Message ID | 1448894605-30097-1-git-send-email-t-kristo@ti.com (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
On 11/30, Tero Kristo wrote: > Errata i810 states that DPLL controller can get stuck while transitioning > to a power saving state, while its M/N ratio is being re-programmed. > > As a workaround, before re-programming the M/N ratio, SW has to ensure > the DPLL cannot start an idle state transition. SW can disable DPLL > idling by setting the DPLL AUTO_DPLL_MODE=0 or keeping a clock request > active by setting a dependent clock domain in SW_WKUP. > > This errata impacts OMAP5 and DRA7 chips, so enable the errata for these. > > Signed-off-by: Tero Kristo <t-kristo@ti.com> > --- Acked-by: Stephen Boyd <sboyd@codeaurora.org>
* Tero Kristo <t-kristo@ti.com> [151130 06:44]: > + /* > + * Errata i810 - DPLL controller can get stuck while transitioning > + * to a power saving state. Software must ensure the DPLL can not > + * transition to a low power state while changing M/N values. > + * Easiest way to accomplish this is to prevent DPLL autoidle > + * before doing the M/N re-program. > + */ > + errata_i810 = ti_clk_get_features()->flags & TI_CLK_ERRATA_I810; > + > + if (errata_i810) { > + ai = omap3_dpll_autoidle_read(clk); > + if (ai) { > + omap3_dpll_deny_idle(clk); > + > + /* OCP barrier */ > + omap3_dpll_autoidle_read(clk); > + } > + } Should we just do this unconditionally? It seems like disabling the autoidle always before reprogramming is a good idea. Regards, Tony
On 12/03/2015 06:48 PM, Tony Lindgren wrote: > * Tero Kristo <t-kristo@ti.com> [151130 06:44]: >> + /* >> + * Errata i810 - DPLL controller can get stuck while transitioning >> + * to a power saving state. Software must ensure the DPLL can not >> + * transition to a low power state while changing M/N values. >> + * Easiest way to accomplish this is to prevent DPLL autoidle >> + * before doing the M/N re-program. >> + */ >> + errata_i810 = ti_clk_get_features()->flags & TI_CLK_ERRATA_I810; >> + >> + if (errata_i810) { >> + ai = omap3_dpll_autoidle_read(clk); >> + if (ai) { >> + omap3_dpll_deny_idle(clk); >> + >> + /* OCP barrier */ >> + omap3_dpll_autoidle_read(clk); >> + } >> + } > > Should we just do this unconditionally? It seems like disabling the > autoidle always before reprogramming is a good idea. Well, that is a few extra register accesses, but given the DPLL re-programming is a slow operation it probably does not matter. Let me spin a new version of this patch, it will avoid the need for the errata flag also. -Tero
* Tero Kristo <t-kristo@ti.com> [151211 00:42]: > On 12/03/2015 06:48 PM, Tony Lindgren wrote: > >* Tero Kristo <t-kristo@ti.com> [151130 06:44]: > >>+ /* > >>+ * Errata i810 - DPLL controller can get stuck while transitioning > >>+ * to a power saving state. Software must ensure the DPLL can not > >>+ * transition to a low power state while changing M/N values. > >>+ * Easiest way to accomplish this is to prevent DPLL autoidle > >>+ * before doing the M/N re-program. > >>+ */ > >>+ errata_i810 = ti_clk_get_features()->flags & TI_CLK_ERRATA_I810; > >>+ > >>+ if (errata_i810) { > >>+ ai = omap3_dpll_autoidle_read(clk); > >>+ if (ai) { > >>+ omap3_dpll_deny_idle(clk); > >>+ > >>+ /* OCP barrier */ > >>+ omap3_dpll_autoidle_read(clk); > >>+ } > >>+ } > > > >Should we just do this unconditionally? It seems like disabling the > >autoidle always before reprogramming is a good idea. > > Well, that is a few extra register accesses, but given the DPLL > re-programming is a slow operation it probably does not matter. Let me spin > a new version of this patch, it will avoid the need for the errata flag > also. OK thanks, Tony
diff --git a/arch/arm/mach-omap2/clock.c b/arch/arm/mach-omap2/clock.c index acb60ed..d058125 100644 --- a/arch/arm/mach-omap2/clock.c +++ b/arch/arm/mach-omap2/clock.c @@ -225,5 +225,9 @@ void __init ti_clk_init_features(void) if (omap_rev() == OMAP3430_REV_ES1_0) features.flags |= TI_CLK_DPLL4_DENY_REPROGRAM; + /* Errata I810 for omap5 / dra7 */ + if (soc_is_omap54xx() || soc_is_dra7xx()) + features.flags |= TI_CLK_ERRATA_I810; + ti_clk_setup_features(&features); } diff --git a/drivers/clk/ti/dpll3xxx.c b/drivers/clk/ti/dpll3xxx.c index f4dec00..1c30038 100644 --- a/drivers/clk/ti/dpll3xxx.c +++ b/drivers/clk/ti/dpll3xxx.c @@ -305,8 +305,9 @@ static void _lookup_sddiv(struct clk_hw_omap *clk, u8 *sd_div, u16 m, u8 n) static int omap3_noncore_dpll_program(struct clk_hw_omap *clk, u16 freqsel) { struct dpll_data *dd = clk->dpll_data; - u8 dco, sd_div; + u8 dco, sd_div, ai = 0; u32 v; + bool errata_i810; /* 3430 ES2 TRM: 4.7.6.9 DPLL Programming Sequence */ _omap3_noncore_dpll_bypass(clk); @@ -350,6 +351,25 @@ static int omap3_noncore_dpll_program(struct clk_hw_omap *clk, u16 freqsel) v |= sd_div << __ffs(dd->sddiv_mask); } + /* + * Errata i810 - DPLL controller can get stuck while transitioning + * to a power saving state. Software must ensure the DPLL can not + * transition to a low power state while changing M/N values. + * Easiest way to accomplish this is to prevent DPLL autoidle + * before doing the M/N re-program. + */ + errata_i810 = ti_clk_get_features()->flags & TI_CLK_ERRATA_I810; + + if (errata_i810) { + ai = omap3_dpll_autoidle_read(clk); + if (ai) { + omap3_dpll_deny_idle(clk); + + /* OCP barrier */ + omap3_dpll_autoidle_read(clk); + } + } + ti_clk_ll_ops->clk_writel(v, dd->mult_div1_reg); /* Set 4X multiplier and low-power mode */ @@ -379,6 +399,9 @@ static int omap3_noncore_dpll_program(struct clk_hw_omap *clk, u16 freqsel) _omap3_noncore_dpll_lock(clk); + if (errata_i810 && ai) + omap3_dpll_allow_idle(clk); + return 0; } diff --git a/include/linux/clk/ti.h b/include/linux/clk/ti.h index 223be69..75205df 100644 --- a/include/linux/clk/ti.h +++ b/include/linux/clk/ti.h @@ -286,6 +286,7 @@ struct ti_clk_features { #define TI_CLK_DPLL_HAS_FREQSEL BIT(0) #define TI_CLK_DPLL4_DENY_REPROGRAM BIT(1) #define TI_CLK_DISABLE_CLKDM_CONTROL BIT(2) +#define TI_CLK_ERRATA_I810 BIT(3) void ti_clk_setup_features(struct ti_clk_features *features); const struct ti_clk_features *ti_clk_get_features(void);
Errata i810 states that DPLL controller can get stuck while transitioning to a power saving state, while its M/N ratio is being re-programmed. As a workaround, before re-programming the M/N ratio, SW has to ensure the DPLL cannot start an idle state transition. SW can disable DPLL idling by setting the DPLL AUTO_DPLL_MODE=0 or keeping a clock request active by setting a dependent clock domain in SW_WKUP. This errata impacts OMAP5 and DRA7 chips, so enable the errata for these. Signed-off-by: Tero Kristo <t-kristo@ti.com> --- arch/arm/mach-omap2/clock.c | 4 ++++ drivers/clk/ti/dpll3xxx.c | 25 ++++++++++++++++++++++++- include/linux/clk/ti.h | 1 + 3 files changed, 29 insertions(+), 1 deletion(-)