-
Notifications
You must be signed in to change notification settings - Fork 365
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[LA.UM.7.1.r1] Merge tag 'LA.UM.7.1.r1-16600-sm8150.0' #2327
base: aosp/LA.UM.7.1.r1
Are you sure you want to change the base?
[LA.UM.7.1.r1] Merge tag 'LA.UM.7.1.r1-16600-sm8150.0' #2327
Conversation
…rithm For voltage based step charging, used in TTF algorithm, improve the code readability by changing ttf_mode from TTF_MODE_V_STEP_CHG to TTF_MODE_VBAT_STEP_CHG. Change-Id: Ia3b7dd0f2fc1b354c5afaa745c1117719fb9df66 Signed-off-by: Sahil Chandna <[email protected]>
To get the information on irq status with default logs added the irq status prints to CAM_ERR_RATE_LIMIT. Change-Id: Ibe6a57e2905aeaa9d8c38cf08ad4d700468f341f Signed-off-by: Tejas Prajapati <[email protected]>
There is a race condition where connector state mode info is updated in sde_encoder_virt_mode_set in crtc_commit thread and mode info getting copied to new connector state in connector_atomic_duplicate_state in hwbinder thread. This is causing new connector state to be in invalid state leading to wrong dsc config and further leading to ping pong timeouts. Removed connector state mode info update in crtc_commit thread as connector state mode info is updated in mode_fixup and does not need updation in sde_encoder_virt_mode_set in crtc_commit thread. Change-Id: Idb05948f7ad0a6af9846a7e1e90c7f934a072815 Signed-off-by: Jayaprakash <[email protected]>
If umd issues flush all, isp driver need to make sure that wait and active list requests also flushed and corresponding buffer fences needed to returned. For active and wait list flush, ife hardware need to stop at frame boundary and release the fences. Ife hardware need to reset to make sure that vfe bus write master fifos are cleared. Start the ife hardware after reset and set the isp context state to sof. Change-Id: Ibc57e64fbcef07de13520f2b9e4cc37c7c62bb37 Signed-off-by: Ravikishore Pampana <[email protected]>
…4.14.c2-display * commit '40c77ca69b266c88ebbfdfdf806ed3033f40840e': cpufreq: schedutil: Fix for limits update with fast switch enabled cpufreq: schedutil: Add fast switch to sugov_limits path soc: qcom: wda: Disable powersave work on deinitialization msm: kgsl: Change data type for GPU ib vote disp: msm: sde: add SSPP CP features to dirty list during IPC msm: ipa3: Fix to recycle buffers diag: Prevent using uninitialized mdlog session mask msm: kgsl: Ignore rgmu disable GDSC error power: smb5-lib: Report liquid presence status msm: kgsl: Add missing check for snapshot IB dump msm: kgsl: Snapshot all GMU memory blocks msm: kgsl: Snapshot GMU TCM registers separetely net: qrtr: mhi: Register for early notifications msm: kgsl: Fix drawqueue timer race condition msm: kgsl: Ensure GMU memory regions don't leak memory msm: kgsl: Update GMU snapshot section IDs ARM: dts: msm: Update GPU snapshot size msm: kgsl: Omit referencing pointers in traces arm64/iommu: handle atomic pool addresses in ->get_sgtable and ->mmap arm64/iommu: handle non-remapped addresses in ->mmap and ->get_sgtable ARM: dts: msm: register IPCR for early MHI notification mhi: cntrl: qcom: register for early error fatal notification mhi: core: add support for early error notifications diag: Update diag get log request structure msm: kgsl: Change throttling counter weight from 15% to 5% NFC: Fix device node probing issue mhi: core: add missing macro for MHI device vote msm: kgsl: show max gpu temperature soc: qcom: qmi_interface: Remove ineffective mutex lock from txn struct soc: qcom: qmi_interface: use qmi txn_lock to avoid use after free of txn sched/fair: Prevent tick path active migration to the same CPU msm: camera: isp: Halt device with the command parsed msm: vidc: add check to avoid out-of-buffer write usb: f_gsi: Don't enable IPA data path if connect channel fails mhi: core: expand MHI vote api to take optional parameter sched: fair: Add strict skip buddy support sched/fair: Allow load bigger task load balance when nr_running is 2 msm: vidc: add additional check to avoid out of bound access USB: configfs: Send DISCONNECT uevent during UDC bind usb: dwc3-msm: Program MSB of doorbell register when using dummy_addr msm: ipa3: Drop WAN TX packets when pipe is down in ssr usb: dwc3-msm: Use dummy buffer as doorbell until GSI is ready ARM: msm: dts: modify rotator line width for sm8150 Revert "drm/msm/sde: fix crtc enable check" ARM: dts: msm: add thermal zone names to sm8150 msm: adsprpc: Fix integer overflow in refcount of map msm: kgsl: Ensure _free_pt() gets called on error case diag: dci: Correct out of bounds check in processing dci pkt rsp usb: dwc3: Flush bh work before disabling dwc3_irq power: step-chg-jeita: update jeita/step ranges to support signed threshold slim-msm-ngd: Create a IPC error logging file soc: qcom: dfc: Purge packets on flow delete Revert "diag: Clear memory device entries during mhi disconnect" ARM: dts: msm: Add GPU coresight properties for SM8150 drm/msm/sde: fix crtc enable check usb: misc: nb7vpq904m: fix possible array overflow msm: ipa3: Handle missing cleanup in IMP_READY state on shutdown msm: npu: update performance mode during load network msm: npu: Allow user app to retrieve NPU version via get_property fbdev: msm: check the length of the external input buffer properly msm: kgsl: Ensure that thermal power level restrictions are enforced net: qrtr: Use skb with fragments in memory failure case media: uvcvideo: Fix 'type' check leading to overflow power: qpnp-qg: Add range checks to FIFO length msm: kgsl: Don't attempt the LLM/GLM handshake for a640 targets power: qpnp-smb5: Do not use ibatt based termination for QG PMICs msm: camera: sensor: Increase the eeprom map buffer count msm: vidc: refine clock calculation for both encoder and decoder msm: vidc: consider pipeline overhead factor for clock calculation msm: camera: reqmgr: Reset previous skipped slot if valid request drm/msm/sde: update mixercount check in writeback usecase msm: camera: crm: Increasing the device handles to 128 f2fs: fix to avoid deadlock in f2fs_read_inline_dir() msm: kgsl: Correct snapshot over-write policy msm: camera: Adding device type to track device handles msm: camera: isp: Add support for initial frame drop esoc: Add mdm error-fatal notifier hook ARM: dts: msm: Increasing shared mem region size for SM6150 and sdmmagpie msm: camera: icp: Prevent out of bound access in acquire msm: camera: mem: Mutex before cpu_begin_access & cpu_end_access msm: camera: cpas: Check the HW state before accessing register msm: camera: cci: Add rd_done to handle read done operation msm: camera: crm: Increase the device handles to 128 ARM: dts: msm: Increase camnoc bw margin for sdmmagpie msm: camera: fd: Remove duplicate "qcom,fd501" property msm: camera: reqmgr: Fix CRM shift one req issue msm: camera: icp: Mapping fw error numbers with error names msm: camera: isp: Prevent out of bounds read msm: camera: csiphy: correct DPHY bring up sequence msm: camera: Fix cpas axi clk rate overflow msm: kgsl: Correctly identify page faulting process msm: camera: reqmgr: Skip reset if no request from UMD Change-Id: I7fb1713ebe3d786c7134db39033860114baa3df9 Signed-off-by: Abhijith Desai <[email protected]>
Enabling format measure helps to find mismatch between the expected sensor width and height with actual sensor width and height. In case of mismatch CSID will give CSID_PATH_ERROR_PIX_COUNT and CSID_PATH_ERROR_LINE_COUNT. Change-Id: I11aefe7d073ec47810564442109981d0e46f9844 Signed-off-by: Rishabh Jain <[email protected]>
…" into dev/msm-4.14-camx
Add fps blob to get the fps value from user space. Epoch interrupt configuration will be depend on the fps value. Configure epoch interrupt to half of the frame if fps value is 60, in rest all cases configure epoch interrupt to 2/3 of the frame. Change-Id: I2544a5e18895a983155c2f01e49920c9e1ec59b9 Signed-off-by: Chandan Kumar Jha <[email protected]>
Dynamic FPS to Detect IRQ delay. If the irq is delayed, the frame drop is reported. This commit helps to recover from the frame drop. We a added a new blob command to report fps from UMD to KMD. This patch Use this fps to calculate If there is a IRQ delay or not. If yes then It sets a flag 'irq_delay_detected'. Change-Id: I3cbcd9275d705f8c2458c7860780fbab615fabc4 Signed-off-by: Vikram Sharma <[email protected]>
Add extra information in the logs during acquire failure. This will ease during debug process by providing enough information from the logs. Change-Id: I0a8789e80ff4477a5596ea0e9c3374ddab085d60 Signed-off-by: Ayush Kumar <[email protected]>
Assign right structures based on csiphy version for atoll. Change-Id: I7dbc2278bbd60a149d3fcc44fa2251419d4511f4 Signed-off-by: Shankar Ravi <[email protected]>
Presently hdlc mode is not getting updated for a peripheral after multimode mdlog exit. Update the peripheral's hdlc mode while closing the mdlog session. Change-Id: I255db98b592648df00e779027232a55c4585de6a Signed-off-by: Manoj Prabhu B <[email protected]>
Camera CAMNOC, ICP, IPE, BPS, CDM, SMMU, JPEG, IFE, PPI nodes are added for chipsets having v150_110 camera. Change-Id: Iea631ba2f7a48fa1a144769097acdc865cfbbbdc Signed-off-by: Chandan Kumar Jha <[email protected]>
…into dev/msm-4.14-camx
…rdown db_timer expires after wq has been flushed in ipa3_teardown_sys_pipe. This causes the wq to be scheduled after pipe teardown. Add workqueue_flushed flag which set after the wq is flushed during teardown. This flag is checked against in the wq function to ensure it is not executed once the wq has been flushed. Change-Id: I176a3a6dbf25985a41decf008ad045647081b871 Acked-by: Priyadarshini Rajagopal <[email protected]> Signed-off-by: Bojun Pan <[email protected]>
PPI is a bridge between CSIPHY and CSID. Responsibilty of this driver is to enable and configure PPI bridge from CSID as per the configuration. Change-Id: I3e2a12043088335f62a54670167306d6084c7221 Signed-off-by: Vikram Sharma <[email protected]>
Ensure that the V-NOC module is reset for 1 All types of system error from firmware. 2 Specific video hardware version. Change-Id: I534cdefd11ebb9217081c2c840f332f7ed7c450d Signed-off-by: Vikash Garodia <[email protected]>
…4.14.c2 * quic/dev/msm-4.14.c2-display: drm/msm/sde: avoid connector state mode update in commit thread Change-Id: Ic413e8d9eaadbe30e160bfd4348f60fe4410ba39 Signed-off-by: Abhijith Desai <[email protected]>
Currently driver is setting the user override speed to maximum speed by checking if override speed is less than the maximum speed. But this check will fail if the maximum speed supported by the controller is super-speed and user is trying to set it to super-speed from a lower speed. Fix this by checking if override speed is equal to or less than maximum speed. Change-Id: Id71cf163548e213ec0180d7f3107395feed2f483 Signed-off-by: Pratham Pratap <[email protected]>
PPI clock sources are added for chipsets having v150_110 camera. Change-Id: Ibab7f009fce45ea309f13501efb2f16fafb87ed6 Signed-off-by: Chandan Kumar Jha <[email protected]>
…ra" into dev/msm-4.14-camx
Plane count maybe passed to sort function as argument without being checked against NULL value in crtc atomic check and blend setup mixer. Add check to use plane count as argument only when it has a valid value. Change-Id: Ibddd9a630fd5afe55c861975d5fce6985daf1401 Signed-off-by: Shubhashree Dhar <[email protected]>
During pm suspend-resume cases, topology is reset and ppsplit flag is set to false based on previous connector state. This leads to incorrect interface config resulting in pptimeout. Made changes to update the ppsplit flag based on current connector state to avoid this. Change-Id: Id42313036eda98e5109083b0d57a778f1262f3fd Signed-off-by: Shubhashree Dhar <[email protected]>
Add the necessary checks to skip the invalid operation of driver remove during the recovery. Change-Id: I12d755b3b19eb4b8a7cc09e3893d778d88201cae Signed-off-by: Naman Padhiar <[email protected]>
Some of user space daemon like ADB queries endpoint descriptor using ioctl to know wMaxPacketSize value. Based on wMaxPacketSize value it queues additional buffer to read zero length packet if expected read buffer size is multiple of wMaxPacketSize. Currently driver is missing handling for super speed plus case which results into sending full speed related USB endpoint descriptor (i.e. wMaxPacketSize 64 bytes). Hence when adbd is expecting 64 bytes packet from host, it queues one more additional buffer. This results into mismatch of ADB command and response when multiple ADB commands are used and causing different ADB related stability issues. Fix this issue adding super speed plus check with FUNCTIONFS_ENDPOINT_DESC ioctl command. Change-Id: I9416295c07c2d98f9d32df43d7e506f975da15a2 Signed-off-by: Mayank Rana <[email protected]>
dpdm_regulator_enable() function turn on the CXO clk to program the phy registers to put PHY in non-driving mode for charger detection. In charger connected case phy driver voting for CXO clock in dpdm_regulator_enable but does not disabling it which is preventing the system from vdd minimization. Hence fix this issue by turning off the CXO clk from dpdm_regulator_enable. Change-Id: Ib3a04eedcef625443077199d5920884f114db82d Signed-off-by: Chandana Kishori Chiluveru <[email protected]>
Update SD card removal event processing logic. Instead of pinging the card to know the card presence rely on card-detect gpio state. On multi-card tray designs, the same card-tray would be used for SD card and SIM cards. If SD card is placed at the rightmost location in the tray, then SIM card may come in contact with SD card power- supply while removing the tray. It may result in SIM damage. For protecting SIM from this issue, in multi-card tray designs, a h/w fix done such that pmic gets a notification of SD card removal event (through hardwiring) and it turns off the SD card voltage regulators immediately. All this will be done much before SD card driver starts processing card removal event. To support this design, SD card driver shouldn't turn-on the regulator while processing card removal event. But the present mmc driver turns-on regulator (multiple times if the card was in suspend state). To avoid turning on SD card regulator in card removal path, updating the card removal processing logic is based on card detect gpio state. Change-Id: I13708a60c9378519713ebec8071ae3b130012a93 Signed-off-by: Veerabhadrarao Badiganti <[email protected]> Signed-off-by: Sarthak Garg <[email protected]>
This reverts commit 9de18ca. Reverting this change as this is introducing race condition with sdcard plug out scenario and leading to device crash. 1. Deferred resume kicks in 2. SDCard is plugged out 3. Deferred resume failed 4. SDCard detection scheduled as resume fails 5. Device crashing Change-Id: I6fd81d6c21be4a0e3139246c9d66959010fd240c Signed-off-by: Ram Prakash Gupta <[email protected]> Signed-off-by: Sarthak Garg <[email protected]>
If resume fails, there is no way to handle it now. Also there's no attempt to recover from it. This leads to lot of warnings while issuing requests. Check for resume errors & reset the stack on error as an attempt to recover from it. Change-Id: Ie4d6d2a34c2c7a8154696e93d85e50d60410e0c2 Signed-off-by: Asutosh Das <[email protected]> Signed-off-by: Ram Prakash Gupta <[email protected]> Signed-off-by: Sarthak Garg <[email protected]>
Sometimes when the card is removed and a request is in queue the resume fails. But there's no way to inform the in-flight request that the card is removed since the resume itself fails and rescan is waiting for claim_host which was acquired by the in-flight request. This request percolates to platform driver which sees that the pre-requisites to issue the request are not met i.e. the clocks are OFF. So it tries to dump the registers and results in a NoC error. Set the card removed when resume fails to avoid this problem. CRs-fixed: 2430862 Change-Id: I171ad435ec11c1212e6528592b8db43cd0171b11 Signed-off-by: Asutosh Das <[email protected]> Signed-off-by: Ram Prakash Gupta <[email protected]> Signed-off-by: Sarthak Garg <[email protected]>
* commit '1d886afeed0208860d2f6f01c677306f06a65a9b': msm: camera: LDAR Debug info dump framework implementation Change-Id: If8be42d36cc3bbf2e82b79bd0a1c6f4517f32e1b Signed-off-by: Sridhar Gujje <[email protected]>
Change-Id: Id639749c1f877e72564aa59b1b3106ade2ec2407
Change-Id: I3aaf70567cf782280a876ee0e549a7cc08c5bf9b
…org/quic/la/kernel/msm-4.14 into HEAD "LA.UM.7.1.r1-16600-sm8150.0"
Hi @tomgus1, should this one be applied on top of the other? |
@pablomh this can be applied without the other but added for continuity and easier tracking. |
Testing this one on yoshino/lilac. Until now no new regressions found everything seems to be fine. Already tested:
In case I spot something unusual I will report back, running this one together with Kernel 4.14.190 PR |
Tim, I've been thinking about this one a lil more... Can you remove and PR separately all the camera related commits? After applying the requested changes, we should then merge the rest. |
Very good idea indeed Angelo, 42 of the 87 commits are merge commits. Removing them, makes the PR a lot more easier to deal with and removes the noise from the content. @tomgus1 if you don't have enough time sorting commits I could help you with that. In general I'd love to see that we are making progress adapting also the other missing fixes from CAF that are contained in https://source.codeaurora.org/quic/la/kernel/msm-4.14/tag/?h=LA.UM.7.1.r1-17400-sm8150.0. My hope would be, that there are more fixes that stabilize/improve the kernel for our phones. The 17400 tag is as well already 8 month old and maybe it contains fixes for the 10 month old tag: 16600?! I wonder if it wouldn't be more time efficient when we try to test all non camera related fixes until 17400 in one go and afterwards testing the camera related stuff in the from Angelo suggested separate PR. It would be very unfortunate to stumble upon problems in 16600 that have been already fixed in 17400. Do they have a consolidated changelog like e.g. kernel.org? I haven't found one. |
Sorting out the merges should be relatively easy as The only reason to "keep" this history is to share the tree with CAF. Meaning all the commits retain their identical hash with CAF rather than appearing cherry-picked, and new(er) tags on the same branch will be properly based on top of that, instead of appearing as a totally different diverging branch. In addition git might not always be able to understand what commits have been applied (especially if things are applied and reverted multiple times), but that's relatively easy to solve with Though, I guess this all sums up to Angelo's:
😁 @stefanhh0 all development now happens on 8.1 tags for "Android 10" which is why 7.1 tags for "Android 9" haven't been added anymore. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
x
So basically we already have at least 75% of this PR in... :p |
Yes, I will go through and prepare a separate PR that collects the very few commits that are missing. |
@stefanhh0 that's a lot of manual labour here! Are you checking whether commits are potentially reverted and re-applied again or is this something that a rebase with conflicts (utilizing |
Well, I wanted to understand what exactly is missing and what can be omitted. I have finished the work already and I will send the PR as soon as I have build it and it and see it running on my phone. Only 16 commits were missing. |
Fair point to be made, rebasing and blindly skipping every commit that conflicts yields me 13 commits. So it's good to have all these checked in the end. |
1aa2204
to
838a06e
Compare
No description provided.