diff mbox series

[net-next,1/6] net: stmmac: add platform library

Message ID E1qfiqd-007TPL-7K@rmk-PC.armlinux.org.uk (mailing list archive)
State New, archived
Headers show
Series net: stmmac: add and use library for setting clock | expand

Commit Message

Russell King (Oracle) Sept. 11, 2023, 3:29 p.m. UTC 
Add a platform library of helper functions for common traits in the
platform drivers. Currently, this is setting the tx clock.

Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
---
 drivers/net/ethernet/stmicro/stmmac/Makefile  |  2 +-
 .../ethernet/stmicro/stmmac/stmmac_plat_lib.c | 29 +++++++++++++++++++
 .../ethernet/stmicro/stmmac/stmmac_plat_lib.h |  8 +++++
 3 files changed, 38 insertions(+), 1 deletion(-)
 create mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_plat_lib.c
 create mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_plat_lib.h

Comments

Paolo Abeni Sept. 12, 2023, 10:18 a.m. UTC | #1 
On Mon, 2023-09-11 at 16:29 +0100, Russell King (Oracle) wrote:
> Add a platform library of helper functions for common traits in the
> platform drivers. Currently, this is setting the tx clock.
> 
> Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>

FTR, two copies of this series landed on the ML. I'm marking the no-
cover-letter version as Superseded, but please have a look at the
comments there, too.

Thanks,

Paolo
Simon Horman Sept. 12, 2023, 2:52 p.m. UTC | #2 
On Mon, Sep 11, 2023 at 04:29:11PM +0100, Russell King (Oracle) wrote:
> Add a platform library of helper functions for common traits in the
> platform drivers. Currently, this is setting the tx clock.
> 
> Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>

Hi Russell,

some minor issues raised by checkpatch follow.

> ---
>  drivers/net/ethernet/stmicro/stmmac/Makefile  |  2 +-
>  .../ethernet/stmicro/stmmac/stmmac_plat_lib.c | 29 +++++++++++++++++++
>  .../ethernet/stmicro/stmmac/stmmac_plat_lib.h |  8 +++++
>  3 files changed, 38 insertions(+), 1 deletion(-)
>  create mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_plat_lib.c
>  create mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_plat_lib.h
> 
> diff --git a/drivers/net/ethernet/stmicro/stmmac/Makefile b/drivers/net/ethernet/stmicro/stmmac/Makefile
> index 5b57aee19267..ba2cbfa0c9d1 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/Makefile
> +++ b/drivers/net/ethernet/stmicro/stmmac/Makefile
> @@ -6,7 +6,7 @@ stmmac-objs:= stmmac_main.o stmmac_ethtool.o stmmac_mdio.o ring_mode.o	\
>  	      mmc_core.o stmmac_hwtstamp.o stmmac_ptp.o dwmac4_descs.o	\
>  	      dwmac4_dma.o dwmac4_lib.o dwmac4_core.o dwmac5.o hwif.o \
>  	      stmmac_tc.o dwxgmac2_core.o dwxgmac2_dma.o dwxgmac2_descs.o \
> -	      stmmac_xdp.o \
> +	      stmmac_xdp.o stmmac_plat_lib.o \
>  	      $(stmmac-y)
>  
>  stmmac-$(CONFIG_STMMAC_SELFTESTS) += stmmac_selftests.o
> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_plat_lib.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_plat_lib.c
> new file mode 100644
> index 000000000000..abb9f512bb0e
> --- /dev/null
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_plat_lib.c
> @@ -0,0 +1,29 @@

Is an SPDX identifier appropriate here?

> +#include <linux/stmmac.h>
> +#include <linux/clk.h>
> +
> +#include "stmmac_plat_lib.h"
> +
> +int dwmac_set_tx_clk_gmii(struct clk *tx_clk, int speed)
> +{
> +	unsigned long rate;
> +
> +	switch (speed) {
> +	case SPEED_1000:
> +		rate = 125000000;
> +		break;
> +
> +	case SPEED_100:
> +		rate = 25000000;
> +		break;
> +
> +	case SPEED_10:
> +		rate = 2500000;
> +		break;
> +
> +	default:
> +		return -ENOTSUPP;

Checkpatch seems to think that EOPNOTSUPP would be more appropriate
as "ENOTSUPP is not a SUSV4 error code".

> +	}
> +
> +	return clk_set_rate(tx_clk, rate);
> +}
> +EXPORT_SYMBOL_GPL(dwmac_set_tx_clk_gmii);

...
Russell King (Oracle) Sept. 12, 2023, 10:20 p.m. UTC | #3 
On Tue, Sep 12, 2023 at 04:52:27PM +0200, Simon Horman wrote:
> On Mon, Sep 11, 2023 at 04:29:11PM +0100, Russell King (Oracle) wrote:
> > +	default:
> > +		return -ENOTSUPP;
> 
> Checkpatch seems to think that EOPNOTSUPP would be more appropriate
> as "ENOTSUPP is not a SUSV4 error code".

It needs to be an error code that clk_set_rate() below isn't going to
return - because if clk_set_rate() does return it, then the users are
going to end up issuing an incorrect error message to the user. I
suspect clk_set_rate() could quite legitimately return -EOPNOTSUPP
or -EINVAL.

Sadly, the CCF implementation of clk_set_rate() doesn't detail what
errors it could return, but it looks like -EBUSY, -EINVAL, or something
from pm_runtime_resume_and_get().

Interestingly, while looking at this, pm_runtime_resume_and_get() can
return '1' if e.g. rpm is disabled and the device is active. It looks
to me like CCF treats that as an error in multiple locations.

> > +	}
> > +
> > +	return clk_set_rate(tx_clk, rate);
> > +}
> > +EXPORT_SYMBOL_GPL(dwmac_set_tx_clk_gmii);
Simon Horman Sept. 14, 2023, 11:24 a.m. UTC | #4 
On Tue, Sep 12, 2023 at 11:20:55PM +0100, Russell King (Oracle) wrote:
> On Tue, Sep 12, 2023 at 04:52:27PM +0200, Simon Horman wrote:
> > On Mon, Sep 11, 2023 at 04:29:11PM +0100, Russell King (Oracle) wrote:
> > > +	default:
> > > +		return -ENOTSUPP;
> > 
> > Checkpatch seems to think that EOPNOTSUPP would be more appropriate
> > as "ENOTSUPP is not a SUSV4 error code".
> 
> It needs to be an error code that clk_set_rate() below isn't going to
> return - because if clk_set_rate() does return it, then the users are
> going to end up issuing an incorrect error message to the user. I
> suspect clk_set_rate() could quite legitimately return -EOPNOTSUPP
> or -EINVAL.
> 
> Sadly, the CCF implementation of clk_set_rate() doesn't detail what
> errors it could return, but it looks like -EBUSY, -EINVAL, or something
> from pm_runtime_resume_and_get().

Thanks Russell,

Understood.

In that case perhaps ENOTSUPP is not such a bad choice as:
a) it seems rather unlikely CCF would use it; and
b) the scope of usage is well contained - the helper and any direct callers.

No further objections from my side :)

> 
> Interestingly, while looking at this, pm_runtime_resume_and_get() can
> return '1' if e.g. rpm is disabled and the device is active. It looks
> to me like CCF treats that as an error in multiple locations.

The plot thickens...
diff mbox series

Patch

diff --git a/drivers/net/ethernet/stmicro/stmmac/Makefile b/drivers/net/ethernet/stmicro/stmmac/Makefile
index 5b57aee19267..ba2cbfa0c9d1 100644
--- a/drivers/net/ethernet/stmicro/stmmac/Makefile
+++ b/drivers/net/ethernet/stmicro/stmmac/Makefile
@@ -6,7 +6,7 @@  stmmac-objs:= stmmac_main.o stmmac_ethtool.o stmmac_mdio.o ring_mode.o	\
 	      mmc_core.o stmmac_hwtstamp.o stmmac_ptp.o dwmac4_descs.o	\
 	      dwmac4_dma.o dwmac4_lib.o dwmac4_core.o dwmac5.o hwif.o \
 	      stmmac_tc.o dwxgmac2_core.o dwxgmac2_dma.o dwxgmac2_descs.o \
-	      stmmac_xdp.o \
+	      stmmac_xdp.o stmmac_plat_lib.o \
 	      $(stmmac-y)
 
 stmmac-$(CONFIG_STMMAC_SELFTESTS) += stmmac_selftests.o
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_plat_lib.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_plat_lib.c
new file mode 100644
index 000000000000..abb9f512bb0e
--- /dev/null
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_plat_lib.c
@@ -0,0 +1,29 @@ 
+#include <linux/stmmac.h>
+#include <linux/clk.h>
+
+#include "stmmac_plat_lib.h"
+
+int dwmac_set_tx_clk_gmii(struct clk *tx_clk, int speed)
+{
+	unsigned long rate;
+
+	switch (speed) {
+	case SPEED_1000:
+		rate = 125000000;
+		break;
+
+	case SPEED_100:
+		rate = 25000000;
+		break;
+
+	case SPEED_10:
+		rate = 2500000;
+		break;
+
+	default:
+		return -ENOTSUPP;
+	}
+
+	return clk_set_rate(tx_clk, rate);
+}
+EXPORT_SYMBOL_GPL(dwmac_set_tx_clk_gmii);
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_plat_lib.h b/drivers/net/ethernet/stmicro/stmmac/stmmac_plat_lib.h
new file mode 100644
index 000000000000..926fdce379b3
--- /dev/null
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_plat_lib.h
@@ -0,0 +1,8 @@ 
+#ifndef STMMAC_PLAT_LIB_H
+#define STMMAC_PLAT_LIB_H
+
+struct clk;
+
+int dwmac_set_tx_clk_gmii(struct clk *tx_clk, int speed);
+
+#endif