan Malchev <malchev@google.com>
if not defined MSM_CAMERA_SENSOR or MSM_ACTUATOR, couldn't build kernel.
add inline functions to fix it.
Change-Id: I24b3c8bb87110e04b4bd06e85e72b5143282b2d3
