There are clk sync issue on TRCM mode for RK3308/RK1808/PX30,
and the previous soluation is to make TX/RX reset at the same
time to fix that. Unfortunately, it fails on PX30-I2S0, because
the reset-TX and reset-RX bits can't be addressed at the same time.
Now, we use another way to fix it:
reset_control_assert(i2s_tdm->tx_reset);
reset_control_assert(i2s_tdm->rx_reset);
|
gate the i2s src clk
|
reset_control_deassert(i2s_tdm->tx_reset);
reset_control_deassert(i2s_tdm->rx_reset);
|
ungate the i2s src clk
as shown above, use the pair gate/ungate src clk to make the
TX/RX reset at the same time.
Change-Id: Ib320ad13350516100eb11c7717e1bd9005abaf39
Signed-off-by: Sugar Zhang <sugar.zhang@rock-chips.com>
TRCM on Master mode require TX/RX enabled at the same time,
or need the one which provide clk enabled at first. In this
situation, one direction's FIFO maybe XRUN which may break
the channels order.
This patch add dmaengine for this mode to handle this situation.
Signed-off-by: Sugar Zhang <sugar.zhang@rock-chips.com>
Change-Id: I831aead0371228661620b54188048e4f3daa717e
TRCM require TX/RX enabled at the same time, or need the one
which provide clk enabled at first. In this situation, one
direction's FIFO maybe XRUN which may break the channels order.
This patch add dmaengine for this mode to handle this situation.
Signed-off-by: Sugar Zhang <sugar.zhang@rock-chips.com>
Change-Id: I30b4ec134b06516d80de4c1d0ccaf916513294ab
Check hwlock user id and object to unlocking someone else's lock
when invoking unlock() API.
Signed-off-by: Frank Wang <frank.wang@rock-chips.com>
Change-Id: I46c256fd703fa56ce3d175ff2d721f53d3615ec0
With this commit, we can set hwlock user id in DT with
"rockchip,hwlock-user-id" property, if it is not set, the driver
use default value 0x01 instead.
Change-Id: Ib7ecc3eaf23ba79d85dc82182a9db0760c6d830c
Signed-off-by: Frank Wang <frank.wang@rock-chips.com>
Set hwlock user id with the "rockchip,hwlock-user-id" property.
Change-Id: I46327ff4146e9c0999def48d67a3538462b82bfb
Signed-off-by: Frank Wang <frank.wang@rock-chips.com>
EP support have been moved to pcie-dw-ep-rockchip.c since 5.10 kernel,
so remove the old EP code from RC driver now.
Signed-off-by: Shawn Lin <shawn.lin@rock-chips.com>
Change-Id: Ia1a5907493a38cf6be5f1a97c5690dd9ecaad3c4
Fixes: c08e952374 ("mtd: spinand: XINCUN: Support new device XCSP1AAPK")
Change-Id: Ifa4d4a875c45b3f20be9839f5bea0f44ff3aeb98
Signed-off-by: Jon Lin <jon.lin@rock-chips.com>
If RTC has been started, stop rtc in pre init will affect
the rtc function and accuracy.
Change-Id: Ic601e8e2d622de95638ca1710c1cff8c5c1a90d5
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
Merge made by the 'ours' strategy.
* commit 'ca7627f2a44b6815b2f23f4da2d0fc996bd20424':
tee: optee: interrupt an RPC depend on shutdown flag
Change-Id: Ia6a2fe4287c0f1b05518c58cb90d5e66be0205cf
drm_atomic_helper_connector_commit() is rockchip private function, so
rename to rockchip_drm_atomic_helper_connector_commit() is more reasonable.
the function is introduced by the following commit:
commit: 88ea93aeb0 ("drm/rockchip: Introduce connector commit")
Signed-off-by: Sandy Huang <hjc@rock-chips.com>
Change-Id: I7d118495024c699529e872e036930a9d3ba70610
when rtk(also some others) wifi/bt combo chips suspend, it
disconnect from bt-remote control unit(RCU). when RCU's
powerkey pressed, it will send ble broadcast to rtk bt chip.
As the rtk chips receive the broadcast, it should wake up
the host, but it can't send bt-hid powerkey to host because
it had lost communication with host.
So,it just send a IRQ pulse to host. The host should resume
from suspend.
Because the RCU just send a IRQ pulse, so rk chips just resume
by IRQ and then go to sleep again for no power-key event.
this is not expected. we expect to power up system wholly.
With this path, we try to power up the system. so we add a
power-up key event when BT IRQ received.
We also can control the "/proc/bluetooth/sleep/powerupkey"
node to enable or disable this function.
Change-Id: Ie59b4a2c4cd2f91820d31835df86565003126465
Signed-off-by: Zou Dengming <marsow.zou@rock-chips.com>
If the hot reset was happening just before the hot_rst_work
was created, NULL pointer dereference was hit.
[ 2.366623] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000100
[ 2.367402] Mem abort info:
...8<...
[ 2.380736] Call trace:
[ 2.380960] __queue_work+0x30/0x5a0
[ 2.381282] queue_work_on+0x88/0xa4
[ 2.381607] rk_pcie_sys_irq_handler+0x1ec/0x280
[ 2.382019] __handle_irq_event_percpu+0x80/0x250
[ 2.382440] handle_irq_event+0x64/0x120
[ 2.382796] handle_fasteoi_irq+0xc8/0x204
[ 2.383162] __handle_domain_irq+0x78/0xe0
[ 2.383528] gic_handle_irq+0xf8/0x364
Fix it by masking all misc interrupts before registering rk_pcie_sys_irq_handler.
And later, they will be unmasked already.
Fixes: 679557456b ("PCIe: dw: rockchip: Delaying the link training after hot reset")
Signed-off-by: Shawn Lin <shawn.lin@rock-chips.com>
Change-Id: I1667e956526faa046a49323ee57b59fdb7a6451d
Part of the SOC supports FSPI CS1N and supports two external devices
for a single FSPI controller. The software provides targeted support.
Change-Id: I1bedf5864b664e94d280a2bc34e7d07212c774d3
Signed-off-by: Jon Lin <jon.lin@rock-chips.com>
Merge made by the 'ours' strategy.
* commit '5f0de7c5d524f0934c83848c6d20526364fc5bb9':
drm/rockchip: vop2: Add devfreq support
Change-Id: I184a624c00cf2bade2a64a929c5d32dd15d75302
Merge made by the 'ours' strategy.
* commit 'a94d0b968ad6275250a6b6bd66f75dcb688ef253':
MALI: rockchip: upgrade bifrost DDK to g21p0-01eac0, from g18p0-01eac0
Change-Id: I01e96d04dd801e7f9d05ac35f934886ca949c1dc
* commit 'e2f9938790a6d9eed3c285c90be45575b6e4c70f':
phy: rockchip-samsung-hdptx-hdmi: Calculate the pll frequency based on the register value
drm/rockchip: analogix_dp: add dual channel mode support
media: rockchip: vicap: support use reserved mem for aov when run with thunderboot mode
mfd: rkx110_x120: reset dsi tx host when generator pattern
media: i2c: cam-sleep-wakeup: optimize sensor wakeup workflow for reset
mfd: rkx110_x120: disable phy when stop video stream
media: i2c: os04d10: add reg list of 1440p@30fps and 360p@120fps
media: i2c: os04d10: fix adapter fastboot issue
Change-Id: I03320bfe99fb028b977197dc04d490288dcc4c34
1. local deserializer and remote devices are associated through i2c-mux
2. remote camera is abstracted as v4l2 subdev
Signed-off-by: Cai Wenzhong <cwz@rock-chips.com>
Change-Id: I8b383651721cc35f7a2f6985ffbb3f0aa6a065a1
