Wakeup_enable bit sets the corresponding bit in the INTR_CFG register.
Also correct target bit to reflect its true position within the
INTR_CFG register. Add this support for Niobe SoC.
Change-Id: Ia0137ade91cdddf4e322f88ca170f4c5e98181a7
Signed-off-by: Meena Pasumarthi <quic_pasumart@quicinc.com>
Currently driver using q2spi_flush_pending_crs() to delete
the pending q2spi packets during port_close. along with this
we should unmap the previous pending RX response buffers
which are not read by User space client application.
Remove unwanted comment from uapi common header file.
Also fix the issue of using geni_resource_lock without initialize.
Change-Id: Ib1ac0fcd338e33f5f331d077034633a44631edf0
Signed-off-by: Chandana Kishori Chiluveru <quic_cchiluve@quicinc.com>
Signal the task waiting on the completion of q2spi->tx_cb in GSI error case
This will help to exit the task without waiting for timeout expiry and
help for UWB framework 1sec timeout cases.
Change-Id: I991b7b01b35d64edb1c743121e76afe5c5d1461d
Signed-off-by: Chandana Kishori Chiluveru <quic_cchiluve@quicinc.com>
Don't disable HID in gsi transfer error path which is causing
disable hid failure and 250msec extra delay and so we can keep
HID disable in q2spi_wakeup_hw_through_gpio function and remove from
check_gsi_transfer_completion function.
In gsi transfer timeout, we can do stop the channels and then start the
channels using gpi_terminate_all function.
Change-Id: I30c595313f9414b070fa17e1c80f516a5fef72bb
Signed-off-by: Jyothi Kumar Seerapu <quic_jseerapu@quicinc.com>
Move dump related modules to first-stage to capture
crash related information as soon as target boots.
Change-Id: I09d700fb12f6097c8cf1f781af2eb6d185cee149
Signed-off-by: Srinivasarao Pathipati <quic_c_spathi@quicinc.com>
The value of register IPCC_REG_RECV_SIGNAL_ENABLE is not persistent,
so need to restore the unmask irq when triggered device wakeup.
Change-Id: Iae1cda73958095f3492972caf7790704a27ee3af
Signed-off-by: Shreyas K K <quic_shrekk@quicinc.com>
Signed-off-by: LADI RAM SAI <quic_lramsai@quicinc.com>
This change to call complete_all with bulk_wait and wait_for_db wait to
signal all the current and future wait_for_completions
to process successfully.
Change-Id: I941fd0425ad8ea42bffd61de9c80addef0600082
Signed-off-by: Chandana Kishori Chiluveru <quic_cchiluve@quicinc.com>
Currently spi geni driver enabling deep sleep flags without
checking deep sleep config. Due to this for levm register
access issues seen. To solve this added deep sleep config
checks.
Change-Id: I1073da76cc6ac066d08063e544b01dd38949d72e
Signed-off-by: Anil Veshala Veshala <quic_aveshala@quicinc.com>
Add virtual channel info print in the error handling of memory
sharing functions.
Change-Id: If0b2b004a8d500db0c8884a0417c758f10245053
Signed-off-by: lixiang <quic_lixian@quicinc.com>
Some hot-pluggable endpoints like SDXC can be removed on the fly. When the
card is removed without gracefully shutting it down. Then PCIe driver can
see a link-down event (as the physical connection has broken) even before
PCIe driver terminates the link.
For such intentional/expected link-down events, we don't need to dump all
the PCIe register info.
This change is to distinguish between normal link-down and link-down due
to card removal and to skip duping debug info in the latter case. It
makes use of the card-presence pin to identify card is present or not.
When it sees a link-down without the card present, then it treats this
link-down as an intentional link-down and skips dumping PCIe registers.
Change-Id: Ifbcf8bae2558b4440ace9c042f6c76a43dce4354
Signed-off-by: Paras Sharma <quic_parass@quicinc.com>
unsigned long uses %lu format specifer, this change updates to same.
Change-Id: I20a3a3c3b705dbd24761bbe9c655fa5d88914a56
Signed-off-by: Tushar Nimkar <quic_tnimkar@quicinc.com>
Signed-off-by: Maulik Shah <quic_mkshah@quicinc.com>
Change is to update the de-emphasis value of NTN3
via i2c register writes to resolve link failure
issues seen on MBB platforms.
Change-Id: If90e5efc19691a14c3d5f12af78df9ff5c2b2a25
Signed-off-by: Subramanian Ananthanarayanan <skananth@codeaurora.org>
Signed-off-by: Krishna chaitanya chundru <quic_krichai@quicinc.com>
Signed-off-by: Yogesh Jadav <quic_yjadav@quicinc.com>
Signed-off-by: Joco Zhang <quic_pingshen@quicinc.com>
1. Enable panicking upon notification timeout and also
increase timeout values to 20 s from the default 10 s.
2. Add PDR timeout configs for notification in niobe
and also increase timeout to 9 sec.
Change-Id: I27a842e776535fb19b61edb4c6e18a3c2cf3dccd
Signed-off-by: Meena Pasumarthi <quic_pasumart@quicinc.com>
Add support for QXR hw platforms in socinfo driver.
Change-Id: Iddbf49c76fa231bdd2a88a63cac2196093f7df36
Signed-off-by: Lijuan Gao <quic_lijuang@quicinc.com>
Currently __qti_flash_led_brightness_set() would be called when
a flash or torch led brightness is set followed immediately by an
hrtimer callback invoked on behalf of an on_time set after
setting the brightness for a switch. When the latter happens
in an atomic/interrupt context and it preempts the former one
running in the process context, an attempt to acquire spin lock
again might occur.
Call Trace:
Core 3 LR: UNKNOWN+0 <fffffff2bc633fe4>
Offending core vcpu context:
Core vcpu3 PC: queued_spin_lock_slowpath+94 <ffffffde691bc96c>
Core vcpu3 LR: _raw_spin_lock[jt]+44 <ffffffde691bbccc>
-------------------------------------------------------------
queued_spin_lock_slowpath+0x94
_raw_spin_lock[jt]+0x44
__qti_flash_led_brightness_set[leds_qti_flash]+0xc8
qti_flash_config_group_symmetry[leds_qti_flash]+0x304
qti_flash_led_symmetry_config[leds_qti_flash]+0x120
qti_flash_switch_enable[leds_qti_flash]+0x40
on_timer_function[leds_qti_flash]+0x1c
__hrtimer_run_queues+0x14c
hrtimer_interrupt+0xdc
arch_timer_handler_virt+0x50
handle_percpu_devid_irq[jt]+0x1e4
generic_handle_domain_irq+0x44
gic_handle_irq+0x4c
call_on_irq_stack+0x3c
el1_interrupt[jt]+0x98
el1h_64_irq_handler+0x18
ret_to_kernel[jt]+0x0
_raw_spin_unlock_irqrestore+0x18
regmap_unlock_spinlock+0x14
regmap_update_bits_base+0x15c
__qti_flash_led_brightness_set[leds_qti_flash]+0x10c
qti_flash_led_brightness_set[leds_qti_flash]+0x98
led_set_brightness[jt]+0xb0
led_trigger_event+0x3c
cam_res_mgr_led_trigger_event[camera][jt]+0x74
cam_flash_ops[camera]+0x114
cam_flash_pmic_apply_setting[camera]+0x588
cam_flash_apply_request[camera]+0x68
__cam_req_mgr_process_req[camera]+0xa24
cam_req_mgr_process_trigger[camera]+0x4ec
cam_req_mgr_process_workq[camera][jt]+0x1b4
cam_req_mgr_process_workq_link_worker[camera]+0x10
process_one_work+0x1e4
Fix this by using spin_{un}lock_irq* variants so that the interrupt
will be disabled and re-enabled when the spin lock is held until
the configuration is done.
Change-Id: Ib88dd9c52c0b10b55e6dd8839732810b6fbbc667
Signed-off-by: Subbaraman Narayanamurthy <quic_subbaram@quicinc.com>
Move dump related modules to first-stage to capture
crash related information as soon as target boots.
Change-Id: I6256bf7cc448fab997e3f2b456a8ea09fbbfb499
Signed-off-by: Wasim Nazir <quic_wasimn@quicinc.com>
"android_rvh_build_perf_domains" vendor hook is used to force building
of perf domain for symmetric cluster system. The vendor hook is called
as part of cpu topology initialization phase, as WALT initialization is
not completed by this time thus setting of override is skipped by WALT
in it's vendor hook and thus topology framework skips building perf
domain for systems with symmetric clusters.
As "android_rvh_build_perf_domains" is independent of WALT state, skip
"walt disabled" check in the vendor hook.
Change-Id: I379c7aec9fd1a19fc6187d8cf85b3f9ef308dcb2
Signed-off-by: Ashay Jaiswal <quic_ashayj@quicinc.com>
Use indirect submission when the ib number is too big to be
submitted inside hfi cmd.
Change-Id: Icba8977e6aa82a2babc9e36cfbdbbfad215fe5e7
Signed-off-by: Jeyaprabu J <quic_jeyaprab@quicinc.com>
For the targets which not describe TBU(Translation Buffer Units)
in device tree, skip accessing of its list_head. Accessing that
list_head results into crash.
[zhenhuah: Adding check in qsmmuv500_log_outstanding_transactions as well]
Change-Id: I298f06b01151285bec3b063d234382371d1738ba
Signed-off-by: Zhenhua Huang <quic_zhenhuah@quicinc.com>
When the device is booted with EUD enabled from bootloader,
dwc3 probe fails. This results in USB functionality and EUD
not working because the PHY is not re-initialized while EUD is
active. This patch resolves the issue by turning on PHY clks
from hsphy_probe and hsphy_init.
Change-Id: I167f984ce37fd795853becb70d89a1fb6eb35bbe
Signed-off-by: Uttkarsh Aggarwal <quic_uaggarwa@quicinc.com>
Signed-off-by: Prashanth K <quic_prashk@quicinc.com>
