-
qcom-arm64-defconfig-for-6.10fefda685 · ·
Qualcomm Arm64 Defconfig updates for v6.10 Ext4 security labels are enabled, as the expectation of setcap being functional is widespread. SC7280 display and GPU clock controllers are enabled, to make available related functionality. The driver for the Novatek display panel found in QCM6490 IDP is enabled. The X1E80100 sound card and reset-gpio drivers are enabled to provide the necessary drivers for this. The regulator driver providing VBUS power on a few different platforms is enabled. ath12k is present on several recent platforms, so this is enabled as well.
-
qcom-drivers-for-6.10166db010 · ·
Qualcomm driver updates for v6.10 The Qualcomm SCM driver initialization order is improved, to avoid any potential for a client to find a half-initialized SCM instance. The handling of download mode bits is updated to not attempt QCOM_SCM_BOOT_SET_DLOAD_MODE if a io-address for the update is specified, and that path is changed to perform a read-modify-write to avoid updating unrelated bits. Error handling is corrected in the peripheral authentication service (PAS) functions, to release interconnect bandwidth votes in the case of an error. An unwanted error print on allocation error is also removed from this code path. The QSEECOM allow list is marked __maybe_unused to avoid build warnings when built with !OF. The error handling related to the interconnect API is cleaned up to avoid handling the impossible IS_ERR() condition. initcall level is bumped to "core" for cmd-db and rpmh-rsc, as dependent drivers like regulators, interconnects and clocks are registered at this level. Another attempt is made to remove the strncpy() usage in cmd-db, this time with strtomem_pad() which has the correct characteristics. The bwmon regmap cache is changed to maple tree. After an attempt to add missing MODULE_DEVICE_TABLEs to debug drivers, the intention of not having them automatically load is documented. Operations on the pmic_glink client list is put under mutual exclusion, to avoid races when clients are being registered. pmic_glink client registered after the firmware notification arrived was not informed that the firmware was up, this is resolved. More DSPs and the apss subsystem is added to the Qualcomm sleep stats driver. Checks for in-flight regulator requests in the RPMh RSC driver is improved to deal with the fact that these occupy multiple registers, so it's insufficient to just to direct address comparison. The socinfo drivers learns about X1 Elite and SMB2360 PMIC. The maintainers entry is split between the linux-arm-msm list and subset that is maintained in the qcom-soc tree, to avoid some confusion about maintainership.
-
qcom-arm64-fixes-for-6.9-2819fe8c9 · ·
One more Qualcomm Arm64 DeviceTree fix for v6.9 On ths SA8155P automotive platform, the wrong gpio controller is defined for the SD-card detect pin, which depending on probe ordering of things cause ethernet to be broken. The card detect pin reference is corrected to solve this problem.
-
qcom-drivers-fixes-for-6.9ed09f81e · ·
Qualcomm driver fix for v6.9 This reworks the memory layout of the argument buffers passed to trusted applications in QSEECOM, to avoid failures and system crashes.
-
qcom-clk-fixes-for-6.90d4ce245 · ·
Qualcomm clock fixes for v6.9 The introduction of support for the external VDD_GFX supply in SA8295P ADP unearthed a lockdep problem in the GDSC code w.r.t regulator supplies. Make the regulator optional, to avoid creating a dummy regulator, on those boards that doesn't use this. While not solving the root cause of the problem, it reduces the impact of the lockdep warning - and it avoids wasting resources. Refactoring of the RPM clock driver accidentally removed num_clks from msm8976. Reintroduce this to get the clocks back.
-
qcom-arm64-fixes-for-6.9ecda8309 · ·
Qualcomm Arm64 DeviceTree fixes for v6.9 This corrects the watchdog IRQ flags for a number of remoteproc instances, which otherwise prevents the driver from probe in the face of a probe deferral. Improvements in other areas, such as USB, have made it possible for CX rail voltage on SC8280XP to be lowered, no longer meeting requirements of active PCIe controllers. Necessary votes are added to these controllers. The MSI definitions for PCIe controllers in SM8450, SM8550, and SM8650 was incorrect, due to a bug in the driver. As this has now been fixed the definition needs to be corrected. Lastly, the SuperSpeed PHY irq of the second USB controller in SC8180x, and the compatible string for X1 Elite domain idle states are corrected.