Use common clk drivers for DaVinci
A message that is not in the archive (sent off-list or not yet received): [email protected]
cover Use common clk drivers for DaVinci
Please ignore the diagram which is ill formatted. Ref the pdf document instead for the clock hardware diagram.
Murali Karicheri
Software Design Engineer
Quoted 155 lines
Quoted 155 lines
-----Original Message-----
From: Karicheri, Muralidharan
Sent: Tuesday, September 18, 2012 2:36 PM
To: Nori, Sekhar; Hilman, Kevin; linux at arm.linux.org.uk; sshtylyov at mvista.com; davinci-linux-open-source at linux.davincidsp.com; linux-arm-kernel at lists.infradead.org; mturquette at linaro.org; arnd at arndb.de
Cc: Chemparathy, Cyril; Karicheri, Muralidharan
Subject: [RFC PATCH 00/10] Use common clk drivers for DaVinciThis is the RFC patch for implementing clock tree for DaVinci SoCs using common clk framework. Currently
arch/arm/mach-davinci/clock.candarch/arm/mach-davinci/psc.cimplements clock drivers for DaVinci. This patch makes these code obsolete and migrate the SoC code to use the common clk based drivers and initialization. This adds two clk drivers specific to DaVinci and Keystone (found in c6x arch such as C6678) devices. Some of the existing clk drivers such as clk-fixed-rate, clk-divider, and clk-mux are re-used in addition to the DaVinci specific drivers to initialize the clock tree for the SoCs. The figure below is provided to help understand the hardware and the code implemented in this patch and more details can be found in SoC specific documentation from Texas Instruments. The figure below is borrowed from chapter 6 of
http://www.ti.com/lit/ug/sprue14c/sprue14c.pdf|-------| |---------| |-------| |--------| CLKIN -----| | | PLL | | | |PLLDIV1 |--SYSCLK1 | MUX |----| mult |-------| MUX |----|--------| OSCIN -----| | | | div | | | | |--------| |-------| | |---------| |---|-------| ---|PLLDIV2 |--SYSCLK2 | | | | | ---------- CLKMODE |---------------| PLLEN | ... | | |--------| | ---|PLLDIV5 |--SYSCLK5 | ---------- |----AUXCLK | |--------| |--| BPDIV |---SYSCLKBP |--------|Main PLL1 structure in DM644x
There are two Main PLLs in DM644x. PLL1 and PLL2. Each of these generate different clocks in the DM device through PLL dividers. Figure above shows this for PLL1. Similar hardware exists for PLL2 and the related output clocks. The hardware is similar in most of the DM SoCs. Some of the recent Keystone devices (c6678 under c6x architecture) include a slight variant of the PLL that implemented different registers for the multipliers and dividers. All of these devices include PLL dividers (PLLDIVx) and Power Sleep controllers (PSC). The SoCs defines various Power Domains and Clock domains and there are PSC modules that controls power and clocks to various hardware IPs of the SoC.
Following drivers are used for various clock hardware blocks:-
CLKIN and OSCIN - clk-fixed-rate (existing driver) MUX - clk-mux (existing driver)
PLLDIVx - clk-divider (existing driver) PLL mult/div - clk-davinci-pll (new driver)
clk-keystone-pll (new driver)
PSC -clk-davinci-psc.c(new driverThe patch series is split in two:- patch 1-5 for drivers and 6-10 for machine specific code. Please note that initially only DM644x is supported. The idea is to review the initial patch and get a feedback on the overall structure of the code organization. The other SoCs will be added in subsequent patch revisions. The driver code implements the features needed to support DM644x and is expected to be enhanced in subsequent patch revisions to support additional SoCs. I have boot tested this on DM6446 EVM I have. Also verified reboot command works and the clock rates are set as before. The patches depends on the following patches that I had sent for review earlier:
davinci spi driver preparation @
https://patchwork.kernel.org/patch/1389321/
davinci watchdog driver preparation @
https://lkml.org/lkml/2012/9/7/634
davinci nand driver preparation @
https://lkml.org/lkml/2012/9/7/635
davinci i2c driver preparation @
https://patchwork.kernel.org/patch/1388841/
davinci gpio driver preparation @
https://lkml.org/lkml/2012/8/31/341[RFC - PATCH] base:pm: prepare driver for common clock framework in the davinci mailing list
Following are to be discussed as part of this patch review.
1.
arch/arm/pm.c. This is configuring PLL controller registers for suspend and resume. It appears that we need to move this code toclk-davinci-pll.c. But I can't find APIs for suspend and resume in the common clk framework. How is this expected to work? Currently i have kept the code ASIS.2. There are usecount=1 in the old clock implementation for dsp, vicp and timer2 clocks.
CLK_IGNORE_UNUSEDflag is used currently to implement the same as disabling these unused clocks causes issues in boot up,NOTE: I am doing this work as a background activity and hence the progress will be slow. Please volunteer to help me in this effort by offering to test or migrating other devices to this framework.
Murali Karicheri (10):
clk:davinci - add Main PLL clock driver
clk:davinci - add PSC clock driver
clk:keystone - add Main PLL clock driver
clk:davinci - common clk driver initialization clk:davinci - add build infrastructure
ARM:davinci - restructure header files for common clk migration ARM:davinci - adding new type forcpu_clksinsoc_infoARM:davinci - migrating to usedavinci_common_clk_initARM:davinci - update the dm644x soc code to use common clk drivers ARM:davinci - switch to common clk frameworkarch/arm/Kconfig | 1 + arch/arm/mach-davinci/Kconfig | 2 + arch/arm/mach-davinci/Makefile | 9 + arch/arm/mach-davinci/common.c | 6 + arch/arm/mach-davinci/devices.c | 2 + arch/arm/mach-davinci/dm644x.c | 301 +++++++++++++++++++++++- arch/arm/mach-davinci/include/mach/common.h | 5 + arch/arm/mach-davinci/include/mach/pll.h | 81 +++++++ arch/arm/mach-davinci/include/mach/psc.h | 2 + arch/arm/mach-davinci/include/mach/time.h | 4 +- arch/arm/mach-davinci/pm.c | 4 + arch/arm/mach-davinci/sleep.S | 5 + arch/arm/mach-davinci/time.c | 13 + drivers/clk/Kconfig | 2 + drivers/clk/Makefile | 2 + drivers/clk/davinci/Kconfig | 42 ++++ drivers/clk/davinci/Makefile | 3 + drivers/clk/davinci/clk-davinci-pll.c | 128 ++++++++++ drivers/clk/davinci/clk-davinci-psc.c | 197 ++++++++++++++++ drivers/clk/davinci/davinci-clock.c | 210 +++++++++++++++++ drivers/clk/keystone/Kconfig | 6 + drivers/clk/keystone/Makefile | 1 + drivers/clk/keystone/clk-keystone-pll.c | 94 ++++++++ include/linux/platform_data/clk-davinci-pll.h | 52 ++++ include/linux/platform_data/clk-davinci-psc.h | 53 +++++ include/linux/platform_data/clk-keystone-pll.h | 34 +++ include/linux/platform_data/davinci-clock.h | 125 ++++++++++ 27 files changed, 1382 insertions(+), 2 deletions(-) create mode 100644 arch/arm/mach-davinci/include/mach/pll.h create mode 100644 drivers/clk/davinci/Kconfig create mode 100644 drivers/clk/davinci/Makefile create mode 100644 drivers/clk/davinci/clk-davinci-pll.c create mode 100644 drivers/clk/davinci/clk-davinci-psc.c create mode 100644 drivers/clk/davinci/davinci-clock.c create mode 100644 drivers/clk/keystone/Kconfig create mode 100644 drivers/clk/keystone/Makefile create mode 100644 drivers/clk/keystone/clk-keystone-pll.c create mode 100644 include/linux/platform_data/clk-davinci-pll.h create mode 100644 include/linux/platform_data/clk-davinci-psc.h create mode 100644 include/linux/platform_data/clk-keystone-pll.h create mode 100644 include/linux/platform_data/davinci-clock.h--
1.7.9.5
A message that is not in the archive (sent off-list or not yet received): [email protected]
RFC PATCH 8/10 ARM:davinci - migrating to use davinci_common_clk_init
Hello.
Quoted 14 lines · Murali Karicheri
The common clk code uses a new function
davinci_common_clk_init()defined indrivers/clk/davinci/davinci-clock.cto initialize the clk drivers. This function is now invoked intime.cas part ofdavinci_timer_init(). Currentlydavinci_clk_init()is called fromdavinci_common_init()which is too early to initialize common clk drivers. Also includepll.hinstead ofclock.hin some of the source files.Signed-off-by Murali Karicheri <[email protected]>arch/arm/mach-davinci/common.c +0
-0
[...]
arch/arm/mach-davinci/common.c void __init davinci_common_init(struct davinci_soc_info *soc_info) · line 111
Quoted 5 lines
arch/arm/mach-davinci/common.c +1
-0@@ -106,7 +108,9 @@ void __init davinci_common_init(struct davinci_soc_info *soc_info) goto err;if (davinci_soc_info.cpu_clks) { +#ifndef CONFIG_COMMON_CLK
Shouldn't you enclose in theis #ifndef the whole *if* statement?
Quoted 5 lines
arch/arm/mach-davinci/common.c +1
-0ret =davinci_clk_init (davinci_soc_info.cpu_clks); +#endif if (ret !=0 ) goto err;
[...]
arch/arm/mach-davinci/sleep.S ENDPROC(davinci_cpu_suspend) · line 190
Quoted 9 lines
arch/arm/mach-davinci/sleep.S +1
-0@@ -183,6 +187,7 @@ ENDPROC(davinci_cpu_suspend) * r1: contains virtual base for DDR2 Power and Sleep controller (PSC) * r2: contains PSC number for DDR2 */ +
Unrelated/unnecessary change?
arch/arm/mach-davinci/sleep.S ENDPROC(davinci_cpu_suspend) · line 193
Quoted 3 lines
arch/arm/mach-davinci/sleep.S +0
-0ENTRY(davinci_ddr_psc_config) /* Set next state in mdctl for DDR2 */ mov r6, #MDCTL
WBR, Sergei