Skip to contentSkip to conversation

mmc: sdhci-of-dwcmshc: Minor cleanups

PATCH Superseded Artem Shimko 3 participants 11 messages May 18 – May 20

Subsystems (2)

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

cover mmc: sdhci-of-dwcmshc: Minor cleanups

Hello,

This series introduces three non-functional cleanups for the Synopsys DWC MSHC driver.

Patch 1 replaces udelay(1) with fsleep(1) in Rockchip-specific paths, following the kernel documentation recommendation for delays in non-atomic contexts.

Patch 2 removes redundant IS_ERR() checks for bus_clk in suspend/resume paths where the called functions already have internal protection.

Patch 3 converts error handling in probe paths to use dev_err_probe() macro for more compact code and better deferred probe debugging. --
Best regards,
Artem

Artem Shimko (3):
mmc: sdhci-of-dwcmshc: improve delay handling with fsleep()
mmc: sdhci-of-dwcmshc: remove redundant IS_ERR() check
mmc: sdhci-of-dwcmshc: use dev_err_probe() to simplify error paths

 drivers/mmc/host/sdhci-of-dwcmshc.c | 41 +++++++++++------------------
 1 file changed, 16 insertions(+), 25 deletions(-)
Signature · 1 line
2.43.0
#1 lore raw

PATCH 1/3 mmc: sdhci-of-dwcmshc: improve delay handling with fsleep()

The driver currently uses udelay() for short delays, which performs busy-waiting and may not be optimal even for very short durations. While udelay() is appropriate for atomic context, these specific delays occur in non-atomic contexts where sleeping is allowed.

Replace udelay(1) with fsleep(1) in dwcmshc_rk3568_set_clock() and rk35xx_sdhci_reset(). The fsleep() function automatically selects the optimal delay mechanism based on the requested duration - for delays <= 10 microseconds it will use udelay() internally, but provides better flexibility for future adjustments and makes the code more consistent with kernel best practices. No functional change is intended.

---
 drivers/mmc/host/sdhci-of-dwcmshc.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/mmc/host/sdhci-of-dwcmshc.c +2 -2
index 0b2158a7e409..8c3ca47c8060 100644
@@ -816,7 +816,7 @@ static void dwcmshc_rk3568_set_clock(struct sdhci_host *host, unsigned int clock
 
 	/* Reset DLL */
 	sdhci_writel(host, BIT(1), DWCMSHC_EMMC_DLL_CTRL);
-	udelay(1);
+	fsleep(1);
 	sdhci_writel(host, 0x0, DWCMSHC_EMMC_DLL_CTRL);
 
 	/*
@@ -895,7 +895,7 @@ static void rk35xx_sdhci_reset(struct sdhci_host *host, u8 mask)
 
 	if (mask & SDHCI_RESET_ALL && priv->reset) {
 		reset_control_assert(priv->reset);
-		udelay(1);
+		fsleep(1);
 		reset_control_deassert(priv->reset);
 	}
 
Signature · 1 line
2.43.0
#2 lore raw

Re: [PATCH 1/3] mmc: sdhci-of-dwcmshc: improve delay handling with fsleep()

Quoted 11 lines · Artem Shimko

The driver currently uses udelay() for short delays, which performs busy-waiting and may not be optimal even for very short durations. While udelay() is appropriate for atomic context, these specific delays occur in non-atomic contexts where sleeping is allowed.

Replace udelay(1) with fsleep(1) in dwcmshc_rk3568_set_clock() and rk35xx_sdhci_reset(). The fsleep() function automatically selects the optimal delay mechanism based on the requested duration - for delays <= 10 microseconds it will use udelay() internally, but provides better flexibility for future adjustments and makes the code more consistent with kernel best practices. No functional change is intended.

Read, what you wrote. With the above commit message this is no point in this patch.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#3 lore raw

Re: [PATCH 1/3] mmc: sdhci-of-dwcmshc: improve delay handling with fsleep()

Hi Andy,

Quoted 2 lines · Andy Shevchenko

Read, what you wrote. With the above commit message this is no point in this patch.

I tried to replace *sleep() with fsleep() API. It seems I misunderstood the hint [1].

Or is that the commit description mistake?

Thank you for reviewing it.

[1] https://lore.kernel.org/all/[email protected]/
--
Best regards,
Artem

#4 lore raw

Re: [PATCH 1/3] mmc: sdhci-of-dwcmshc: improve delay handling with fsleep()

Quoted 7 lines · Artem Shimko
Quoted 2 lines · Andy Shevchenko

Read, what you wrote. With the above commit message this is no point in this patch.

I tried to replace *sleep() with fsleep() API. It seems I misunderstood the hint [1].

Yep, the hint was basically about uses of usleep_range(). udelay for 1 or 2 microseconds is fine as is. Of course for the sake of consistency it can be moved to fsleep() (when the context is non-atomic), but solely udelay(1) --> fsleep(1) makes no needed churn.

Quoted 5 lines

Or is that the commit description mistake?

Thank you for reviewing it.

[1] https://lore.kernel.org/all/[email protected]/

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#5 lore raw

Re: [PATCH 1/3] mmc: sdhci-of-dwcmshc: improve delay handling with fsleep()

Hi Andy,

Quoted 1 line · Andy Shevchenko

Yep, the hint was basically about uses of usleep_range().

Just to clarify before sending V2, for example (midpoint of the range):
s/usleep_range(1000, 2000)/fsleep(1500)/ ?

Quoted 1 line

microseconds is fine as is.

Since they occur in non-atomic context
(inside set_clock and reset functions), would it be worth replacing them with fsleep(1) as well for consistency? =)

Thank you!
--
Best regards,
Artem Shimko

Quoted 12 lines
Quoted 6 lines

Or is that the commit description mistake?

Thank you for reviewing it.

[1] https://lore.kernel.org/all/[email protected]/

--
With Best Regards,
Andy Shevchenko

#6 lore raw

PATCH 2/3 mmc: sdhci-of-dwcmshc: remove redundant IS_ERR() check

In reply to Artem Shimko

The clk_disable_unprepare() function has internal protection against ERR_PTR and NULL pointers (IS_ERR_OR_NULL). Remove the redundant IS_ERR() check for bus_clk in dwcmshc_suspend() and in the error path of dwcmshc_resume() to simplify the code.

Note that the clk_prepare_enable() call in dwcmshc_resume() must retain its IS_ERR() check because clk_prepare() only handles NULL pointers, not ERR_PTR.

No functional change intended.

---
 drivers/mmc/host/sdhci-of-dwcmshc.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)
drivers/mmc/host/sdhci-of-dwcmshc.c +2 -4
index 8c3ca47c8060..29bb46783ee5 100644
@@ -2564,8 +2564,7 @@ static int dwcmshc_suspend(struct device *dev)
 		return ret;
 
 	clk_disable_unprepare(pltfm_host->clk);
-	if (!IS_ERR(priv->bus_clk))
-		clk_disable_unprepare(priv->bus_clk);
+	clk_disable_unprepare(priv->bus_clk);
 
 	clk_bulk_disable_unprepare(priv->num_other_clks, priv->other_clks);
 
@@ -2608,8 +2607,7 @@ static int dwcmshc_resume(struct device *dev)
 disable_other_clks:
 	clk_bulk_disable_unprepare(priv->num_other_clks, priv->other_clks);
 disable_bus_clk:
-	if (!IS_ERR(priv->bus_clk))
-		clk_disable_unprepare(priv->bus_clk);
+	clk_disable_unprepare(priv->bus_clk);
 disable_clk:
 	clk_disable_unprepare(pltfm_host->clk);
 	return ret;
Signature · 1 line
2.43.0
maintainer Acked-by on v1 2/3
#7 lore raw

Re: [PATCH 2/3] mmc: sdhci-of-dwcmshc: remove redundant IS_ERR() check

Quoted 5 lines · Artem Shimko

The clk_disable_unprepare() function has internal protection against ERR_PTR and NULL pointers (IS_ERR_OR_NULL). Remove the redundant IS_ERR() check for bus_clk in dwcmshc_suspend() and in the error path of dwcmshc_resume() to simplify the code.

Quoted 7 lines

Note that the clk_prepare_enable() call in dwcmshc_resume() must retain its IS_ERR() check because clk_prepare() only handles NULL pointers, not ERR_PTR.

No functional change intended.

Quoted 28 lines
---
 drivers/mmc/host/sdhci-of-dwcmshc.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)
drivers/mmc/host/sdhci-of-dwcmshc.c +2 -4
index 8c3ca47c8060..29bb46783ee5 100644
@@ -2564,8 +2564,7 @@ static int dwcmshc_suspend(struct device *dev)
 		return ret;
 
 	clk_disable_unprepare(pltfm_host->clk);
-	if (!IS_ERR(priv->bus_clk))
-		clk_disable_unprepare(priv->bus_clk);
+	clk_disable_unprepare(priv->bus_clk);
 
 	clk_bulk_disable_unprepare(priv->num_other_clks, priv->other_clks);
 
@@ -2608,8 +2607,7 @@ static int dwcmshc_resume(struct device *dev)
 disable_other_clks:
 	clk_bulk_disable_unprepare(priv->num_other_clks, priv->other_clks);
 disable_bus_clk:
-	if (!IS_ERR(priv->bus_clk))
-		clk_disable_unprepare(priv->bus_clk);
+	clk_disable_unprepare(priv->bus_clk);
 disable_clk:
 	clk_disable_unprepare(pltfm_host->clk);
 	return ret;
#8 lore raw

PATCH 3/3 mmc: sdhci-of-dwcmshc: use dev_err_probe() to simplify error paths

In reply to Artem Shimko

Replace common pattern of dev_err() + return with dev_err_probe() in probe functions and their callees. This macro provides standardized error message format with symbolic error names and adds deferred probe debugging information.

The conversion makes the code more compact and ensures consistent error logging across all initialization paths.

---
 drivers/mmc/host/sdhci-of-dwcmshc.c | 31 +++++++++++------------------
 1 file changed, 12 insertions(+), 19 deletions(-)
drivers/mmc/host/sdhci-of-dwcmshc.c +12 -19
index 29bb46783ee5..ccd5f4d6fec0 100644
@@ -917,11 +917,9 @@ static int dwcmshc_rk35xx_init(struct device *dev, struct sdhci_host *host,
 		return -ENOMEM;
 
 	priv->reset = devm_reset_control_array_get_optional_exclusive(mmc_dev(host->mmc));
-	if (IS_ERR(priv->reset)) {
-		err = PTR_ERR(priv->reset);
-		dev_err(mmc_dev(host->mmc), "failed to get reset control %d\n", err);
-		return err;
-	}
+	if (IS_ERR(priv->reset))
+		return dev_err_probe(mmc_dev(host->mmc), PTR_ERR(priv->reset),
+				     "failed to get reset control\n");
 
 	err = dwcmshc_get_enable_other_clks(mmc_dev(host->mmc), dwc_priv,
 					    ARRAY_SIZE(clk_ids), clk_ids);
@@ -1781,10 +1779,8 @@ static int eic7700_init(struct device *dev, struct sdhci_host *host, struct dwcm
 	dwc_priv->priv = priv;
 
 	ret = sdhci_eic7700_reset_init(dev, dwc_priv->priv);
-	if (ret) {
-		dev_err(dev, "failed to reset\n");
-		return ret;
-	}
+	if (ret)
+		return dev_err_probe(dev, ret, "failed to reset\n");
 
 	ret = dwcmshc_get_enable_other_clks(mmc_dev(host->mmc), dwc_priv,
 					    ARRAY_SIZE(clk_ids), clk_ids);
@@ -1792,16 +1788,14 @@ static int eic7700_init(struct device *dev, struct sdhci_host *host, struct dwcm
 		return ret;
 
 	ret = of_parse_phandle_with_fixed_args(dev->of_node, "eswin,hsp-sp-csr", 2, 0, &args);
-	if (ret) {
-		dev_err(dev, "Fail to parse 'eswin,hsp-sp-csr' phandle (%d)\n", ret);
-		return ret;
-	}
+	if (ret)
+		return dev_err_probe(dev, ret, "Fail to parse 'eswin,hsp-sp-csr' phandle\n");
 
 	hsp_regmap = syscon_node_to_regmap(args.np);
 	if (IS_ERR(hsp_regmap)) {
-		dev_err(dev, "Failed to get regmap for 'eswin,hsp-sp-csr'\n");
 		of_node_put(args.np);
-		return PTR_ERR(hsp_regmap);
+		return dev_err_probe(dev, PTR_ERR(hsp_regmap),
+				     "Failed to get regmap for 'eswin,hsp-sp-csr'\n");
 	}
 	hsp_int_status = args.args[0];
 	hsp_pwr_ctrl = args.args[1];
@@ -2408,10 +2402,9 @@ static int dwcmshc_probe(struct platform_device *pdev)
 	u32 extra, caps;
 
 	pltfm_data = device_get_match_data(&pdev->dev);
-	if (!pltfm_data) {
-		dev_err(&pdev->dev, "Error: No device match data found\n");
-		return -ENODEV;
-	}
+	if (!pltfm_data)
+		return dev_err_probe(&pdev->dev, -ENODEV,
+				     "Error: No device match data found\n");
 
 	host = sdhci_pltfm_init(pdev, &pltfm_data->pdata,
 				sizeof(struct dwcmshc_priv));
Signature · 1 line
2.43.0
maintainer
#9 lore raw

Re: [PATCH 3/3] mmc: sdhci-of-dwcmshc: use dev_err_probe() to simplify error paths

Quoted 67 lines · Artem Shimko

Replace common pattern of dev_err() + return with dev_err_probe() in probe functions and their callees. This macro provides standardized error message format with symbolic error names and adds deferred probe debugging information.

The conversion makes the code more compact and ensures consistent error logging across all initialization paths.

---
 drivers/mmc/host/sdhci-of-dwcmshc.c | 31 +++++++++++------------------
 1 file changed, 12 insertions(+), 19 deletions(-)
drivers/mmc/host/sdhci-of-dwcmshc.c +9 -15
index 29bb46783ee5..ccd5f4d6fec0 100644
@@ -917,11 +917,9 @@ static int dwcmshc_rk35xx_init(struct device *dev, struct sdhci_host *host,
 		return -ENOMEM;
 
 	priv->reset = devm_reset_control_array_get_optional_exclusive(mmc_dev(host->mmc));
-	if (IS_ERR(priv->reset)) {
-		err = PTR_ERR(priv->reset);
-		dev_err(mmc_dev(host->mmc), "failed to get reset control %d\n", err);
-		return err;
-	}
+	if (IS_ERR(priv->reset))
+		return dev_err_probe(mmc_dev(host->mmc), PTR_ERR(priv->reset),
+				     "failed to get reset control\n");
 
 	err = dwcmshc_get_enable_other_clks(mmc_dev(host->mmc), dwc_priv,
 					    ARRAY_SIZE(clk_ids), clk_ids);
@@ -1781,10 +1779,8 @@ static int eic7700_init(struct device *dev, struct sdhci_host *host, struct dwcm
 	dwc_priv->priv = priv;
 
 	ret = sdhci_eic7700_reset_init(dev, dwc_priv->priv);
-	if (ret) {
-		dev_err(dev, "failed to reset\n");
-		return ret;
-	}
+	if (ret)
+		return dev_err_probe(dev, ret, "failed to reset\n");
 
 	ret = dwcmshc_get_enable_other_clks(mmc_dev(host->mmc), dwc_priv,
 					    ARRAY_SIZE(clk_ids), clk_ids);
@@ -1792,16 +1788,14 @@ static int eic7700_init(struct device *dev, struct sdhci_host *host, struct dwcm
 		return ret;
 
 	ret = of_parse_phandle_with_fixed_args(dev->of_node, "eswin,hsp-sp-csr", 2, 0, &args);
-	if (ret) {
-		dev_err(dev, "Fail to parse 'eswin,hsp-sp-csr' phandle (%d)\n", ret);
-		return ret;
-	}
+	if (ret)
+		return dev_err_probe(dev, ret, "Fail to parse 'eswin,hsp-sp-csr' phandle\n");
 
 	hsp_regmap = syscon_node_to_regmap(args.np);
 	if (IS_ERR(hsp_regmap)) {
-		dev_err(dev, "Failed to get regmap for 'eswin,hsp-sp-csr'\n");
 		of_node_put(args.np);
-		return PTR_ERR(hsp_regmap);
+		return dev_err_probe(dev, PTR_ERR(hsp_regmap),
+				     "Failed to get regmap for 'eswin,hsp-sp-csr'\n");
 	}
 	hsp_int_status = args.args[0];
 	hsp_pwr_ctrl = args.args[1];

drivers/mmc/host/sdhci-of-dwcmshc.c static int dwcmshc_probe(struct platform_device *pdev) · line 2407

Quoted 10 lines
drivers/mmc/host/sdhci-of-dwcmshc.c +3 -4
@@ -2408,10 +2402,9 @@ static int dwcmshc_probe(struct platform_device *pdev)
 	u32 extra, caps;
 
 	pltfm_data = device_get_match_data(&pdev->dev);
-	if (!pltfm_data) {
-		dev_err(&pdev->dev, "Error: No device match data found\n");
-		return -ENODEV;
-	}
+	if (!pltfm_data)
+		return dev_err_probe(&pdev->dev, -ENODEV,
+				     "Error: No device match data found\n");

"Error: " is especially redundant with dev_err_probe(), and it can all be on 1 line e.g.

		return dev_err_probe(&pdev->dev, -ENODEV, "No device match data found\n");
Quoted 3 lines
drivers/mmc/host/sdhci-of-dwcmshc.c +0 -0
 	host = sdhci_pltfm_init(pdev, &pltfm_data->pdata,
 				sizeof(struct dwcmshc_priv));
#10 lore raw

Re: [PATCH 3/3] mmc: sdhci-of-dwcmshc: use dev_err_probe() to simplify error paths

Hi Adrian,

Quoted 5 lines · Adrian Hunter

drivers/mmc/host/sdhci-of-dwcmshc.c static int dwcmshc_probe(struct platform_device *pdev) · line 2407

Quoted 3 lines
drivers/mmc/host/sdhci-of-dwcmshc.c +3 -0
+     if (!pltfm_data)
+             return dev_err_probe(&pdev->dev, -ENODEV,
+                                  "Error: No device match data found\n");

"Error: " is especially redundant with dev_err_probe(),

Missed that.

Quoted 1 line

and it can all be on 1 line e.g.

Got it.

Thank you for your review!
--
Best Regards,
Artem