Hello Marek,

On 20.11.25 12:56, Marek Vasut wrote:
On 11/20/25 10:08 AM, Heiko Schocher wrote:

Hello Heiko,

diff --git a/drivers/i2c/i2c-uclass.c b/drivers/i2c/i2c-uclass.c
index 380a9f8f3ad..d0e9dad7202 100644
--- a/drivers/i2c/i2c-uclass.c
+++ b/drivers/i2c/i2c-uclass.c
@@ -749,8 +749,11 @@ static int i2c_post_probe(struct udevice *dev)
  #if CONFIG_IS_ENABLED(OF_REAL)
      struct dm_i2c_bus *i2c = dev_get_uclass_priv(dev);
-    i2c->speed_hz = dev_read_u32_default(dev, "clock-frequency",
-                         I2C_SPEED_STANDARD_RATE);
+    if (!dev_has_ofnode(dev))
+        i2c->speed_hz = I2C_SPEED_STANDARD_RATE;
+    else
+        i2c->speed_hz = dev_read_u32_default(dev, "clock-frequency",
+                             I2C_SPEED_STANDARD_RATE);
      return dm_i2c_set_bus_speed(dev, i2c->speed_hz);
  #else


This sounds more as a bug of dev_read_u32_default() to me and should
be fixed there, else we need to make this check around each call of
dev_read_u32_default() ...

And looking into u-boot:/drivers/core/read.c there are more
candidates for such a fix (all with "default" in function name?)

and such a fix would also make the dev_has_ofnode(dev) check in
i2c_child_post_bind() obsolete.

Or may we fix the default functions in drivers/core/ofnode.c ?
No, this was the previous implementation which is now superseded by this one, see [1] [2]. The goal is to fix the entry points which pass invalid content further into U-Boot DT handling code, so the DT validation happens only one, on DT input, and not again later.

Ah, okay, so

Reviewed-by: Heiko Schocher <[email protected]>

Also, please wait for feedback on [3] [4] [5] [6] before possibly applying this 
one.

[1] 
https://lore.kernel.org/u-boot/[email protected]/
[2] 
https://lore.kernel.org/u-boot/[email protected]/

[3] 
https://lore.kernel.org/u-boot/[email protected]/
[4] 
https://lore.kernel.org/u-boot/[email protected]/
[5] 
https://lore.kernel.org/u-boot/[email protected]/
[6] 
https://lore.kernel.org/u-boot/[email protected]/

May you ping me if they are in? But for me it is fine too, if Tom
picks it up directly...

Thanks!

bye,
Heiko
--
Nabla Software Engineering
HRB 40522 Augsburg
Phone: +49 821 45592596
E-Mail: [email protected]
Geschäftsführer : Stefano Babic

Reply via email to