Skip to contentSkip to conversation

soc: imx: imx93_lpm: return -EPROBE_DEFER when SE firmware not ready

PATCH In review Bratiranjan Acharya 2 participants 4 messages Sep 2 – Nov 2 Arm i.MX and Layerscape

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 soc: imx: imx93_lpm: return -EPROBE_DEFER when SE firmware not ready

imx_get_se_data_info() returns NULL during early boot because the EdgeLock Enclave firmware takes ~900ms to register its services. Returning -ENODEV here permanently drops the driver with no retry, leaving DDR DVFS and lpm-cooling non-functional.

Return -EPROBE_DEFER so the driver retries once SE firmware is up.

Tested on i.MX93: probe defers once, succeeds on retry, lpm-cooling and DDR frequency scaling both operational.

Fixes 711990e353bb ("LF-13910: soc: imx93: lpm changes to update exported apis")

---
v2: Add Reviewed-by from Peng Fan

 drivers/soc/imx/imx93_lpm.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/soc/imx/imx93_lpm.c +1 -1
index bfa52ff8f..26d72e3e2 100644
@@ -621,7 +621,7 @@ static int imx93_lpm_probe(struct platform_device *pdev)
 	se_data = imx_get_se_data_info(SOC_ID_OF_IMX93, 0);
 	if (!se_data) {
 		dev_err(&pdev->dev, "get se-fw2 failed\n");
-		return -ENODEV;
+		return -EPROBE_DEFER;
 	}
 
 	/*
Signature · 1 line
2.34.1
#1 lore raw
Quoted 10 lines · Bratiranjan Acharya

imx_get_se_data_info() returns NULL during early boot because the EdgeLock Enclave firmware takes ~900ms to register its services. Returning -ENODEV here permanently drops the driver with no retry, leaving DDR DVFS and lpm-cooling non-functional.

Return -EPROBE_DEFER so the driver retries once SE firmware is up.

Tested on i.MX93: probe defers once, succeeds on retry, lpm-cooling and DDR frequency scaling both operational.

Quoted 6 lines
Fixes 711990e353bb ("LF-13910: soc: imx93: lpm changes to update exported apis")

---
v2: Add Reviewed-by from Peng Fan

 drivers/soc/imx/imx93_lpm.c | 2 +-

I have seen this file at upstream tree

https://elixir.bootlin.com/linux/v7.3-rc1/source/drivers/soc/imx

Frank

Quoted 17 lines
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/soc/imx/imx93_lpm.c +1 -2
index bfa52ff8f..26d72e3e2 100644
@@ -621,7 +621,7 @@ static int imx93_lpm_probe(struct platform_device *pdev)
 	se_data = imx_get_se_data_info(SOC_ID_OF_IMX93, 0);
 	if (!se_data) {
 		dev_err(&pdev->dev, "get se-fw2 failed\n");
-		return -ENODEV;
+		return -EPROBE_DEFER;
 	}
 
 	/*
--

2.34.1

#3 lore raw
In reply to Bratiranjan Acharya

Hi Frank,

Thanks for pointing this out, apologies for the delayed response.

I checked the upstream tree (torvalds/linux master). The file drivers/soc/imx/imx93_lpm.c is not present in mainline yet -- only imx93-src.c, soc-imx.c, soc-imx8m.c and soc-imx9.c exist under drivers/soc/imx/ in Linus's tree.

Are you referring to the NXP staging/arm-soc tree? If the file is on its way upstream, I am happy to re-spin the patch against that tree once it lands, or submit it separately upstream if that is preferred.

Thanks,
Bratiranjan