[PATCH 02/19] ARM: OMAP4: PM: save/restore all CM1/2 settings in OFF mode

[t-kristo@ti.com (Tero Kristo)]

On Wed, 2012-05-09 at 16:27 -0700, Kevin Hilman wrote:
> Tero Kristo <t-kristo@ti.com> writes:
> 
> > From: Rajendra Nayak <rnayak@ti.com>
> >
> > Restore all CM1/2 module registers as they are lost in OFF mode.
> 
> Except they are still lost since nobody calls these new functions (in
> this patch.)  :)
> 
> For ease of review, it's preferred to add the *users* of new code in the
> same patch as the new code.

I'll fix this for the next version. I think this same comment applies to
patch #3 also.

> 
> > [nm at ti.com: minor clean ups]
> > Signed-off-by: Nishanth Menon <nm@ti.com>
> > Signed-off-by: Rajendra Nayak <rnayak@ti.com>
> > Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
> > Signed-off-by: Axel Haslam <axelhaslam@gmail.com>
> > Signed-off-by: Tero Kristo <t-kristo@ti.com>
> > ---
> >  arch/arm/mach-omap2/cm44xx.c |  322 ++++++++++++++++++++++++++++++++++++++++++
> >  arch/arm/mach-omap2/cm44xx.h |    2 +
> >  2 files changed, 324 insertions(+), 0 deletions(-)
> >
> > diff --git a/arch/arm/mach-omap2/cm44xx.c b/arch/arm/mach-omap2/cm44xx.c
> > index 535d66e..fb5465b 100644
> > --- a/arch/arm/mach-omap2/cm44xx.c
> > +++ b/arch/arm/mach-omap2/cm44xx.c
> > @@ -21,8 +21,11 @@
> >  #include "iomap.h"
> >  #include "common.h"
> >  #include "cm.h"
> > +#include "cm44xx.h"
> >  #include "cm1_44xx.h"
> >  #include "cm2_44xx.h"
> > +#include "cminst44xx.h"
> > +#include "prcm44xx.h"
> >  #include "cm-regbits-44xx.h"
> >  
> >  /* CM1 hardware module low-level functions */
> > @@ -50,3 +53,322 @@ void omap4_cm2_write_inst_reg(u32 val, s16 inst, u16 reg)
> >  {
> >  	__raw_writel(val, OMAP44XX_CM2_REGADDR(inst, reg));
> >  }
> > +
> > +#define MAX_CM_REGISTERS 51
> > +
> > +struct omap4_cm_reg {
> > +	u16 offset;
> > +	u32 val;
> > +};
> > +
> > +struct omap4_cm_regs {
> > +	u32 mod_off;
> > +	u32 no_reg;
> 
> minor: do these need to be u32?

u16 should be good enough to save space I guess, I'll try changing this
and see what happens.

> 
> > +	struct omap4_cm_reg reg[MAX_CM_REGISTERS];
> > +};
> > +
> > +static struct omap4_cm_regs cm1_regs[] = {
> > +	/* OMAP4430_CM1_OCP_SOCKET_MOD */
> > +	{ .mod_off = OMAP4430_CM1_OCP_SOCKET_INST, .no_reg = 1,
> > +	{{.offset = OMAP4_CM_CM1_PROFILING_CLKCTRL_OFFSET} },
> 
> For readability sake, I'd prefer to see this line indented.  And why
> the extra space before the final '}'}

These tables are horrible, would be better to get rid of them completely
but I guess that is not possible... I'll look at the indentation.

> 
> > +	},
> 
> [...]
> 
> > +static void omap4_cm1_prepare_off(void)
> > +{
> > +	u32 i, j;
> > +	struct omap4_cm_regs *cm_reg = cm1_regs;
> > +
> > +	for (i = 0; i < ARRAY_SIZE(cm1_regs); i++, cm_reg++) {
> > +		for (j = 0; j < cm_reg->no_reg; j++) {
> > +			cm_reg->reg[j].val =
> > +			    omap4_cminst_read_inst_reg(OMAP4430_CM1_PARTITION,
> > +						       cm_reg->mod_off,
> > +						       cm_reg->reg[j].offset);
> > +		}
> > +	}
> > +}
> > +
> > +static void omap4_cm2_prepare_off(void)
> > +{
> > +	u32 i, j;
> > +	struct omap4_cm_regs *cm_reg = cm2_regs;
> > +
> > +	for (i = 0; i < ARRAY_SIZE(cm2_regs); i++, cm_reg++) {
> > +		for (j = 0; j < cm_reg->no_reg; j++) {
> > +			cm_reg->reg[j].val =
> > +			    omap4_cminst_read_inst_reg(OMAP4430_CM2_PARTITION,
> > +						       cm_reg->mod_off,
> > +						       cm_reg->reg[j].offset);
> > +		}
> > +	}
> > +}
> 
> > +static void omap4_cm1_resume_off(void)
> > +{
> > +	u32 i, j;
> > +	struct omap4_cm_regs *cm_reg = cm1_regs;
> > +
> > +	for (i = 0; i < ARRAY_SIZE(cm1_regs); i++, cm_reg++) {
> > +		for (j = 0; j < cm_reg->no_reg; j++) {
> > +			omap4_cminst_write_inst_reg(cm_reg->reg[j].val,
> > +						    OMAP4430_CM1_PARTITION,
> > +						    cm_reg->mod_off,
> > +						    cm_reg->reg[j].offset);
> > +		}
> > +	}
> > +}
> > +
> > +static void omap4_cm2_resume_off(void)
> > +{
> > +	u32 i, j;
> > +	struct omap4_cm_regs *cm_reg = cm2_regs;
> > +
> > +	for (i = 0; i < ARRAY_SIZE(cm2_regs); i++, cm_reg++) {
> > +		for (j = 0; j < cm_reg->no_reg; j++) {
> > +			omap4_cminst_write_inst_reg(cm_reg->reg[j].val,
> > +						    OMAP4430_CM2_PARTITION,
> > +						    cm_reg->mod_off,
> > +						    cm_reg->reg[j].offset);
> > +		}
> > +	}
> > +}
> 
> Notice the two prpare functions (and resume functions) are basically
> identical, except for the partition.
> 
> How about adding a .partition field to the struct so there can be a
> single function}

Yea, should be possible. I'll fix this for next rev.

-Tero