On Fri 01 Mar 2024 at 16:29, Alexey Romanov <[email protected]> wrote:
> Amlogic crypto IP doesn't take a clock input on some
> SoCs: AXG / A1 / S4 / G12. So make it optional.
>
I commented this patch on v2 and the comment keep on being un-addressed.
The SoC either:
* has a clock that is required for the IP to work
* Or does not
It is not something you are free to provide or not.
For the record, I find very hard believe that some SoC would have clock,
and other would not, for the same HW.
Isn't it more likely that the clock just happens to be left enabled by
the bootloader on some SoC and it conviently allows to ignore it ?
> Signed-off-by: Alexey Romanov <[email protected]>
> ---
> drivers/crypto/amlogic/amlogic-gxl-core.c | 14 +++-----------
> 1 file changed, 3 insertions(+), 11 deletions(-)
>
> diff --git a/drivers/crypto/amlogic/amlogic-gxl-core.c b/drivers/crypto/amlogic/amlogic-gxl-core.c
> index e9e733ed98e0..a3a69a59f476 100644
> --- a/drivers/crypto/amlogic/amlogic-gxl-core.c
> +++ b/drivers/crypto/amlogic/amlogic-gxl-core.c
> @@ -269,16 +269,11 @@ static int meson_crypto_probe(struct platform_device *pdev)
> dev_err(&pdev->dev, "Cannot request MMIO err=%d\n", err);
> return err;
> }
> - mc->busclk = devm_clk_get(&pdev->dev, "blkmv");
> +
> + mc->busclk = devm_clk_get_optional_enabled(&pdev->dev, "blkmv");
> if (IS_ERR(mc->busclk)) {
> err = PTR_ERR(mc->busclk);
> - dev_err(&pdev->dev, "Cannot get core clock err=%d\n", err);
> - return err;
> - }
> -
> - err = clk_prepare_enable(mc->busclk);
> - if (err != 0) {
> - dev_err(&pdev->dev, "Cannot prepare_enable busclk\n");
> + dev_err(&pdev->dev, "Cannot get and enable core clock err=%d\n", err);
> return err;
> }
>
> @@ -306,7 +301,6 @@ static int meson_crypto_probe(struct platform_device *pdev)
> meson_unregister_algs(mc);
> error_flow:
> meson_free_chanlist(mc, mc->flow_cnt - 1);
> - clk_disable_unprepare(mc->busclk);
> return err;
> }
>
> @@ -321,8 +315,6 @@ static void meson_crypto_remove(struct platform_device *pdev)
> meson_unregister_algs(mc);
>
> meson_free_chanlist(mc, mc->flow_cnt - 1);
> -
> - clk_disable_unprepare(mc->busclk);
> }
>
> static const struct meson_pdata meson_gxl_pdata = {
--
Jerome
Hello Jerome,
On Fri, Mar 01, 2024 at 04:21:20PM +0100, Jerome Brunet wrote:
>
> On Fri 01 Mar 2024 at 16:29, Alexey Romanov <[email protected]> wrote:
>
> > Amlogic crypto IP doesn't take a clock input on some
> > SoCs: AXG / A1 / S4 / G12. So make it optional.
> >
>
> I commented this patch on v2 and the comment keep on being un-addressed.
>
> The SoC either:
> * has a clock that is required for the IP to work
> * Or does not
>
> It is not something you are free to provide or not.
>
> For the record, I find very hard believe that some SoC would have clock,
> and other would not, for the same HW.
>
> Isn't it more likely that the clock just happens to be left enabled by
> the bootloader on some SoC and it conviently allows to ignore it ?
S905X and newer SoC's uses DMA engine for crypto HW and they
don't required clock input to work. Clock input is needed for
blkmv engine.
Therefore, I'm not sure that it is needed for GXL too (and the second
interrupt line). I tested it on vim1 board witouht them and everything
works correctly.
>
> > Signed-off-by: Alexey Romanov <[email protected]>
> > ---
> > drivers/crypto/amlogic/amlogic-gxl-core.c | 14 +++-----------
> > 1 file changed, 3 insertions(+), 11 deletions(-)
> >
> > diff --git a/drivers/crypto/amlogic/amlogic-gxl-core.c b/drivers/crypto/amlogic/amlogic-gxl-core.c
> > index e9e733ed98e0..a3a69a59f476 100644
> > --- a/drivers/crypto/amlogic/amlogic-gxl-core.c
> > +++ b/drivers/crypto/amlogic/amlogic-gxl-core.c
> > @@ -269,16 +269,11 @@ static int meson_crypto_probe(struct platform_device *pdev)
> > dev_err(&pdev->dev, "Cannot request MMIO err=%d\n", err);
> > return err;
> > }
> > - mc->busclk = devm_clk_get(&pdev->dev, "blkmv");
> > +
> > + mc->busclk = devm_clk_get_optional_enabled(&pdev->dev, "blkmv");
> > if (IS_ERR(mc->busclk)) {
> > err = PTR_ERR(mc->busclk);
> > - dev_err(&pdev->dev, "Cannot get core clock err=%d\n", err);
> > - return err;
> > - }
> > -
> > - err = clk_prepare_enable(mc->busclk);
> > - if (err != 0) {
> > - dev_err(&pdev->dev, "Cannot prepare_enable busclk\n");
> > + dev_err(&pdev->dev, "Cannot get and enable core clock err=%d\n", err);
> > return err;
> > }
> >
> > @@ -306,7 +301,6 @@ static int meson_crypto_probe(struct platform_device *pdev)
> > meson_unregister_algs(mc);
> > error_flow:
> > meson_free_chanlist(mc, mc->flow_cnt - 1);
> > - clk_disable_unprepare(mc->busclk);
> > return err;
> > }
> >
> > @@ -321,8 +315,6 @@ static void meson_crypto_remove(struct platform_device *pdev)
> > meson_unregister_algs(mc);
> >
> > meson_free_chanlist(mc, mc->flow_cnt - 1);
> > -
> > - clk_disable_unprepare(mc->busclk);
> > }
> >
> > static const struct meson_pdata meson_gxl_pdata = {
>
>
> --
> Jerome
--
Thank you,
Alexey
On Mon, Mar 04, 2024 at 01:49:46PM +0000, Alexey Romanov wrote:
> Hello Jerome,
>
> On Fri, Mar 01, 2024 at 04:21:20PM +0100, Jerome Brunet wrote:
> >
> > On Fri 01 Mar 2024 at 16:29, Alexey Romanov <[email protected]> wrote:
> >
> > > Amlogic crypto IP doesn't take a clock input on some
> > > SoCs: AXG / A1 / S4 / G12. So make it optional.
> > >
> >
> > I commented this patch on v2 and the comment keep on being un-addressed.
> >
> > The SoC either:
> > * has a clock that is required for the IP to work
> > * Or does not
> >
> > It is not something you are free to provide or not.
> >
> > For the record, I find very hard believe that some SoC would have clock,
> > and other would not, for the same HW.
> >
> > Isn't it more likely that the clock just happens to be left enabled by
> > the bootloader on some SoC and it conviently allows to ignore it ?
>
>
> S905X and newer SoC's uses DMA engine for crypto HW and they
> don't required clock input to work. Clock input is needed for
> blkmv engine.
>
> Therefore, I'm not sure that it is needed for GXL too (and the second
> interrupt line). I tested it on vim1 board witouht them and everything
> works correctly.
Amlogic says that the crypto HW based on DMA engine doesn't require
a clock input. GXL uses DMA engine, so, I think we have to remove
whole clock controller calls in the next series from driver/dts/bindings.
And the second interrupt line from crypto node in meson-gxl.dtsi too.
>
> >
> > > Signed-off-by: Alexey Romanov <[email protected]>
> > > ---
> > > drivers/crypto/amlogic/amlogic-gxl-core.c | 14 +++-----------
> > > 1 file changed, 3 insertions(+), 11 deletions(-)
> > >
> > > diff --git a/drivers/crypto/amlogic/amlogic-gxl-core.c b/drivers/crypto/amlogic/amlogic-gxl-core.c
> > > index e9e733ed98e0..a3a69a59f476 100644
> > > --- a/drivers/crypto/amlogic/amlogic-gxl-core.c
> > > +++ b/drivers/crypto/amlogic/amlogic-gxl-core.c
> > > @@ -269,16 +269,11 @@ static int meson_crypto_probe(struct platform_device *pdev)
> > > dev_err(&pdev->dev, "Cannot request MMIO err=%d\n", err);
> > > return err;
> > > }
> > > - mc->busclk = devm_clk_get(&pdev->dev, "blkmv");
> > > +
> > > + mc->busclk = devm_clk_get_optional_enabled(&pdev->dev, "blkmv");
> > > if (IS_ERR(mc->busclk)) {
> > > err = PTR_ERR(mc->busclk);
> > > - dev_err(&pdev->dev, "Cannot get core clock err=%d\n", err);
> > > - return err;
> > > - }
> > > -
> > > - err = clk_prepare_enable(mc->busclk);
> > > - if (err != 0) {
> > > - dev_err(&pdev->dev, "Cannot prepare_enable busclk\n");
> > > + dev_err(&pdev->dev, "Cannot get and enable core clock err=%d\n", err);
> > > return err;
> > > }
> > >
> > > @@ -306,7 +301,6 @@ static int meson_crypto_probe(struct platform_device *pdev)
> > > meson_unregister_algs(mc);
> > > error_flow:
> > > meson_free_chanlist(mc, mc->flow_cnt - 1);
> > > - clk_disable_unprepare(mc->busclk);
> > > return err;
> > > }
> > >
> > > @@ -321,8 +315,6 @@ static void meson_crypto_remove(struct platform_device *pdev)
> > > meson_unregister_algs(mc);
> > >
> > > meson_free_chanlist(mc, mc->flow_cnt - 1);
> > > -
> > > - clk_disable_unprepare(mc->busclk);
> > > }
> > >
> > > static const struct meson_pdata meson_gxl_pdata = {
> >
> >
> > --
> > Jerome
>
> --
> Thank you,
> Alexey
--
Thank you,
Alexey