we found mclk maybe not precise as required because of PLL,
but it still can be used and no side effect. for example, if we
require mclk 11289600, but get 11289598, it doesn't matter.
so using DIV_ROUND_CLOSEST to fix it.
Change-Id: If8453a7a08b319da81b07d572b02247bd7e7bd27
Signed-off-by: Sugar Zhang <sugar.zhang@rock-chips.com>
This commit would enable the VDPU and RKVDEC devices.
The VDPU works in the non combo mode.
Change-Id: I643350d5a2ac17759984fda2e95fb2b82701e7cf
Signed-off-by: Randy Li <randy.li@rock-chips.com>
Even the same type video IP would request a different numbers
of reset control.
From the RK3328 times, the video IP also request decrease the
frequency of the clock to lower than 300 MHZ before resetting.
It seems no hard to apply it into the previous platform.
Change-Id: Iacf1accf24c8776bb8b425b613e6e34215380203
Signed-off-by: Randy Li <randy.li@rock-chips.com>
Jung and I meet some problem the video decoder, so
we just release the VDPU standalone this time.
It seems that the iommu can't attach to two different
IP at the same time.
Change-Id: I24d73cd5ab2c3d32da6ef29661061c7fda9186f2
Signed-off-by: Randy Li <randy.li@rock-chips.com>
Add "rockchip,rk3228-dw-mshc", "rockchip,rk3288-dw-mshc" for
dwmmc on rk322x platform.
Change-Id: I8217d237260a33ce5b115080cf4d41ad4a5733e8
Signed-off-by: Shawn Lin <shawn.lin@rock-chips.com>
There are some race conditions related to phy power on/off
and otg charger detection work, otg sm work. I can find at
least three race conditions at present.
Race condition[1]:
The first race condition involving phy power on/off which
may be caused by the following case.
Test on rk3399 evaluation board Type-C0, connect to PC usb
port with Type-C cable, then phy power on/off operation may
be done twice because of race condition between phy driver
and usb controller driver.
CPU 0:
- rockchip_usb2phy_bvalid_irq()
- rockchip_usb2phy_otg_sm_work()
- detect connect to PC usb, do phy power on
- rockchip_usb2phy_power_on()
CPU 1:
- dwc3 driver do runtime resume process
- dwc3_runtime_resume()
- dwc3_core_init()
- phy_power_on()
- rockchip_usb2phy_power_on()
Although we use a suspended flag in rockchip_usb2phy_power_on()
to avoid doing the same things twice, but it's not enough to
prevent race condition if phy driver and usb controller driver
access the rockchip_usb2phy_power_on() at the same time. This
race condition may cause clk management unbalanced.
Race condition[2]:
The second race condition related to phy power on/off and otg
charger detection work. We need to keep the usb phy staying in
suspend mode when do usb charger detection. But now it don't
have any protection to prevent the other threads to operate phy
during charger detection.
The problem can also be easily reproduced on rk3399 evaluation
board Type-C0 when connect to PC usb port with Type-C cable.
CPU 0:
- rockchip_chg_detect_work()
- power off phy and start to do charge detection work
CPU 1:
- dwc3 driver do runtime resume process
- dwc3_runtime_resume()
- dwc3_core_init()
- phy_power_on()
- power on phy again
This race condition may cause charger detection and later usb
enumeration abnormally.
Race condition[3]:
The third race condition involving otg sm work. The otg sm
work can be interrupted by bvalid irq, and the bvalid irq
handler rockchip_usb2phy_bvalid_irq() will do otg sm work,
which may cause unknown error.
This patch uses mutex lock to protect the phy operations,
otg charger detection work and otg sm work.
Change-Id: Ic6845a10b3e69fe9ae6cf0b2d4e2beb098232abd
Signed-off-by: William Wu <william.wu@rock-chips.com>
This reverts commit 7e705c4974eaa8abaf44cb1542d3ec49d520fde8.
Change-Id: I498ade43de012f65ea39624bd2982b4a84bcbf54
Signed-off-by: Mark Yao <mark.yao@rock-chips.com>
The logo framework use state->rotation may conflict to common drm
update, cause display abnormal
Change-Id: I09b6b898a7606cd05371af1f4b25254945923d0d
Signed-off-by: Mark Yao <mark.yao@rock-chips.com>
The jtag clk no driver to handle them.
But this clk need enable,so make it as critical.
The ddrphy/ddrupctl clks no driver to handle them,
Chip design requirements for these clock to always on,
The pmu_hclk_otg0 is Chip design defect, must be always on,
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
(cherry picked from git.kernel.org mmind/linux-rockchip.git v4.13-clk/next
commit 223c24be74)
Conflicts:
drivers/clk/rockchip/clk-rk3368.c
Change-Id: I31c1c7efb7a83652501a7f53ff5931d9f308f736
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
The atclk/dbg/jtag/hsic-xin12m/pclk_core clks no driver to handle them.
But this clks need enable,so make it as ignore_unused for now.
The ddrupctl0/ddrupctl1/publ0/publ1 clks no driver to handle them,
Chip design requirements for these clock to always on,
The pmu_hclk_otg0 is Chip design defect, must be always on,
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
(cherry picked from git.kernel.org mmind/linux-rockchip.git v4.13-clk/next
commit 55bb6a633c)
Conflicts:
drivers/clk/rockchip/clk-rk3288.c
Change-Id: I6271a903deb9ca21b5e74fd2c1ad4cf69f7021e1
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
No driver to handle this clk yet, but chip design requiress for this clock
supplying the ddr controller to be always on.
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
(cherry picked from git.kernel.org mmind/linux-rockchip.git v4.13-clk/next
commit f2893aaba4)
Change-Id: I3cd9578f73a69eb0f09d1f40c22ee55b393149aa
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
The shared uart-pll is on boot a child of the apll that can get changed
by cpu frequency scaling. So move it away to the more stable gpll to
make sure the uart doesn't break on cpu frequency changes.
This turned up during the 4.11 merge-window when commit
6a171b2993 ("serial: 8250_dw: Allow hardware flow control to be used")
added general termios enablement making the uart on rk3036 change
frequency and thus making it susceptible for the frequency scaling issue.
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
(cherry picked from git.kernel.org mmind/linux-rockchip.git v4.13-clk/next
commit f8ba2d68e5)
Change-Id: Ia8683d7b49523284043457727665d7e58d1551ec
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
The mux_pll_src_apll_dpll_gpll_usb480m_p parent list was missing a ","
between the 3rd and 4th parent names, making them fall together and thus
lookups fail. Fix that.
Fixes: 5190c08b29 ("clk: rockchip: add clock controller for rk3036")
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
(cherry picked from git.kernel.org mmind/linux-rockchip.git v4.13-clk/next
commit 9b1b23f03a)
Change-Id: I535b64fc7c902a4e9c64b4b803bb03126b7ba110
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
The jtag/bus/peri/initmem/rom/stimer/phy clks no driver to handle them.
But this clks need enable,so make it as critical.
The ddrupctl/ddrmon/ddrphy clks no driver to handle them,
Chip design requirements for these clock to always on,
The hclk_otg_pmu is Chip design defect, must be always on,
The new document will update the description of this clock.
All these non-noc/non-arbi clocks,IC suggest always on,
Because it's have some order limitation, between the NOC clock switch
and bus IDLE(or pd on/off).
The software is not very good to solve this constraint.
Always on these clocks, has no effect on the system power consumption.
The new document will update the description of these clock.
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
(cherry picked from git.kernel.org mmind/linux-rockchip.git v4.13-clk/next
commit f18c0994cd)
Conflicts:
drivers/clk/rockchip/clk-rk3228.c
Change-Id: Ie2c4c8d2c73a62efe96e64a3ec638970e82051d1
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
This patch support using appended device tree blob to zImage and
supplementing the appended DTB with traditional ATAG information.
Change-Id: I8e8e63513c17544fdafd9107fda425740c63220e
Signed-off-by: Frank Wang <frank.wang@rock-chips.com>
Building with CONFIG_CC_STACKPROTECTOR_STRONG triggers protection code
generation under CONFIG_ARM_ATAG_DTB_COMPAT but this is too early for
being able to use any of the stack_chk code. Explicitly disable it for
only the atags_to_fdt bits.
Change-Id: Ib1f66cc4083b4f04d713c3c70610b8a337a6b0ff
Suggested-by: zhxihu <zhxihu@marvell.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Frank Wang <frank.wang@rock-chips.com>
(cherry picked from commit 7f66cd3f54)
Add the clock tree definition for the new rk3128 SoC.
And it also applies to the RK3126 SoC.
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
(cherry picked from git.kernel.org mmind/linux-rockchip.git v4.13-clk/next
commit f6022e88fa)
Change-Id: Ib933e398bc8e40d8659bc1cdc419116f48f6ae30
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
Add the dt-bindings header for the rk3128,
that gets shared between the clock controller and
the clock references in the dts.
Add softreset ID for rk3128.
And it also applies to the RK3126 SoC.
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
(cherry picked from git.kernel.org mmind/linux-rockchip.git v4.13-clk/next
commit b20841b9e0)
Change-Id: I70c055570319abe4547ac2a42b9139c7248abb13
Signed-off-by: Elaine Zhang <zhangqing@rock-chips.com>
When LPAE is activated, the dma_addr_t type is u64,
but pointer is still 32bit on arm32 platform.
1. %pad is used to print dma_addr_t type in log.
2. The member paddr(dma_addr_t type) in struct shm is cast
to unsigned long when it needs to be cast to a pointer. The cast
is fine as the value of paddr in struct shm is always less than 4G.
Change-Id: I1e2112796f657759dfa845258ea19558cb84c4ec
Signed-off-by: Zhang Zhijie <zhangzj@rock-chips.com>
we think 'if (fiq_pt_regs.pstate & 0x10)' doesn't make any
sense, use sp_el1 from bl31 delivers is ok.
Change-Id: I0792d76e39912b4ca5484b029761daac05cd719b
Signed-off-by: chenjh <chenjh@rock-chips.com>
Add CLK_SET_RATE_PARENT and CLK_SET_RATE_NO_REPARENT
for dclk_lcdc.
Change-Id: I19a4a8e5f9e2cc5fda8b70f1b632dccd538e02a0
Signed-off-by: Zheng Yang <zhengyang@rock-chips.com>
when other devices bind failed,drm will unbind and re-bind all devices.
if don't cancel the delayed work but flush and destroy workqueue directly,
kernel point is likely to become NULL.
Change-Id: Ib48704186ee298cbd4daac1cdbbac5fb3906b6bb
Signed-off-by: algea.cao <algea.cao@rock-chips.com>
When tested usb device through force device mode method,
we found that usb device failed to connect to usb host
in the following case.
1. Use micro usb 2.0 OTG interface.
2. Plug in otg cable, and the id pin was pulled down
to Ground.
3. User space force usb to enter device mode through
'echo 2 > /sys/bus/platform/drivers/usb20_otg/force_usb_mode'
4. Use usb 2.0 Standard-A to Standard-A cable assembly,
plug into otg cable receptor on one side, and connect
to PC on the other side.
5. PC fail to enumerate our device, because of usb driver
logical issue.
This is because that the dwc_otg_pcd_check_vbus_work()
only enable usb to start connecting if check the bvalid
and iddig is high. But in the above test case, the iddig
is low, so fail to start connection work. In this patch,
we enable usb to connect if iddig is high or usb is in
force device mode.
In addition, fix some coding style to increase the readability.
Change-Id: I08f1a4e6e7e5fb246b1716a20d4572d8b866f238
Signed-off-by: William Wu <william.wu@rock-chips.com>
