Skip to contentSkip to conversation

mmc: renesas_sdhi: add checks for pinctrl_lookup_state

PATCH In review Chuhong Yuan 4 participants 8 messages Oct 18, 2019 – Nov 24, 2019

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 mmc: renesas_sdhi: add checks for pinctrl_lookup_state

renesas_sdhi_probe misses checks for pinctrl_lookup_state and may miss failures.
Add checks for them to fix the problem.

---
 drivers/mmc/host/renesas_sdhi_core.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/mmc/host/renesas_sdhi_core.c +5 -0
index d4ada5cca2d1..dc5ad6632df3 100644
@@ -694,8 +694,13 @@ int renesas_sdhi_probe(struct platform_device *pdev,
 	if (!IS_ERR(priv->pinctrl)) {
 		priv->pins_default = pinctrl_lookup_state(priv->pinctrl,
 						PINCTRL_STATE_DEFAULT);
+		if (IS_ERR(priv->pins_default))
+			return PTR_ERR(priv->pins_default);
+
 		priv->pins_uhs = pinctrl_lookup_state(priv->pinctrl,
 						"state_uhs");
+		if (IS_ERR(priv->pins_uhs))
+			return PTR_ERR(priv->pins_uhs);
 	}
 
 	host = tmio_mmc_host_alloc(pdev, mmc_data);
Signature · 1 line
2.20.1
#1 lore raw
Quoted 16 lines · Chuhong Yuan

renesas_sdhi_probe misses checks for pinctrl_lookup_state and may miss failures.
Add checks for them to fix the problem.

---
 drivers/mmc/host/renesas_sdhi_core.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/mmc/host/renesas_sdhi_core.c +0 -0
index d4ada5cca2d1..dc5ad6632df3 100644
@@ -694,8 +694,13 @@ int renesas_sdhi_probe(struct platform_device *pdev,
        if (!IS_ERR(priv->pinctrl)) {

drivers/mmc/host/renesas_sdhi_core.c int renesas_sdhi_probe(struct platform_device *pdev, · line 704

Quoted 10 lines
drivers/mmc/host/renesas_sdhi_core.c +5 -0
                priv->pins_default = pinctrl_lookup_state(priv->pinctrl,
                                                PINCTRL_STATE_DEFAULT);
+               if (IS_ERR(priv->pins_default))
+                       return PTR_ERR(priv->pins_default);
+
                priv->pins_uhs = pinctrl_lookup_state(priv->pinctrl,
                                                "state_uhs");
+               if (IS_ERR(priv->pins_uhs))
+                       return PTR_ERR(priv->pins_uhs);
        }

This looks correct to me, as I guess if there is a pinctrl specified for device node of the controller, it means that it should be used!?

I understand that this is only used for those variants that supports UHS-I via the renesas_sdhi_start_signal_voltage_switch(). Wolfram, is this fine you think?

Quoted 5 lines
drivers/mmc/host/renesas_sdhi_core.c +0 -1
        host = tmio_mmc_host_alloc(pdev, mmc_data);
--

2.20.1

Kind regards
Uffe

#2 lore raw
Quoted 4 lines

I understand that this is only used for those variants that supports UHS-I via the renesas_sdhi_start_signal_voltage_switch(). Wolfram, is this fine you think?

I'll double check tomorrow.

#3 lore raw
In reply to Ulf Hansson
Quoted 31 lines · Ulf Hansson
Quoted 16 lines · Chuhong Yuan

renesas_sdhi_probe misses checks for pinctrl_lookup_state and may miss failures.
Add checks for them to fix the problem.

---
 drivers/mmc/host/renesas_sdhi_core.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/mmc/host/renesas_sdhi_core.c +0 -0
index d4ada5cca2d1..dc5ad6632df3 100644
@@ -694,8 +694,13 @@ int renesas_sdhi_probe(struct platform_device *pdev,
        if (!IS_ERR(priv->pinctrl)) {

drivers/mmc/host/renesas_sdhi_core.c int renesas_sdhi_probe(struct platform_device *pdev, · line 704

Quoted 10 lines
drivers/mmc/host/renesas_sdhi_core.c +5 -0
                priv->pins_default = pinctrl_lookup_state(priv->pinctrl,
                                                PINCTRL_STATE_DEFAULT);
+               if (IS_ERR(priv->pins_default))
+                       return PTR_ERR(priv->pins_default);
+
                priv->pins_uhs = pinctrl_lookup_state(priv->pinctrl,
                                                "state_uhs");
+               if (IS_ERR(priv->pins_uhs))
+                       return PTR_ERR(priv->pins_uhs);
        }

This looks correct to me, as I guess if there is a pinctrl specified for device node of the controller, it means that it should be used!?

Quoted 3 lines

I understand that this is only used for those variants that supports UHS-I via the renesas_sdhi_start_signal_voltage_switch(). Wolfram, is this fine you think?

Well, I don't like to bail out because this error is not fatal for basic operations. How about releasing priv->pinctrl again with an additional warning that pinctrl settings are broken and will prevent 1.8v modes?

Opinions?

#4 lore raw
Quoted 37 lines · Wolfram Sang
Quoted 31 lines · Ulf Hansson
Quoted 16 lines · Chuhong Yuan

renesas_sdhi_probe misses checks for pinctrl_lookup_state and may miss failures.
Add checks for them to fix the problem.

---
 drivers/mmc/host/renesas_sdhi_core.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/mmc/host/renesas_sdhi_core.c +0 -0
index d4ada5cca2d1..dc5ad6632df3 100644
@@ -694,8 +694,13 @@ int renesas_sdhi_probe(struct platform_device *pdev,
        if (!IS_ERR(priv->pinctrl)) {

drivers/mmc/host/renesas_sdhi_core.c int renesas_sdhi_probe(struct platform_device *pdev, · line 704

Quoted 10 lines
drivers/mmc/host/renesas_sdhi_core.c +5 -0
                priv->pins_default = pinctrl_lookup_state(priv->pinctrl,
                                                PINCTRL_STATE_DEFAULT);
+               if (IS_ERR(priv->pins_default))
+                       return PTR_ERR(priv->pins_default);
+
                priv->pins_uhs = pinctrl_lookup_state(priv->pinctrl,
                                                "state_uhs");
+               if (IS_ERR(priv->pins_uhs))
+                       return PTR_ERR(priv->pins_uhs);
        }

This looks correct to me, as I guess if there is a pinctrl specified for device node of the controller, it means that it should be used!?

Quoted 3 lines

I understand that this is only used for those variants that supports UHS-I via the renesas_sdhi_start_signal_voltage_switch(). Wolfram, is this fine you think?

Quoted 5 lines

Well, I don't like to bail out because this error is not fatal for basic operations. How about releasing priv->pinctrl again with an additional warning that pinctrl settings are broken and will prevent 1.8v modes?

Opinions?

Hmm, from a mmc driver probe point of view, I don't quite share this approach.

I would rather fail as it forces the DTB to be corrected immediately, rather than trusting some developer to look at a warning in a log. The point is, in such a case it may never get fixed, if the product is shipped with the wrong DTB.

My concern at this point is rather to break existing DTBs, but it seems that should not be an issue, right?

Kind regards
Uffe

#5 lore raw

HI Ulf,

Quoted 4 lines
Quoted 4 lines

Well, I don't like to bail out because this error is not fatal for basic operations. How about releasing priv->pinctrl again with an additional warning that pinctrl settings are broken and will prevent 1.8v modes?

Quoted 8 lines
Quoted 1 line

Opinions?

Hmm, from a mmc driver probe point of view, I don't quite share this approach.

I would rather fail as it forces the DTB to be corrected immediately, rather than trusting some developer to look at a warning in a log. The point is, in such a case it may never get fixed, if the product is shipped with the wrong DTB.

I could agree to this arguement, iff the only way pinctrl_select fails is a DT misconfiguration. I am not sure if this is true right now, and we can't be sure about the future. Or?

Quoted 2 lines

My concern at this point is rather to break existing DTBs, but it seems that should not be an issue, right?

AFAIK not for the upstream ones, I don't know the customer ones.

I still prefer to not bail out.

Kind regards,

Wolfram

#6 lore raw

Hi Wolfram,

Quoted 8 lines · Wolfram Sang
Quoted 8 lines
Quoted 5 lines

Well, I don't like to bail out because this error is not fatal for basic operations. How about releasing priv->pinctrl again with an additional warning that pinctrl settings are broken and will prevent 1.8v modes?

Opinions?

Hmm, from a mmc driver probe point of view, I don't quite share this approach.

Quoted 8 lines
Quoted 4 lines

I would rather fail as it forces the DTB to be corrected immediately, rather than trusting some developer to look at a warning in a log. The point is, in such a case it may never get fixed, if the product is shipped with the wrong DTB.

I could agree to this arguement, iff the only way pinctrl_select fails is a DT misconfiguration. I am not sure if this is true right now, and we can't be sure about the future. Or?

Isn't "state_uhs" optional? So bailing out if it doesn't exist is wrong. "default" should be there, if the device node has pinctrl properties.

renesas_sdhi_start_signal_voltage_switch() already handles fallback to 3v3 operation.

Gr{oetje,eeting}s,

Geert

Signature · 5 lines
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds
#7 lore raw
Quoted 24 lines · Geert Uytterhoeven

Hi Wolfram,

Quoted 8 lines · Wolfram Sang
Quoted 8 lines
Quoted 5 lines

Well, I don't like to bail out because this error is not fatal for basic operations. How about releasing priv->pinctrl again with an additional warning that pinctrl settings are broken and will prevent 1.8v modes?

Opinions?

Hmm, from a mmc driver probe point of view, I don't quite share this approach.

Quoted 8 lines
Quoted 4 lines

I would rather fail as it forces the DTB to be corrected immediately, rather than trusting some developer to look at a warning in a log. The point is, in such a case it may never get fixed, if the product is shipped with the wrong DTB.

I could agree to this arguement, iff the only way pinctrl_select fails is a DT misconfiguration. I am not sure if this is true right now, and we can't be sure about the future. Or?

Isn't "state_uhs" optional? So bailing out if it doesn't exist is wrong. "default" should be there, if the device node has pinctrl properties.

Quoted 7 lines

renesas_sdhi_start_signal_voltage_switch() already handles fallback to 3v3 operation.

Gr{oetje,eeting}s,

Geert

Just ping this thread.
I am not clear about what is the conclusion of how to deal with the error? Should I resend a new version of the patch?

Regards,
Chuhong

Quoted 6 lines

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds