The driver, once it found a divider, tries to round it up by increasing
the least significant bit of the fractional part by one when the
round_up argument is set and there's a remainder.
However, since it increases the divider it will actually reduce the
clock rate below what we were asking for, leading to issues with
clk_set_min_rate() that will complain that our rounded clock rate is
below the minimum of the rate.
Since the dividers are fairly precise already, let's remove that part so
that we can have clk_set_min_rate() working.
This is effectively a revert of 9c95b32ca0 ("clk: bcm2835: add a round
up ability to the clock divisor").
Fixes: 9c95b32ca0 ("clk: bcm2835: add a round up ability to the clock divisor")
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Acked-by: Stephen Boyd <sboyd@kernel.org>
Reviewed-by: Nicolas Saenz Julienne <nsaenz@kernel.org>
Tested-by: Nicolas Saenz Julienne <nsaenz@kernel.org> # boot and basic functionality
Tested-by: Michael Stapelberg <michael@stapelberg.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210922125419.4125779-3-maxime@cerno.tech
The driver currently tries to pick the closest rate that is lower than
the rate being requested.
This causes an issue with clk_set_min_rate() since it actively checks
for the rounded rate to be above the minimum that was just set.
Let's change the logic a bit to pick the closest rate to the requested
rate, no matter if it's actually higher or lower.
Fixes: 6d18b8adbe ("clk: bcm2835: Support for clock parent selection")
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Acked-by: Stephen Boyd <sboyd@kernel.org>
Reviewed-by: Nicolas Saenz Julienne <nsaenz@kernel.org>
Tested-by: Nicolas Saenz Julienne <nsaenz@kernel.org> # boot and basic functionality
Tested-by: Michael Stapelberg <michael@stapelberg.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210922125419.4125779-2-maxime@cerno.tech
Use devm_rpi_firmware_get() so as to make sure we release RPi's firmware
interface when unbinding the device.
Signed-off-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Acked-by: Stephen Boyd <sboyd@kernel.org>
drivers/clk/bcm/clk-raspberrypi.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
Pull clk updates from Stephen Boyd:
"This contains no changes to the core framework. It is a collection of
various clk driver updates.
The biggest driver updates in terms of lines of code is the Allwinner
driver, closely followed by the Qualcomm and Mediatek drivers. All of
those hit high because we add so many lines of clk data. Coming in
fourth place is i.MX which also adds a bunch of clk data. This
accounts for the new driver additions this time around.
Otherwise the patches are lots of little cleanups and fixes for
various clk drivers that have baked in linux-next for a while. I
suppose one highlight or theme is that more clk drivers are being
updated to work as modules, which is interesting to see such critical
SoC infrastructure work as a loadable module.
New Drivers:
- Support qcom SM8150/SM8250 video and display clks
- Support Mediatek MT8167 clks
- Add clock for CRC block found on vf610 SoCs
- Add support for the Renesas R-Car V3U (R8A779A0) SoC
- Add support for the VSP for Resizing clock on Renesas RZ/G1H
- Support Allwinner A100 SoC clks
Removed Drivers:
- Remove i.MX21 clock driver, as i.MX21 platform support is being
dropped
Updates:
- Change how qcom's display port clks work
- Small non-critical fixes for TI clk driver
- Remove various unused variables in clk drivers
- Allow Rockchip clk driver to be a module
- Remove most __clk_lookup() calls in Samsung drivers (yay!)
- Support building i.MX ARMv8 platforms clock driver as module
- Some kerneldoc fixes here and there
- A couple of minor i.MX clk data corrections
- Update audio clock inverter and fdiv2 flag on Amlogic g12
- Make amlogic clk drivers configurable in Kconfig
- Fix Renesas VSP clock names to match corrected hardware
documentation
- Sigma-delta modulation on Allwinner R40
- Various fixes for at91 clk driver
- Use semicolons instead of commas in some places
- Mark some variables const so they can move to RO memory"
* tag 'clk-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux: (102 commits)
clk: imx8mq: Fix usdhc parents order
clk: qcom: gdsc: Keep RETAIN_FF bit set if gdsc is already on
clk: Restrict CLK_HSDK to ARC_SOC_HSDK
clk: at91: sam9x60: support only two programmable clocks
clk: ingenic: Respect CLK_SET_RATE_PARENT in .round_rate
clk: ingenic: Don't tag custom clocks with CLK_SET_RATE_PARENT
clk: ingenic: Don't use CLK_SET_RATE_GATE for PLL
clk: ingenic: Use readl_poll_timeout instead of custom loop
clk: ingenic: Use to_clk_info() macro for all clocks
clk: bcm2835: add missing release if devm_clk_hw_register fails
clk: at91: clk-sam9x60-pll: remove unused variable
clk: at91: clk-main: update key before writing AT91_CKGR_MOR
clk: at91: remove the checking of parent_name
clk: clk-prima2: fix return value check in prima2_clk_init()
clk: mmp2: Fix the display clock divider base
clk: pxa: Constify static struct clk_ops
clk: baikal-t1: Mark Ethernet PLL as critical
clk: qoriq: modify MAX_PLL_DIV to 32
clk: axi-clkgen: Set power bits for fractional mode
clk: axi-clkgen: Add support for fractional dividers
...
Contrary to previous SoCs, bcm2711 doesn't have a prescaler in the PLL
feedback loop. Bypass it by zeroing fb_prediv_mask when running on
bcm2711.
Note that, since the prediv configuration bits were re-purposed, this
was triggering miscalculations on all clocks hanging from the VPU clock,
notably the aux UART, making its output unintelligible.
Fixes: 42de9ad400 ("clk: bcm2835: Add BCM2711_CLOCK_EMMC2 support")
Reported-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
Link: https://lore.kernel.org/r/20200730182619.23246-1-nsaenzjulienne@suse.de
Tested-by: Nathan Chancellor <natechancellor@gmail.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
The RaspberryPi4 firmware actually exposes more clocks than are currently
handled by the driver and we will need to change some of them directly
based on the pixel rate for the display related clocks, or the load for the
GPU.
Since the firmware implements DVFS, this rate change can have a number of
side-effects, including adjusting the various PLL voltages or the PLL
parents. The firmware also implements thermal throttling, so even some
thermal pressure can change those parameters behind Linux back.
DVFS is currently implemented on the arm, core, h264, v3d, isp and hevc
clocks, so updating any of them using the MMIO driver (and thus behind the
firmware's back) can lead to troubles, the arm clock obviously being the
most problematic.
In order to make Linux play as nice as possible with those constraints, it
makes sense to rely on the firmware clocks as much as possible. However,
the firmware doesn't seem to provide some equivalents to their MMIO
counterparts, so we can't really replace that driver entirely.
Fortunately, the firmware has an interface to discover the clocks it
exposes.
Let's use it to discover, register the clocks in the clocks framework and
then expose them through the device tree for consumers to use them.
Cc: Michael Turquette <mturquette@baylibre.com>
Cc: Stephen Boyd <sboyd@kernel.org>
Cc: linux-clk@vger.kernel.org
Acked-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
Reviewed-by: Stephen Boyd <sboyd@kernel.org>
Tested-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/438d73962741a8c5f7c689319b7443b930a87fde.1592210452.git-series.maxime@cerno.tech
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
bcm2835_debugfs_clock_reg32 is never changed and can therefore be made
const.
This allows the compiler to put it in the text section instead of the
data section.
Before:
text data bss dec hex filename
26598 16088 64 42750 a6fe drivers/clk/bcm/clk-bcm2835.o
After:
text data bss dec hex filename
26662 16024 64 42750 a6fe drivers/clk/bcm/clk-bcm2835.o
Signed-off-by: Rikard Falkeborn <rikard.falkeborn@gmail.com>
Link: https://lkml.kernel.org/r/20200508220238.4883-1-rikard.falkeborn@gmail.com
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
There are four different callback functions that are used for the
clk_register callback that all have different second parameter types.
bcm2835_register_pll -> struct bcm2835_pll_data
bcm2835_register_pll_divider -> struct bcm2835_pll_divider_data
bcm2835_register_clock -> struct bcm2835_clock_data
bcm2835_register_date -> struct bcm2835_gate_data
These callbacks are cast to bcm2835_clk_register so that there is no
error about incompatible pointer types. Unfortunately, this is a control
flow integrity violation, which verifies that the callback function's
types match the prototypes exactly before jumping.
[ 0.857913] CFI failure (target: 0xffffff9334a81820):
[ 0.857977] WARNING: CPU: 3 PID: 35 at kernel/cfi.c:29 __cfi_check_fail+0x50/0x58
[ 0.857985] Modules linked in:
[ 0.858007] CPU: 3 PID: 35 Comm: kworker/3:1 Not tainted 4.19.123-v8-01301-gdbb48f16956e4-dirty #1
[ 0.858015] Hardware name: Raspberry Pi 3 Model B Rev 1.2 (DT)
[ 0.858031] Workqueue: events 0xffffff9334a925c8
[ 0.858046] pstate: 60000005 (nZCv daif -PAN -UAO)
[ 0.858058] pc : __cfi_check_fail+0x50/0x58
[ 0.858070] lr : __cfi_check_fail+0x50/0x58
[ 0.858078] sp : ffffff800814ba90
[ 0.858086] x29: ffffff800814ba90 x28: 000fffffffdfff3d
[ 0.858101] x27: 00000000002000c2 x26: ffffff93355fdb18
[ 0.858116] x25: 0000000000000000 x24: ffffff9334a81820
[ 0.858131] x23: ffffff93357f3580 x22: ffffff9334af1000
[ 0.858146] x21: a79b57e88f8ebc81 x20: ffffff93357f3580
[ 0.858161] x19: ffffff9334a81820 x18: fffffff679769070
[ 0.858175] x17: 0000000000000000 x16: 0000000000000000
[ 0.858190] x15: 0000000000000004 x14: 000000000000003c
[ 0.858205] x13: 0000000000003044 x12: 0000000000000000
[ 0.858220] x11: b57e91cd641bae00 x10: b57e91cd641bae00
[ 0.858235] x9 : b57e91cd641bae00 x8 : b57e91cd641bae00
[ 0.858250] x7 : 0000000000000000 x6 : ffffff933591d4e5
[ 0.858264] x5 : 0000000000000000 x4 : 0000000000000000
[ 0.858279] x3 : ffffff800814b718 x2 : ffffff9334a84818
[ 0.858293] x1 : ffffff9334bba66c x0 : 0000000000000029
[ 0.858308] Call trace:
[ 0.858321] __cfi_check_fail+0x50/0x58
[ 0.858337] __cfi_check+0x3ab3c/0x4467c
[ 0.858351] bcm2835_clk_probe+0x210/0x2dc
[ 0.858369] platform_drv_probe+0xb0/0xfc
[ 0.858380] really_probe+0x4a0/0x5a8
[ 0.858391] driver_probe_device+0x68/0x104
[ 0.858403] __device_attach_driver+0x100/0x148
[ 0.858418] bus_for_each_drv+0xb0/0x12c
[ 0.858431] __device_attach.llvm.17225159516306086099+0xc0/0x168
[ 0.858443] bus_probe_device+0x44/0xfc
[ 0.858455] deferred_probe_work_func+0xa0/0xe0
[ 0.858472] process_one_work+0x210/0x538
[ 0.858485] worker_thread+0x2e8/0x478
[ 0.858500] kthread+0x154/0x164
[ 0.858515] ret_from_fork+0x10/0x18
To fix this, change the second parameter of all functions void * and use
a local variable with the correct type so that everything works
properly. With this, the only use of bcm2835_clk_register is in struct
bcm2835_clk_desc so we can just remove it and use the type directly.
Fixes: 56eb3a2ed9 ("clk: bcm2835: remove use of BCM2835_CLOCK_COUNT in driver")
Link: https://github.com/ClangBuiltLinux/linux/issues/1028
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Link: https://lkml.kernel.org/r/20200516080806.1459784-2-natechancellor@gmail.com
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
bcm2835_register_gate is used as a callback for the clk_register member
of bcm2835_clk_desc, which expects a struct clk_hw * return type but
bcm2835_register_gate returns a struct clk *.
This discrepancy is hidden by the fact that bcm2835_register_gate is
cast to the typedef bcm2835_clk_register by the _REGISTER macro. This
turns out to be a control flow integrity violation, which is how this
was noticed.
Change the return type of bcm2835_register_gate to be struct clk_hw *
and use clk_hw_register_gate to do so. This should be a non-functional
change as clk_register_gate calls clk_hw_register_gate anyways but this
is needed to avoid issues with further changes.
Fixes: b19f009d45 ("clk: bcm2835: Migrate to clk_hw based registration and OF APIs")
Link: https://github.com/ClangBuiltLinux/linux/issues/1028
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Link: https://lkml.kernel.org/r/20200516080806.1459784-1-natechancellor@gmail.com
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
The VPU firmware assume that the PLLD_PER isn't modified by the ARM core.
Otherwise this could cause firmware lookups. So mark the clock as critical
to avoid this.
Signed-off-by: Stefan Wahren <wahrenst@gmx.net>
Reviewed-by: Eric Anholt <eric@anholt.net>
The new BCM2711 supports an additional clock for the emmc2 block.
So add a new compatible and register this clock only for BCM2711.
Signed-off-by: Stefan Wahren <wahrenst@gmx.net>
Reviewed-by: Matthias Brugger <mbrugger@suse.com>
Acked-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Eric Anholt <eric@anholt.net>
In order to support SoC specific clocks (e.g. emmc2 for BCM2711), we
extend the description with a SoC support flag. This approach avoids long
and mostly redundant lists of clock IDs. Since PLLH is specific to
BCM2835, we register only rest of the clocks as common to all SoC.
Suggested-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Stefan Wahren <wahrenst@gmx.net>
Reviewed-by: Matthias Brugger <mbrugger@suse.com>
Acked-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Eric Anholt <eric@anholt.net>
