| Message ID | 1335510185-7906-10-git-send-email-richard.zhao@freescale.com |
|---|---|
| State | New |
| Headers | show |
On Fri, Apr 27, 2012 at 03:03:03PM +0800, Richard Zhao wrote: > It's used by audio drivers. > > Signed-off-by: Richard Zhao <richard.zhao@freescale.com> > --- > arch/arm/mach-imx/clk-imx6q.c | 1 + > 1 files changed, 1 insertions(+), 0 deletions(-) > > diff --git a/arch/arm/mach-imx/clk-imx6q.c b/arch/arm/mach-imx/clk-imx6q.c > index f40a35d..9a03dcc 100644 > --- a/arch/arm/mach-imx/clk-imx6q.c > +++ b/arch/arm/mach-imx/clk-imx6q.c > @@ -418,6 +418,7 @@ int __init mx6q_clocks_init(void) > clk_register_clkdev(clk[sdma], NULL, "20ec000.sdma"); > clk_register_clkdev(clk[dummy], NULL, "20bc000.wdog"); > clk_register_clkdev(clk[dummy], NULL, "20c0000.wdog"); > + clk_register_clkdev(clk[ssi1], NULL, "2028000.ssi"); The ssi clock needs a general cleanup on all i.MX just like I cleaned up the other units. The SSI unit has at least a register clock and a baud clock. What the driver requests and enables is the register clock. The baud clock is currently unused and is needed only for master mode (which is not implemented in the ssi driver) So where we want to come to is: clk_register_clkdev(clk[ipg], "ipg", "2028000.ssi"); clk_register_clkdev(clk[ssi1], "baud", "2028000.ssi"); Sascha
On Fri, Apr 27, 2012 at 10:04:12AM +0200, Sascha Hauer wrote: > On Fri, Apr 27, 2012 at 03:03:03PM +0800, Richard Zhao wrote: > > It's used by audio drivers. > > > > Signed-off-by: Richard Zhao <richard.zhao@freescale.com> > > --- > > arch/arm/mach-imx/clk-imx6q.c | 1 + > > 1 files changed, 1 insertions(+), 0 deletions(-) > > > > diff --git a/arch/arm/mach-imx/clk-imx6q.c b/arch/arm/mach-imx/clk-imx6q.c > > index f40a35d..9a03dcc 100644 > > --- a/arch/arm/mach-imx/clk-imx6q.c > > +++ b/arch/arm/mach-imx/clk-imx6q.c > > @@ -418,6 +418,7 @@ int __init mx6q_clocks_init(void) > > clk_register_clkdev(clk[sdma], NULL, "20ec000.sdma"); > > clk_register_clkdev(clk[dummy], NULL, "20bc000.wdog"); > > clk_register_clkdev(clk[dummy], NULL, "20c0000.wdog"); > > + clk_register_clkdev(clk[ssi1], NULL, "2028000.ssi"); > > The ssi clock needs a general cleanup on all i.MX just like I cleaned up > the other units. The SSI unit has at least a register clock and a baud > clock. What the driver requests and enables is the register clock. > > The baud clock is currently unused and is needed only for master mode > (which is not implemented in the ssi driver) Are you sure for that? If I don't enable clk[ssi1], the ssi will not work. > > So where we want to come to is: > > clk_register_clkdev(clk[ipg], "ipg", "2028000.ssi"); ssi don't have ipg gate. We can let it always on for imx6q. Thanks Richard > clk_register_clkdev(clk[ssi1], "baud", "2028000.ssi"); > > Sascha > > > -- > Pengutronix e.K. | | > Industrial Linux Solutions | http://www.pengutronix.de/ | > Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0 | > Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 | >
On Fri, Apr 27, 2012 at 05:10:56PM +0800, Richard Zhao wrote: > On Fri, Apr 27, 2012 at 10:04:12AM +0200, Sascha Hauer wrote: > > On Fri, Apr 27, 2012 at 03:03:03PM +0800, Richard Zhao wrote: > > > It's used by audio drivers. > > > > > > Signed-off-by: Richard Zhao <richard.zhao@freescale.com> > > > --- > > > arch/arm/mach-imx/clk-imx6q.c | 1 + > > > 1 files changed, 1 insertions(+), 0 deletions(-) > > > > > > diff --git a/arch/arm/mach-imx/clk-imx6q.c b/arch/arm/mach-imx/clk-imx6q.c > > > index f40a35d..9a03dcc 100644 > > > --- a/arch/arm/mach-imx/clk-imx6q.c > > > +++ b/arch/arm/mach-imx/clk-imx6q.c > > > @@ -418,6 +418,7 @@ int __init mx6q_clocks_init(void) > > > clk_register_clkdev(clk[sdma], NULL, "20ec000.sdma"); > > > clk_register_clkdev(clk[dummy], NULL, "20bc000.wdog"); > > > clk_register_clkdev(clk[dummy], NULL, "20c0000.wdog"); > > > + clk_register_clkdev(clk[ssi1], NULL, "2028000.ssi"); > > > > The ssi clock needs a general cleanup on all i.MX just like I cleaned up > > the other units. The SSI unit has at least a register clock and a baud > > clock. What the driver requests and enables is the register clock. > > > > The baud clock is currently unused and is needed only for master mode > > (which is not implemented in the ssi driver) > Are you sure for that? If I don't enable clk[ssi1], the ssi will not > work. > > > > > So where we want to come to is: > > > > clk_register_clkdev(clk[ipg], "ipg", "2028000.ssi"); > ssi don't have ipg gate. We can let it always on for imx6q. Can you please ask your IC guys for clarification? For example on i.MX5 we have a ssi ipg clock and a ssi serial clock. Both can be gated with two individual gate bits. The i.MX6 datasheet (and also several other i.MX datasheets) is quite nebulous. The i.MX6 has only one gate bit for each SSI unit, but it's not clear if this bit actually gates both the ipg and serial clock or only one of them. My general idea is that each unit in the SoC has different input clocks. In the driver we need to clk_get() the input clocks. If a given SoC has no software control over some of the devices input clocks, then we need to provide a dummy for this, because other SoCs have control over the clock. Sascha
On Fri, Apr 27, 2012 at 11:25:29AM +0200, Sascha Hauer wrote: > On Fri, Apr 27, 2012 at 05:10:56PM +0800, Richard Zhao wrote: > > On Fri, Apr 27, 2012 at 10:04:12AM +0200, Sascha Hauer wrote: > > > On Fri, Apr 27, 2012 at 03:03:03PM +0800, Richard Zhao wrote: > > > > It's used by audio drivers. > > > > > > > > Signed-off-by: Richard Zhao <richard.zhao@freescale.com> > > > > --- > > > > arch/arm/mach-imx/clk-imx6q.c | 1 + > > > > 1 files changed, 1 insertions(+), 0 deletions(-) > > > > > > > > diff --git a/arch/arm/mach-imx/clk-imx6q.c b/arch/arm/mach-imx/clk-imx6q.c > > > > index f40a35d..9a03dcc 100644 > > > > --- a/arch/arm/mach-imx/clk-imx6q.c > > > > +++ b/arch/arm/mach-imx/clk-imx6q.c > > > > @@ -418,6 +418,7 @@ int __init mx6q_clocks_init(void) > > > > clk_register_clkdev(clk[sdma], NULL, "20ec000.sdma"); > > > > clk_register_clkdev(clk[dummy], NULL, "20bc000.wdog"); > > > > clk_register_clkdev(clk[dummy], NULL, "20c0000.wdog"); > > > > + clk_register_clkdev(clk[ssi1], NULL, "2028000.ssi"); > > > > > > The ssi clock needs a general cleanup on all i.MX just like I cleaned up > > > the other units. The SSI unit has at least a register clock and a baud > > > clock. What the driver requests and enables is the register clock. > > > > > > The baud clock is currently unused and is needed only for master mode > > > (which is not implemented in the ssi driver) > > Are you sure for that? If I don't enable clk[ssi1], the ssi will not > > work. > > > > > > > > So where we want to come to is: > > > > > > clk_register_clkdev(clk[ipg], "ipg", "2028000.ssi"); > > ssi don't have ipg gate. We can let it always on for imx6q. > > Can you please ask your IC guys for clarification? > > For example on i.MX5 we have a ssi ipg clock and a ssi serial clock. > Both can be gated with two individual gate bits. > > The i.MX6 datasheet (and also several other i.MX datasheets) is quite > nebulous. The i.MX6 has only one gate bit for each SSI unit, but > it's not clear if this bit actually gates both the ipg and serial > clock or only one of them. You're right. ipg and serial clocks share the same gate. How do we handle it? I think it's not the only one and won't be last one. > > My general idea is that each unit in the SoC has different input clocks. > In the driver we need to clk_get() the input clocks. If a given SoC has > no software control over some of the devices input clocks, then we need > to provide a dummy for this, because other SoCs have control over the > clock. Great. Thanks Richard > > Sascha > > -- > Pengutronix e.K. | | > Industrial Linux Solutions | http://www.pengutronix.de/ | > Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0 | > Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 | > > _______________________________________________ > linux-arm-kernel mailing list > linux-arm-kernel@lists.infradead.org > http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
On Mon, Apr 30, 2012 at 10:01:46AM +0800, Richard Zhao wrote: > > > > > > > > clk_register_clkdev(clk[ipg], "ipg", "2028000.ssi"); > > > ssi don't have ipg gate. We can let it always on for imx6q. > > > > Can you please ask your IC guys for clarification? > > > > For example on i.MX5 we have a ssi ipg clock and a ssi serial clock. > > Both can be gated with two individual gate bits. > > > > The i.MX6 datasheet (and also several other i.MX datasheets) is quite > > nebulous. The i.MX6 has only one gate bit for each SSI unit, but > > it's not clear if this bit actually gates both the ipg and serial > > clock or only one of them. > You're right. ipg and serial clocks share the same gate. How do we > handle it? I think it's not the only one and won't be last one. We don't have support for a single gate gating two clocks right now and I don't know how this fits into the clock framework. We could pretend that only the ipg clock is gateable because this is needed anyway when the SSI unit is used. It seems we need a proper solution for this later. Sascha
On Mon, Apr 30, 2012 at 02:18:57PM +0200, Sascha Hauer wrote: > On Mon, Apr 30, 2012 at 10:01:46AM +0800, Richard Zhao wrote: > > > > > > > > > > clk_register_clkdev(clk[ipg], "ipg", "2028000.ssi"); > > > > ssi don't have ipg gate. We can let it always on for imx6q. > > > > > > Can you please ask your IC guys for clarification? > > > > > > For example on i.MX5 we have a ssi ipg clock and a ssi serial clock. > > > Both can be gated with two individual gate bits. > > > > > > The i.MX6 datasheet (and also several other i.MX datasheets) is quite > > > nebulous. The i.MX6 has only one gate bit for each SSI unit, but > > > it's not clear if this bit actually gates both the ipg and serial > > > clock or only one of them. > > You're right. ipg and serial clocks share the same gate. How do we > > handle it? I think it's not the only one and won't be last one. > > We don't have support for a single gate gating two clocks right now and > I don't know how this fits into the clock framework. > We could pretend that only the ipg clock is gateable because this is > needed anyway when the SSI unit is used. Shawn, will you add clk ssi_ipg? Thanks Richard > > It seems we need a proper solution for this later. > > Sascha > > > -- > Pengutronix e.K. | | > Industrial Linux Solutions | http://www.pengutronix.de/ | > Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0 | > Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 | >
On Wed, May 02, 2012 at 06:34:41PM +0800, Richard Zhao wrote: > Shawn, will you add clk ssi_ipg? > Ok, will do after Sascha's clk series gets settled on arm-soc tree.
On Wed, May 02, 2012 at 10:34:17PM +0800, Shawn Guo wrote: > On Wed, May 02, 2012 at 06:34:41PM +0800, Richard Zhao wrote: > > Shawn, will you add clk ssi_ipg? > > > Ok, will do after Sascha's clk series gets settled on arm-soc tree. I have a branch based on Mikes clk-next as of today, I will send it tomorrow. Sascha
diff --git a/arch/arm/mach-imx/clk-imx6q.c b/arch/arm/mach-imx/clk-imx6q.c index f40a35d..9a03dcc 100644 --- a/arch/arm/mach-imx/clk-imx6q.c +++ b/arch/arm/mach-imx/clk-imx6q.c @@ -418,6 +418,7 @@ int __init mx6q_clocks_init(void) clk_register_clkdev(clk[sdma], NULL, "20ec000.sdma"); clk_register_clkdev(clk[dummy], NULL, "20bc000.wdog"); clk_register_clkdev(clk[dummy], NULL, "20c0000.wdog"); + clk_register_clkdev(clk[ssi1], NULL, "2028000.ssi"); for (i = 0; i < ARRAY_SIZE(clks_init_on); i++) { c = clk_get_sys(clks_init_on[i], NULL);
It's used by audio drivers. Signed-off-by: Richard Zhao <richard.zhao@freescale.com> --- arch/arm/mach-imx/clk-imx6q.c | 1 + 1 files changed, 1 insertions(+), 0 deletions(-)