Hi Jon,
On 06/04/2012 09:16 AM, Rajendra Nayak wrote:
+/* struct clksel_rate.flags possibilities */
+#define RATE_IN_242X (1<< 0)
+#define RATE_IN_243X (1<< 1)
+#define RATE_IN_3430ES1 (1<< 2) /* 3430ES1 rates only */
+#define RATE_IN_3430ES2PLUS (1<< 3) /* 3430 ES>= 2 rates only */
+#define RATE_IN_36XX (1<< 4)
+#define RATE_IN_4430 (1<< 5)
+#define RATE_IN_TI816X (1<< 6)
+#define RATE_IN_4460 (1<< 7)
+#define RATE_IN_AM33XX (1<< 8)
+#define RATE_IN_TI814X (1<< 9)
+
+#define RATE_IN_24XX (RATE_IN_242X | RATE_IN_243X)
+#define RATE_IN_34XX (RATE_IN_3430ES1 | RATE_IN_3430ES2PLUS)
+#define RATE_IN_3XXX (RATE_IN_34XX | RATE_IN_36XX)
+#define RATE_IN_44XX (RATE_IN_4430 | RATE_IN_4460)
+
+/* RATE_IN_3430ES2PLUS_36XX includes 34xx/35xx with ES>=2, and all
36xx/37xx */
+#define RATE_IN_3430ES2PLUS_36XX (RATE_IN_3430ES2PLUS |
RATE_IN_36XX)
+
+/* Platform flags for the clkdev-OMAP integration code */
+#define CK_242X (1<< 4)
+#define CK_243X (1<< 5) /* 243x, 253x */
+#define CK_3430ES1 (1<< 6) /* 34xxES1 only */
+#define CK_3430ES2PLUS (1<< 7) /* 34xxES2, ES3,
non-Sitara 35xx only */
+#define CK_3505 (1<< 8)
+#define CK_3517 (1<< 9)
+#define CK_36XX (1<< 10) /* 36xx/37xx-specific clocks */
+#define CK_443X (1<< 11)
+#define CK_TI816X (1<< 12)
+#define CK_446X (1<< 13)
+
+#define CK_34XX (CK_3430ES1 | CK_3430ES2PLUS)
+#define CK_AM35XX (CK_3505 | CK_3517) /* all Sitara AM35xx */
+#define CK_3XXX (CK_34XX | CK_AM35XX | CK_36XX)
I am not sure why we should duplicate these defines in an OMAP2 specific
header. What not just leave in plat clock.h where we have all the
RATE_IN_xxx and CK_xxx for all OMAP devices?
These get removed from the file which is used for OMAP1 in a later
patch. Like I said the idea was to separate out whats needed for OMAP1
(using legacy struct clk) and OMAP2+ (using common struct clk) with
both headers residing in respective mach-omap folders. (The RFC I posted
still had the OMAP1 file in plat-omap)
But these definitions are unrelated to whether you use common clock or
legacy. So why move them?
I am really fine either way, I don;t see a big issue in keeping these
in plat clock.h and some others with ifdefs around for COMMON_CLK.
The bigger issue is moving OMAP1 to common clk and I don't plan to do
it as part of this series.
+/**
+ * struct clksel_rate - register bitfield values corresponding to
clk divisors
+ * @val: register bitfield value (shifted to bit 0)
+ * @div: clock divisor corresponding to @val
+ * @flags: (see "struct clksel_rate.flags possibilities" above)
+ *
+ * @val should match the value of a read from struct clk.clksel_reg
+ * AND'ed with struct clk.clksel_mask, shifted right to bit 0.
+ *
+ * @div is the divisor that should be applied to the parent clock's
rate
+ * to produce the current clock's rate.
+ */
+struct clksel_rate {
+ u32 val;
+ u8 div;
+ u8 flags;
+};
+
+/**
+ * struct clksel - available parent clocks, and a pointer to their
divisors
+ * @parent: struct clk * to a possible parent clock
+ * @rates: available divisors for this parent clock
+ *
+ * A struct clksel is always associated with one or more struct clks
+ * and one or more struct clksel_rates.
+ */
+struct clksel {
+ struct clk *parent;
+ const struct clksel_rate *rates;
+};
These above clksel structures would be need for omap1 devices so that we
could use the clock framework to set the parent clock. So why not keep
in plat clock.h?
We could, but that alone wouldn't be enough if we move OMAP2+ alone to
common clk, it would mean we duplicate the clksel handling code too,
and if we do that, maybe its not that bad to just duplicate a couple
more struct definitions.
So I have tested the legacy clksel code on omap1 and it works. So I
would hope that the common clock version of clksel would work too for
omap1 (if we move omap1 to the common clock framework).
Yes, *if we move omap1 to common clock* :-)
regards,
Rajendra
+
+struct clk_hw_omap_ops;
+
+struct clk_hw_omap {
+ struct clk_hw hw;
+ struct list_head node;
+ unsigned long fixed_rate;
+ u8 fixed_div;
+ void __iomem *enable_reg;
+ u8 enable_bit;
+ u8 flags;
+#ifdef CONFIG_ARCH_OMAP2PLUS
+ void __iomem *clksel_reg;
+ u32 clksel_mask;
+ const struct clksel *clksel;
+ struct dpll_data *dpll_data;
+ const char *clkdm_name;
+ struct clockdomain *clkdm;
+#else
+ u8 rate_offset;
+ u8 src_offset;
+#endif
+ const struct clk_hw_omap_ops *ops;
+};
+
+struct clk_hw_omap_ops {
+ void (*find_idlest)(struct clk_hw_omap *oclk,
+ void __iomem **idlest_reg,
+ u8 *idlest_bit, u8 *idlest_val);
+ void (*find_companion)(struct clk_hw_omap *oclk,
+ void __iomem **other_reg,
+ u8 *other_bit);
+ void (*allow_idle)(struct clk_hw_omap *oclk);
+ void (*deny_idle)(struct clk_hw_omap *oclk);
+};
The above clk_hw_xxx would also be needed for omap1 too, right?
Yes, when OMAP1 moves to common clk *and* if we find enough
commonalities in clk_hw_xxxx accross OMAP1 and OMAP2+.
Else it would make sense to keep them in separate mach folders.
I guess I was hoping that these would be put somewhere in plat-omap, may
be cclock.h so that the intent would be so that omap1 could use them. By
putting them in mach-omap2, then if we find that omap1 can use them we
are going to need to move them.
Moving all this into mach-omap2, appears to be going the other direction
I was expecting. In other words, really separating omap1 and omap2 code.
I was hoping we could consolidate it more.
Cheers
Jon
--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
