summaryrefslogtreecommitdiffstats
path: root/services
Commit message (Collapse)AuthorAgeFilesLines
...
* | Encapsulate the unbindService with try catch blockLuke Zhang2016-07-021-1/+5
| | | | | | | | | | | | | | | | | | | | BT got stuck in mContext.unbindService when it tried to recover from error. The best way to fix it is to encapsulate the unbindService with try catch block CRs_fixed: 996389 Change-Id: I51469c312537b31317481096b0f9822e4a611576
* | Bluetooth: Access mediaplayerlist in synchronized wayAyan Ghosh2016-07-021-67/+73
| | | | | | | | | | | | | | | | Access mediaplayerlist in synchronized way to make sure no transient update is tried to be worked upon in multithreaded environment. Change-Id: I0e7810ac89e0ddc19d57b32c679450aa7cb87273 CRs-Fixed: 979840
* | WifiDisplayController: Fix p2p disconnect before rtsp teardown completesManoj Kumar AVM2016-07-021-16/+81
| | | | | | | | | | | | | | | | Disconnect p2p only after onDisplayDisconnected is received which serves as a notification from native layers of WFD that RTSP message sequence as part of teardown sequence is completed Change-Id: I3f678a76d172725d904013476226810faecfbd83
* | combo bug when location reported by providers do not come with extrasKevin Tang2016-07-021-4/+5
| | | | | | | | | | | | | | | | | | LocationManagerService tries to create an extra bundle and stick it into the location object, but it is not doing it correctly. As a result, combo drops report. Change-Id: I6f5886dd7f14819ecc7a4ed96c867a0ec486faf7 CRs-Fixed: 962746
* | Ignore IMS type APN when notify data connectionRichard LIU2016-07-021-1/+2
| | | | | | | | | | | | Do not add IMS type APN in mConnectedApns. Change-Id: I339a2c411f94572c00e9df48deb38aab1046e581
* | MSIM: Restrict data activity change to listeners with matching subIdMuhammed Siju2016-07-021-1/+2
| | | | | | | | | | | | | | | | | | | | Currently data activity change callback is invoked on all listeners irresepctive of subId. This causes wrong data activity value to be updated from other sub. To fix this add subId match before invoking callbacks on listeners. Change-Id: If218ea3ead2b837d9c2d1cac4d353c1d35c5babe CRs-Fixed: 957838
* | Revert "Revert "Only show keyguard panel if on lockscreen + no activity on top""Danesh M2016-06-291-1/+3
| | | | | | | | | | | | | | | | CYNGNOS-2980 This reverts commit 226797c8e2aba40ecbfb646f2ef5085893dff46a. Change-Id: I0440faf2dd07b427f639277eec2f3cfa89a7de6e
* | WindowAnimator : Check KEYGUARD_PANEL when deciding to show/hide windowDanesh M2016-06-292-1/+13
| | | | | | | | | | | | | | | | | | | | | | By default, if blur is enabled we don't hide the window behind since we want the user to see it. This is not true for when a KEYGUARD_PANEL is active since we can't see anything behind it anyway. Only apply blur logic if KEYGUARD_PANEL is not added. CYNGNOS-2980 Change-Id: Icecae14455d6e68665bb688e3f6dc4ea9d74e88c
* | PhoneWindowManager : Check CM_SETUP_WIZARD_COMPLETED for isUserSetupCompleteDanesh M2016-06-291-2/+2
| | | | | | | | | | | | | | | | | | | | | | During certain flows (such as restoring device data), google's setup wizard sets USER_SETUP_COMPLETED which forces certain things like immersive clings to show when they are not suppose to. Add additional checks to ensure these clings only appear after setup wizard has completed CYNGNOS-3077 Change-Id: I111b3a06e1531bb4265d55bc8a8dcdb02f80cc0b
* | Disable proximity check on power key properlyLuK13372016-06-281-1/+1
| | | | | | | | Change-Id: If1ca0fa02805ce2fbe5a9ac1ab56a5e62beca4e1
* | fw: GlobalActions: Always dismiss any existing dialogTom Marshall2016-06-281-3/+2
| | | | | | | | | | | | | | | | | | | | | | The check for "mUiContext == null" is always false after the initial dialog is created, causing the dismiss logic to never trigger. All code uses getUiContext() now, so the check is not necessary. Remove it and re-align the create/dismiss logic with AOSP. Jira: BACON-4069 Change-Id: Ibba7b38a5fdc8410824db6d0a6f726f4a2822c9d
* | SystemServer: Do not start Atlas Service on lowram devices.Trishool Narayanasetty2016-06-261-1/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | Provide an on/off switch to control atlas service at start-up. This service is disabled by default on lowram devices to save memory. Atlas service is responsible for pre-loading drawables which can be shareable among the applications. Found that the probability of using same drawables in different apps concurrently is very less. However it allocates fixed size of ion buffer based on config. So by disabling it this memory overhead can be avoided. We are safe to disable this service as it does not block any functionality. Change-Id: I14599125c717b61f3ea5cded376fcc1802281756
* | Revert "Only show keyguard panel if on lockscreen + no activity on top"Danesh M2016-06-161-3/+1
| | | | | | | | | | | | | | | | | | Breaks SHOW_WHEN_LOCKED for secure lockscreen + blur + LLS CYNGNOS-3057 This reverts commit 28e023d0c621f00e357f52264baaedf72c1eeee7. Change-Id: Iccec0129e793b910af02bf0a8f60ec9a2ca226a2
* | Themes: Expose Power DialogsBryan Owens2016-06-143-3/+3
| | | | | | | | | | | | | | | | This will give themes the ability to invert the three power dialogs without editing a main resource Change-Id: Ibeab94add8ce8c18955576dc49c441883feb18d6 Signed-off-by: Bryan Owens <djbryan3540@gmail.com>
* | Revert "base: start nfc service prior to systemui"Roman Birg2016-06-101-28/+0
| | | | | | | | | | | | | | | | | | | | We no longer need to start NFC before anything else, System UI will properly handle getting a proper refernce to the Nfc Adapter. This reverts commit 1882a02f21f9e79550000e2b68a37acf21f26ec2. Ticket: OPO-755 Change-Id: I14514746a55fe89e6960f4e453309da5d21012f0
* | UsbDeviceManager: Remove charging from persisted function compositionTom Marshall2016-06-101-0/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | CM 12.1 supported the function composition "charging" to indicate no function compositions, only charging. CM 13.0 does not support this. Instead, mtp is used (see commits to fix b/21429947 for an explanation why this is done). When a user upgrades from 12.1 to 13.0 and the persisted function composition is "charging", modifications to the function composition, such as to add adb or mtp, will timeout and fail. This introduces a three second latency on the change. Removing the charging function when reading the persisted function composition avoids the delay. Jira: BACON-4912, OPO-765 Change-Id: Ibbf4f7ce12f97aa046fa214383580ef4d191ce25
* | Merge tag 'android-6.0.1_r46' into HEADJessica Wagantall2016-06-072-4/+4
|\ \ | |/ | | | | | | | | | | Android 6.0.1 release 46 # gpg: Signature made Mon 06 Jun 2016 10:38:22 AM PDT using DSA key ID 9AB10E78 # gpg: Can't check signature: public key not found
| * Kill the real/isolated uid group, not the ApplicationInfo uidChristopher Tate2016-05-122-4/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | This is a direct reimplementation in L of fixes applied to N in these two commits: 8dc8d37c1d7d694016f1ec2b3cea5fb723567be8 e8741d23d2dd05c4cb3fed5ee6a4040ee96a60e3 Bug 19285814 Change-Id: I59bcc8f1d41c426e9da635bea9ad1d7c6756d5aa Resolve merge conflict when cp'ing ag/941553 to mnc-mr1-release branch
* | powermanager: Extend scope of lock when updating blocked UIDsSteve Kondik2016-06-031-4/+4
| | | | | | | | | | | | | | * Stop WTFs from the system, we need to hold the lock when calling updatePowerStateLocked as one might expect. Change-Id: I344e93cd4ffef915f1684063a496b9e9bbfcb9a6
* | Only show keyguard panel if on lockscreen + no activity on topDanesh M2016-06-031-1/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | There is a bug in window manager, where if the keyboard is collapsed, the reference to the window is lost when destroying its surface. This in turn prevents the ime window from being reordered later on. For normal apps this is not an issue, but since we're above normal apps (KEYGUARD_PANEL), this rises an issue for us in this scenario : 1) Set password lockscreen 2) Swipe up to unlock 3) Collapse keyboard 4) Launch emergency dialer (SHOW_WHEN_LOCKED) 5) Press home button Notice that all touch events are going to the external keyguard. If the keyboard is not collapsed in step 3, everything works and the reason is the reference to the input method window in window manager isn't lost and it gets re-ordered behind the windows. To workaround this issue, avoid removing/re-adding the window and instead rely on window manager to control its visibility. Change-Id: I59dd41547902ae5e23bbeb0a01ee27f11fc29dfc
* | fw: Fix power key interception for PRIVATE_FLAG_PREVENT_POWER_KEYSultanxda2016-06-021-3/+3
| | | | | | | | | | TICKET: CYNGNOS-2958 Change-Id: I67a7469257a1617b1c999f5533c5863014595dcc
* | Frameworks: Add per app controls for LP keyguard notifications (1/2)Lars Greiss2016-06-013-1/+48
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Nice done by google but the UX is a problem especially for ppl who are using a lot apps and just want to see from important apps the notifications on the lockscreen. This commit adds the ability to - enable/disable per app the keyguard notification at all - enable/disable per app ongoing notifications on the keyguard We handle this over the app policy conf file like the other per app notification options. CRACKLING-1127 Change-Id: Ib166db1b1673aeaea132c8eeb16c650d2f254a82
* | fw: reintroduce PRIVATE_FLAG_PREVENT_POWER_KEYScott Mertz2016-05-311-1/+2
| | | | | | | | | | | | | | | | | | Switching the PREVENT_SYSTEM_KEYS forces the application to handle every system key. Some apps may not want to handle every single key, so allow them to just handle the power key. CYNGNOS-2958 Change-Id: I43fd65e67b909e6d52548f110ff9997b829b3e15
* | services: Kick off to CMSystemServer for external service init.Adnan Begovic2016-05-311-9/+21
| | | | | | | | Change-Id: I2dfe51a4cadfe02f0295be36fb30a304219047ff
* | am: clear state for every protected app invocation.Adnan Begovic2016-05-311-0/+3
| | | | | | | | | | Change-Id: I7766e050b8bedb46ff9293a5d61451e6c998fb0b TICKET: CYNGNOS-2962
* | appops: Respect screen interactivity before creating dialogs.Adnan Begovic2016-05-273-4/+60
| | | | | | | | | | | | | | | | | | | | | | | | | | If the device's screen is currently off, do not queue ask runnables who cannot be interacted with. Since these events are gating mechanisms for closing an IPC loop, they need to happen when the user is interacting with the device. Likewise, on screen off, clear the queue of every op as they become unnecessary. Change-Id: Ie930d200839c9408e882510c6bc3ede37ea889ef TICKET: CYNGNOS-2869
* | appops: Implement concept of delayedcount.Adnan Begovic2016-05-272-3/+46
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | High frequency request ops will be delayed until their ignore count ceiling is met. This is to mitigate the overloading the main activity manager service handler and having watchdog kill our service. Google play services likes to share its uid with numerous packages to avoid having to grant permissions from the users perspective and thus is the worst example of overloading this queue -- so, to not encourage bad behavior, we move them to the back of the line. NOTE: these values are magic, and may need tuning. Ideally we'd want a ringbuffer or token bucket here to do proper rate limiting. Change-Id: I5c3e88807abc80f9700dd68dcecd87dac4626de7 TICKET: CYNGNOS-2869
* | Add CM version to dropbox entriesScott Mertz2016-05-251-0/+2
| | | | | | | | Change-Id: Ie318f8b18294a5c7c26ba344a546de81564cc64d
* | pm: Use sourcerecord if available for protected app validation.Adnan Begovic2016-05-252-6/+6
| | | | | | | | | | | | | | | | | | | | | | A source record serves as a beneficial historical record to see if an activity start has been redirected numerous times. This fixes issues where applications that redirect implicitly and are protected would constantly spam to be authed. Change-Id: Ibeb9eae4279a0cdd65635392316eabb485adfa27 TICKET: PAELLA-216 FEIJ-160 FEIJ-177
* | fw: Move HOME key interception logic to interceptKeyBeforeDispatchingjtang2016-05-241-7/+6
| | | | | | | | | | TICKET: FEIJ-522 Change-Id: I290016413c6cad6f5c35f3ef2ee01779991eece8
* | Only call uncrypt when we're encryptedPat Erley2016-05-231-1/+4
| | | | | | | | | | | | | | | | | | | | | | | | We were always generating a block map for encrypted update, even when the device was not encrypted. This leads to a spectacular failure. Fix this by only calling uncrypt when we're encrypted. Additionally, only pass block.map as the update file in the case that the device was encrypted and requires it. NIGHTLIES-3012 Change-Id: Ia34eb5115ac4365605fd57f76179854a6042c5e4
* | fw: Repurpose PREVENT_POWER_KEY for preventing multiple system keys.Adnan Begovic2016-05-231-1/+22
| | | | | | | | | | | | | | | | | | | | | | Allows applications like camera, and various testing applications, to intercept certain system keys to verify behavior or leverage it in creative ways. Limited to signature/privileged. Change-Id: Iacd8735af2aabafbf1fc062d8c565b7102dc6865 TICKET: FEIJ-522
* | IMS: Add support of notifyPreciseCallState for multi simSuresh Koleti2016-05-211-3/+41
| | | | | | | | | | | | | | Extend notifyPreciseCallState for multi SIM. Change-Id: I52adcd55462d6ceae5defacbbe2cb057f4484099 CRs-Fixed: 957251
* | server: Whitelist non-INTERNET/non-CELLULAR Network for meteringBryse Flowers2016-05-212-2/+12
| | | | | | | | | | | | | | | | | | | | Avoid setting a quota or counting statistics for non-internet and non-cellular Networks. Fixes an issue with metering feature where mobile data usage is incorrectly applied to IMS or other non-internet PDN. Change-Id: Ib68f5e560ea1dd230747e4f367081501b70cfc60 CRs-Fixed: 984189
* | Release alarm wakelock for blocked uid.Susheel Yadagiri2016-05-191-37/+24
| | | | | | | | | | | | | | | | | | If an alarm is triggered for a uid and if the same uid is blocked and onSendFinished is not called for that alarm intent for BLOCKED_UID_CHECK_INTERVAL or greater then release the wakelock as that uid is being blocked by data power manager. Change-Id: Ieb4dbcc81e273e692fc5db45ac75a4a4341e8c2f
* | SyncManager: Adjust sync remaining time calculation.yingying2016-05-191-2/+1
| | | | | | | | | | | | | | | | | | The sync remaining millis is not right if the period is changed, so changed to set the sync remaining millis as period reduce the time since last run. CRs-fixed: 794898 Change-Id: Ic3f9fbdd8963814af7715f0f005114b6c79afa55
* | Themes: Fix theming of cmsdk assetsd34d2016-05-191-0/+2
| | | | | | | | | | | | | | Missed the pkgid assignment bits in I4f64d53d82c9edafbe624c3b2eafcda7332587fb Change-Id: Ie442c8922f0d9eac3ed5bab70a99d3e065b719c4 TICKET: CYNGNOS-2723
* | am: Handle unchecked activity starts for protected components.Adnan Begovic2016-05-192-4/+46
| | | | | | | | | | | | | | | | | | | | | | | | Previously if you received a notification from a protected app, since AM would state that the calling package was also the target package, the protected apps implementation would allow you to launch into the application. Mitigate this by hooking into the unchecked activity start stack (pending intent launches) globally. Change-Id: I0371593ade9e4af2554962873d89a0f82a639b57 TICKET: PAELLA-216 FEIJ-160 FEIJ-177
* | usb: ensure accessory detached sent on re-attachKhalid Zubair2016-05-181-1/+7
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | It's possible for USB state change events to get swallowed if the cable is reconnected quickly because state changes are de-bounced with a 1 second interval. As a result an accessory detached event will not be sent if it re-attaches quickly. However, UsbDeviceManager will send an attached event when the USB connection is configured. This causes apps to misbehave and attempt to open the accessory without closing the previous one. The call to openAccessory fails (with EBUSY) because the underlying /dev/usb_accessory chardev is already open exclusively by the first accessory. Fix this issue by sending a detached event before sending an attached event if the previous accessory was not cleaned up properly. This change fixes a failure in CTSVerifier's USB Accessory Test where the test fails if you reconnect the USB cable quickly. OPO-688, FEIJ-471 Change-Id: I49690553213e866d200da1fb22c83348c6eb3491
* | pm: Avoid installing prebundle to other users that were uninstalled by owner.Adnan Begovic2016-05-173-11/+107
| | | | | | | | | | | | | | | | | | | | | | Otherwise the copied app data (since these are treated similarily to system packages on user creation) would be purged of the granted permissions in packages.xml. Also, it just doesn't make sense from a users perspective. Change-Id: I9a72eb9a9f0bea9401e90daa86c8fa32e866c380 TICKET: OPO-695
* | ShutdownThread: Decrypt before stopping storage managerPat Erley2016-05-161-6/+7
| | | | | | | | | | | | | | | | uncrypt needs access to the file to uncrypt it. If we stop the storage manager service, uncrypting fails, leading to failed updates. Change-Id: Ifc75c990ebf8f0a2f11b340cd39cd469b46c6f68
* | Expose lock screen blurring as a setting [2/2]d34d2016-05-161-6/+41
| | | | | | | | | | | | | | | | | | This will provide control over blurring since the current implementation assumes it is always on provided the device config specifies it supports blur. Change-Id: If6d1f1290473df20bda7e96e42571285bd2d8967 TICKET: CYNGNOS-2610
* | Fix incorrect behavior for statusbar and quicksettings panelYinlin Cai2016-05-141-1/+1
| | | | | | | | | | | | | | | | Incorrectly disableForUser() is called in disable2(). Use disable2ForUser() instead of disableForUser(). Bug: 27688623 Change-Id: Iadb6aec82ccfe95a90e4a3d212fbd14d73093982
* | ActivityManager: fix perf boost NPERoman Birg2016-05-131-1/+1
| | | | | | | | | | | | | | | | Able to get an NPE by factory resetting a device, and then just rebooting it. Change-Id: I6f5bbcf59dc20a75a209d6d70e6c3e35fff01406 Signed-off-by: Roman Birg <roman@cyngn.com>
* | CMSettings : Move force_show_navbar to globalDanesh M2016-05-133-10/+10
| | | | | | | | | | | | | | | | | | | | | | Keep feature inline with 12.1, only allow owner to control the feature and mirror across users. Also add additional checks for moved settings. OPO-696 Change-Id: I81c942ab106fead12020107b99c0795a5b9aac98
* | Acquire wakelock until dream service acknowledges dream state changeScott Mertz2016-05-121-5/+55
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Fixes a bug where the system can enter suspend before the dream service can handle the state change. This manifests in the DozeService failing to enable the sensors used to determine if it should pulse the screen by the time the system goes into full suspend. This doesn't completely address the asynchronous nature of the DreamManagerService startDream method. There's still 2 more asynchronous calls before the ACTION_DREAMING_STARTED is sent out, but testing shows this is 'good enough'. OPO-681 Change-Id: Id6ddc33bc367c1a0182df95374817577006a5b98
* | WindowManagerService : Ensure we update both size/positionDanesh M2016-05-121-0/+1
| | | | | | | | | | | | | | | | | | | | | | computeShownBounds is necessary to ensure shownFrame is up to date when setSurfaceBoundariesLocked is invoked to ensure any needed size and positional changes go as part of one transaction. Relying on animateLocked to send changes results in size/position being sent as two separate transactions and surface flinger drawing a frame which looks incorrect according to what the window specified. Change-Id: If881047c060c73db0037fda7da1f66a89dd45487
* | pm: Don't throw if invalid package setting or classname.Adnan Begovic2016-05-111-8/+2
| | | | | | | | | | | | | | | | | | Since an application could be updating in the background while the query for isComponentProtected occurs, the package setting for it can be null. Instead of throwing here, return false. Change-Id: Idb8f11785961ddf2b567e93977a9ca4650ab2119 TICKET: CYNGNOS-2689
* | only use proximity checks on wakeup keysRoman Birg2016-05-111-1/+2
| | | | | | | | | | | | | | | | | | | | The power button should always wake the screen regardless of the proximity; only check proximity when the wake up key code is WAKEUP. Ticket: CYNGNOS-2579 Change-Id: I686ad22d513a3d939be3a0a5c3b752339b2383d3 Signed-off-by: Roman Birg <roman@cyngn.com>
* | am: Don't protect apps that atttempt to grant uri perms.Adnan Begovic2016-05-111-2/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | If an application is delegating uri permissions through the disambiguation dialog, we need to ignore the global protected apps concept to make sure the functionality still works. TODO: Refactor the protected apps implementation to function with startActivityAsCaller within the system process and the android package. TICKET: FEIJ-420 Change-Id: I8ba974a7b25dffe654d32e859064ae86cd9dc7f8
generated by cgit v1.1 at 2025-01-08 17:30:04 +0000